migration and inequality in germany 1870-1913 (oxford historical monographs)

OXFORD HISTORICAL MONOGRAPHS

editors

j . maddicott r. j . w. evansj . harris b. ward-perkinsj . robertson r. service

p. a . slack

This page intentionally left blank

Migration and Inequalityin Germany1870–1913

OLIVER GRANT

CLARENDON PRESS � OXFORD

3Great Clarendon Street, Oxford ox2 6dp

Oxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,

and education by publishing worldwide in

Oxford New York

Auckland Cape Town Dar es Salaam Hong Kong Karachi

Kuala Lumpur Madrid Melbourne Mexico City NairobiNew Delhi Shanghai Taipei Toronto

With oYces in

Argentina Austria Brazil Chile Czech Republic France Greece

Guatemala Hungary Italy Japan Poland Portugal SingaporeSouth Korea Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trade mark of Oxford University Press

in the UK and in certain other countries

Published in the United States

by Oxford University Press Inc., New York

� Oliver Grant 2005

The moral rights of the author have been assertedDatabase right Oxford University Press (maker)

First published 2005.

All rights reserved. No part of this publication may be reproduced,

stored in a retrieval system, or transmitted, in any form or by any means,without the prior permission in writing of Oxford University Press,

or as expressly permitted by law, or under terms agreed with the appropriatereprographics rights organization. Enquiries concerning reproduction

outside the scope of the above should be sent to the Rights Department,

Oxford University Press, at the address above

You must not circulate this book in any other binding or coverand you must impose this same condition on any acquirer

British Library Cataloguing in Publication Data

Data available

Library of Congress Cataloging in Publication Data

Data available

Typeset by SPI Publisher Services, Pondicherry, India

Printed in Great Britainon acid-free paper by

Biddles Ltd.King’s Lynn, Norfolk.

ISBN 0-19-927656-0

1 3 5 7 9 10 8 6 4 2

PREFACE

This study has its origins in a D.Phil. thesis, ‘Internal Migration in

Germany 1870–1913’, which was motivated by the view that migration

is the central problem of an industrializing society. The initial stimulus

came from a visit to China in the spring of 1995, a visit which reawoke

my interest in the process of economic growth and its associated prob-

lems and opportunities. This combined with an older fascination with

the course of German history to produce the thesis which has now

evolved into this book.

Comparison is one of the most powerful tools at the disposal of a

historian, and one of the themes of this book is that historians should

not be timid or reticent about this. They should be prepared to make

broad comparisons across time and across continents. They should

insist that theirs is a discipline which can contribute to the study of

major issues of current concern: the causes of war and peace, economic

growth and the creation of modern industrial societies, the foundations

of democracy and the establishment of individual rights. If historians

turn their backs on these issues, then others will Wll this gap.

But, while historians can bring to these subjects the additional per-

spective which is gained by historical analysis, they should also be

aware of the researches on contemporary societies which social scientists

are making, and consider the ways that these can bring new perspec-

tives to bear on historical issues. This means confronting the methods

that social scientists bring to bear on these issues, in particular model-

building and the use of statistical analysis. Both will be used in the

chapters that follow.

The way that the thesis has evolved into this book has been

inXuenced by the experience of designing and teaching a postgraduate

course on comparative industrialization at Oxford University, entitled

‘Industrialization in Europe, North America and East Asia since 1700’.

I would like to thank Professor Avner OVer for having encouraged me

to do this. I also thank those students who have taken the course for

many interesting discussions on the problems of industrializing soci-

eties.

When the problems of Imperial Germany are set in this wider con-

text, it soon becomes apparent that, to a large extent, this was a success-

ful and progressive society, which dealt with the problems of rapid

industrialization unusually well. It was in many respects the Wrst

‘Modell Deutschland’.

I owe thanks to many people who have helped directly and indirectly

over the past three years. I have beneWted from exemplary supervision

from Professors Charles Feinstein and Hartmut Pogge von Strand-

mann. I have much enjoyed our conversations. I have learnt much from

their observations and suggestions.

I had the additional good fortune to have had two excellent examin-

ers in Professors Niall Ferguson and Albrecht Ritschl. This work has

been considerably improved as a result of their comments.

Individual chapters have been read by Anthony Atkinson, Avner

OVer, Hans-Joachim Voth, Liam Brunt, James Foreman-Peck, Tom

Nicholas, and Eric Grimmer-Solem. I thank all these for their time and

attention.

I owe much to the library of the London School of Economics. This

has a large collection of German statistical materials on open access.

I am particularly grateful that the German statistical collection has

remained on open access during the rebuilding of the main library.

I would like to express my thanks to the staV of the Nuneham Cour-

tenay depository of the Bodleian Library, to Chris Newbould and John

Slater. The depository is not well suited to the arrival of visitors with

access permits, who sit in corridors and cause obstructions. Their toler-

ance was much appreciated.

I owe thanks to my wife and children, the arrival of the latter having

coincided with the beginning of this study. They have provided much

diversion and, at times, a necessary sense of proportion.

But my Wnal thanks are owed to my farm staV, to Paul Burns, Russell

Donovan, and Alan Roberts, who have taken on additional responsibil-

ities while I have been working on this study. It is beyond question

that, without their eVorts, this would not have been possible.

O.W.G.

18 June 2004

vi Preface

CONTENTS

1. Imperial Germany as an Example of Industrialization

under Labour Surplus Conditions 1

2. Sources of Inequality in Rural Germany 22

3. The Pattern of Migration, 1870–1913 56

4. Migration in Germany 1870–1913: A Statistical

Analysis 97

5. Demography and Migration 124

6. Migration, Farm Size, and the Condition of the

Agricultural Labourer 181

7. Agricultural Productivity, Labour Surplus, and

Migration 215

8. Migration and Urban Labour Markets 253

9. Industrialization, Migration, and Inequality 293

10. Challenging the Kehrite View of Imperial Germany 330

References and Sources 361

Index 389

This page intentionally left blank

1

Imperial Germany as an Example

of Industrialization under

Labour Surplus Conditions

1. Introduction : ‘A house div ided ’—M igration

and Inequality in Imperial Germany , 1870–1913

‘A house divided cannot stand’ is Abraham Lincoln’s precis of the

issues facing the United States in the Civil War; it can also serve as a

summary of the views of a school of German historians, the Kehrite

school. The key concept of this school is the ‘primacy of internal polit-

ics’: the view that the internal weaknesses of pre-1914 Germany led to a

series of expedients which made German foreign policy a disruptive

force in pre-1914 European diplomacy, and which prevented the settle-

ment of outstanding disputes by any means short of a general European

conXict. The core of this policy was the building of a Xeet large enough

to challenge British naval hegemony.1

What were the sources of the internal weakness of Imperial Germany?

Historians have produced a list of suggested Xaws in the polity and social

structure of Imperial Germany: a series of ‘skirted decisions’ which

explain Germany’s failure to make political progress commensurate with

its economic advances.2 In this analysis, the fundamental assumption is

1 Kehr’s own work, originally written in the 1930s, is available in English, Kehr (1975)and (1977). Other important works in the Kehrite tradition include Wehler (1969) and(1985), Berghahn (1971), Stegemann (1977), and the collection in Sturmer (1976). Theexistence of a Kehrite school has been denied by Puhle (1978) and there is some disagree-ment over what the key Kehrite concept really is; for Eley (1986) it is either ‘the primacyof pre-industrial traditions’, p. 42, or the concept of Sammlungspolitik, p. 111. For thepurposes of this study, the key Kehrite concept is the ‘primacy of internal politics’:meaning by this, that the Kaiserreich suVered from internal problems, which had anadverse eVect on foreign policy to an extent which is unusual compared to other states.This last point is rarely made explicitly, but it is important if the concept is to have anygenuine force. For a simple statement of this view, see Rohl (1969), 10.

2 The phrase ‘skirted decisions’ is taken from the title of an essay in Mommsen (1995).Mommsen’s work stresses the political crisis produced by the inability of the Bismarckian

that there were options which, if taken, would have alleviated the social

strains and enabled Germany to become a normal, modern, democratic

industrial state. And the implication of this is that such a state would

have been a more stable and predictable element of the pre-1914 diplo-

matic order.

These ‘skirted decisions’ were not just political and diplomatic: they

were derived from a fundamental unwillingness of the ‘pre-industrial’

elites of Imperial Germany to modernize structures and institutions

which had their roots in the traditions of an agrarian society.3 This

produced a bureaucracy recruited from a narrow and unrepresentative

social base, a political system which limited democratic control over the

executive, and a social order in which large numbers felt themselves

to be excluded or marginalized. In short, an outcome of ‘incomplete

modernization’, derived in part from the Bismarckian origins of the

Kaiserreich in a ‘revolution from above’.

Historians outside the group normally identiWed as Kehrite, or Kehr-

inspired, have accepted parts of this argument, or put it forward in

diVerent guises. The broader concept of a Sammlungspolitik, policies

designed to rally ‘loyal’ forces in support of the Imperial government,

has replaced Kehr’s narrower emphasis on the trade-oV of agricultural

protection for battleship construction. Modern studies place greater

stress on social and cultural divisions as the ultimate sources of the

weakness of the political structure: the rise of extremist ideas, the isol-

ation of important social strata, and the domination of political parties

by economic interest groups. Instead of a conscious ‘Flucht nach Vorn’,

an attempt to solve internal problems by foreign policy adventures,

more recent studies have set German foreign policy in the context of

a society where Social Darwinism had a ‘special resonance’, where

‘chauvinistic bile’ had a ‘broad appeal’, where paciWsm was ‘relatively

weak’ and ‘its opposite was correspondingly powerful’. Under these

circumstances, the ‘restless, querulous’ nature of German foreign

policy, the ‘fatalistic brinkmanship’ of the civilian leadership in the

constitution to adapt to changing economic and social conditions, and argues that ‘in July1914 the ‘‘latent crisis’’ of the German political system spilled over onto the internationalstage, setting in train a world war’, Mommsen (1995), 204. This can be regarded as avariation on the basic Kehrite theme.

3 According to Wehler (1985), 236, Bismarck, and the ‘Wilhelmine polycracy’ thatfollowed him, were determined to avoid the spread of democracy and moves towardsequal rights. The militancy of the property-owning elites, and their determination topreserve their privileges at all costs, is stressed in, for example, Puhle (1976), 341.

2 Industrialization under Labour Surplus Conditions

crisis of July 1914, are seen, implicitly or explicitly, as expressions of

deeper Xaws.4

This study puts forward an alternative hypothesis: that many of the

internal stresses and strains of Imperial Germany were the inevitable

consequences of industrialization, and that there was little that German

political leaders could have done about this. Even if the political elites

had been prepared to make concessions this would not have produced

an immediate shift to a more harmonious society: the option of a ‘social

market economy’ was not available in pre-1914 circumstances. Instead,

the maturing of the industrial economy was a necessary precondition

for the attainment of social stability and political maturity.

The point that the transition from an agrarian to an industrial econ-

omy inevitably creates social and political strains is not a new one. Other

societies had also experienced these tensions. But the process of indus-

trialization in Germany was diVerent from that in the pioneer industrial-

izing countries because Germany had the advantages, and the

disadvantages, of ‘relative backwardness’, as originally pointed out by

Alexander Gerschenkron.5 Germany could make use of advanced indus-

trial technology developed elsewhere. This meant that the pace of

German industrialization was faster. It meant that the gap between

productivity in the mechanized industrial sectors and the traditional

sectors (agriculture and handcraft) was much greater. Transfers of

labour between these sectors was a major contribution to overall growth.

But the same transfers brought about social and institutional disruption.

Pre-1914 Germany was the Wrst economy to experience the tele-

scoped process of industrial development typical of those countries

which make their way to the front rank of industrial powers having

started from a position of relative backwardness. Many historians have

noted the rapidity of the process of social and economic transformation

in Imperial Germany. But the full implications of this have not been

examined.6 One important consequence is that the analysis of German

economy in the period needs to make use of concepts and tools taken

4 The quoted phrases are all from Blackbourn (1997), 426, 433, 446, 452, and 459. Thisis not a work in the Kehrite tradition as it is normally understood, but it is indicative ofthe way that Kehrite ideas have penetrated the historical mainstream. Works which dis-cuss foreign policy issues in isolation from the domestic context are rare.

5 Gerschenkron (1962); the main pioneer industrializing country was, of course, GreatBritain, but France also experienced a prolonged period of relatively slow industrialdevelopment, as emphasized by Keyder and O’Brien (1978).

6 One study which has made use of insights from development economics in theanalysis of inequality trends in nineteenth-century Germany is Dumke (1991).

Industrialization under Labour Surplus Conditions 3

from the Weld of development economics. The concept of development

economics as a separate branch of economics rests on the idea that

‘relatively backward’ countries face diVerent problems and opportun-

ities to those which arise in more mature economies, or in societies

which have a more drawn out period of transition.

Much of the analysis which will be presented in this study makes use

of ideas whose origins lie in development economics: the ‘Kuznets

Curve’ relating income inequality to industrialization; the ‘Lewis Model’

of development under labour surplus conditions. Many societies have

found industrialization to be a disruptive process, and the path to matur-

ity has often been a rocky one even for those societies which have

successfully made the transition. The problems which Germany faced

have reoccurred in other contexts.

The historical debate over the pace of political progress relative to

economic development in Imperial Germany has a modern counterpart

in arguments over the role of authoritarian governments, or ‘Asian

values’, in the success of the East Asian ‘newly industrializing countries’

(or NICs). Developmental dictatorships such as the Park regime in

South Korea or the Kuomintang government in Taiwan have presided

over the initial stages of rapid industrialization, and political liberaliza-

tion has come later, when the most disruptive phase has been com-

pleted.7 It has been argued, by the World Bank amongst others, that an

‘insulated bureaucracy’ is an important component of a successful long-

term growth strategy.8 If so, then the alleged backwardness of the

German political system may have been advantageous in economic

terms: the state was able to focus on long-term goals rather than being

obliged, by the pressures of democratic politics, to resort to short-term

expedients which had damaging long-term consequences.9

There are now a number of examples of societies which have had

authoritarian governments during the early stages of industrialization,

and which have successfully managed the transition to democracy once

this phase is past. Besides the Korean and Taiwanese cases just cited,

there are other examples: Chile post-Pinochet, Spain after Franco, the

7 The role of authoritarian states in economic development in East Asia has beenexamined in a number of works; see for example Amsden (1989), Amsden (2001), andWade (1990).

8 World Bank (1994), 167.9 Many Third World countries have a problem of ‘urban bias’ as a result of policies

which were intended to improve urban conditions, but which then create privileged urbanelites.


re-establishment of democracy in Brazil following military rule in the

1960s. Why were these societies able to make this transition when

Germany could not? It cannot, surely, be argued that social and polit-

ical tensions were any greater in Germany in the 1880s and 1890s than

in Chile at the time of the coup against Allende, or Spain just before the

outbreak of the civil war.

There were, undeniably, particular factors in the case of each of these

societies which helped to cause a reduction in the potential for social

and political conXict as industrialization proceeded. This study does not

propose a general theory of politics and industrial development. But

one particular factor is that the process of transition was not derailed by

any major external catastrophe. These countries did not face genuine

strategic dilemmas which in turn impinged on domestic politics. They

were not required to work out solutions to internal problems of transi-

tion at the same time as they dealt with foreign policy issues, which

were themselves moving towards a crisis for independent reasons.

This alternative view of pre-1914 Germany stresses the essential

autonomy of the foreign policy problems which beset the German gov-

ernment. In this ‘anti-Kehrite’ formulation, ‘revolutions from above’ are

commonplace; the processes of economic and political advance do not go

hand in hand but follow diVerent trajectories, so that ‘incomplete mod-

ernization’ is the normal state for an industrializing society; Imperial

Germany was not moving towards an internally generated catastrophe,

but was a society with as good a chance of achieving full economic

maturity, social modernization, and political democratization as any

other. The decisive factor in the equation, which shifted German history

onto a diVerent course, was the outbreak of war in 1914.10

2. M igration and Inequality in the

Development Process

Migration has a key role in industrialization, and it is a principal cause

of the social dislocation which accompanies the early stages of economic

growth in modern industrial societies. This will inevitably be a period

of high migration. It is the opportunity to shift labour out of low

productivity sectors such as agriculture which makes it possible to

10 Arguments about the outbreak of war are summarized in Pogge von Strandmann(1988), see also the collection edited by Schollgen (1990).


achieve high rates of growth in underdeveloped economies. Indeed, one

of the deWning features which distinguishes ‘mature’ from ‘developing’

economies is that, typically, economic growth in the latter is accompan-

ied by large-scale migration from the countryside into cities, while in

the former it is not.

One view which provides a simple analysis of the eVects of migration

out of agriculture is the model of economic development Wrst set out byArthur Lewis in 1955: ‘a model of development with unlimited supplies

of labour’, generally known to development economists as the Lewis

Model.11 In his article, Lewis analysed the eVects of industrial develop-

ment in the context of a labour market where there is surplus labour, as

commonly found in countries in the early stages of industrialization.

The model predicts that, so long as capitalists can draw on these

supplies of surplus labour, then the fruits of industrial progress will go

largely to the holders of capital, in the form of rising proWts, and there

will be little increase in the level of real wages. Given that the owners of

capital are likely to be predominantly rich, and few in number, it can be

seen that this prediction will lead to rising inequality in cities and other

industrial areas, with a consequent increase in social divisions, and, also,

to a feeling amongst industrial workers that they are not gaining a fair

reward for their labours.

For students of German history the attraction of the Lewis Model is

that it provides a possible explanation both of the rapid growth of the

German economy 1870–1914 and of the tensions that accompanied this.

Moreover it also suggests a reason for one of these problems: the rise of

the socialist vote. Lewis named the two phases he described ‘classical’

(when there is a labour surplus and wages are held down) and ‘neo-

classical’ (when the labour surplus is exhausted and wages start to rise).

But one might note the similarity between his model and that of Karl

Marx and call the Wrst phase ‘Marxist’ or ‘quasi-Marxist’. According to

the model, this is a period when the capitalist economy behaves in a

way which seems to bear out some of Marx’s predictions: proWts rise as

a percentage of the national income; wages lag behind productivity

increases.12

11 Lewis (1954). For a modern discussion see Hayami (1997), 72–5 and 170–2.12 Lewis did not assume, as Marx did, that technology was always labour-saving. In a

purely Marxist model capital accumulation leads to increased competition, a fall in therate of proWt, and increasingly severe economic crises, so the ultimate level of the share ofproWts in national income is not clearly determined. However, Kolakowski has pointed outthat Marx’s views on whether capitalism led to the absolute or the relative impoverish-ment of the working class are also ‘by no means unequivocal’, Kolakowski (1978), i. 289.


If the economy behaved as Lewis predicted, then the experience of

workers competing in urban labour markets with apparently unlimited

supplies of migrants from rural areas would be a dispiriting one. The

formation of trade unions could lead to an improvement in workers’

bargaining strength, but unions would have to be cautious about press-

ing claims for higher wages given the ability of employers to bring in

large numbers of migrants with no previous experience of trade union

activity. Trade unions would be more eVective where the workers they

represented possessed skills which could not easily be taught to newly

arrived migrants, as was the case for printers and other members of the

so-called ‘labour aristocracy’. These would be easier to organize.

In this context, the appeal of radical Marxist ideas would be under-

standable, together with a pessimistic view of the likely beneWts from

non-revolutionary trade union activity. But the diVerence between

Lewis’s model and Marx’s is that Lewis’s labour surplus phrase comes

to an end eventually. When this happens, urban labour markets are no

longer dominated by the Xow of surplus labour out of agriculture, and

employers who need to attract labour to new industries are obliged to

oVer superior wage levels to those obtained by fully employed workers

elsewhere. Trade unions can bargain from a position of strength and

union membership can oVer beneWts to unskilled as well as skilled

workers. The economy enters a new phase in which wages rise in line

with productivity and the proWt share stabilizes or may even decline.

The operation of the Lewis Model is illustrated in Figure 1.1. The

introduction of new technology causes a rise in productivity in

the advanced or industrial sectors. But with unlimited labour supplies,

the wage level is Wxed. Consequently, the gains from productivity in-

creases go exclusively to the owners of capital, producing a rise in the

proWts share. It is only when the labour surplus is exhausted—a

moment sometimes called the ‘turning point’ in the development eco-

nomics literature—that wages can rise above the Xoor set by the ready

availability of surplus labour in the traditional or rural sectors.13 Once

this has occurred the economy shifts to a new phase of development,

represented by the neo-classical model of mainstream economics, where

there are limitations on the supply of labour as well as capital. In this

phase, both capital and labour share in the fruits of economic develop-

ment. In mature capitalist economies real wages have tended to rise at

least as fast as the growth of productivity, if not somewhat faster. This

13 The term ‘turning point’ was coined by Fei and Ranis (1964).


has meant that the share of wages in national output has been stable or

tending to increase.14

The implication of the Lewis Model is that this historical moment, the

‘turning point’ as the economy moves out of the labour surplus phase,

represents a new opportunity to achieve political and social stabilization.

There are now reasons for workers to favour ‘normal’ trade union activ-

ity, aiming to secure steady improvements in wages and other working

conditions, over more radical alternatives. ‘Revisionist’ views should gain

ground. Provided that the political system oVers a reasonable chance of

similar progress, leading to increased participation in political decision-

making, and that the whole system is not destabilized by some major

catastrophe, a society which has passed the ‘turning point’ of the Lewis

Model has a much better prospect of achieving full political, social, and

economic maturity—becoming a ‘normal’ modern, liberal, capitalist

democracy—than one which is still in the labour surplus phase.

The Lewis Model in this pure form has a clear and abrupt transition

from the labour surplus to the neo-classical model. In practice,

economic development does not divide quite so easily into phases.

14 The Wgure shows the eVect on factor shares if wages rise in line with productivity.This is not an automatic consequence of the transition to a ‘neo-classical’ economy,although it is broadly consistent with the experience of most OECD countries in thetwentieth century.

Turning point

Productivity

Profits

Labour surplus model Neo-classical model

Wages

Fig. 1.1. The pure Lewis Model: changing factor shares

during industrialization


Moreover, there is generally some rise in wages during the early stages

of industrialization, even if the rise is not as fast as the increase in

productivity. Initially, industry can rely on nearby sources of labour.

But once this ‘local labour surplus’ is exhausted, wages have to be

raised to attract migrants from regions which are more remote from the

main industrial areas. This period of slow wage increases, which lag

behind rising productivity, can last until all sources of surplus labour

are used up. Then wages start to rise in line with productivity.

Figure 1.2 illustrates this ‘modiWed Lewis Model’. In the early stages

of industrialization the beneWts still go predominantly to the owners of

capital, even though there is some increase in wages. The transition to

the neo-classical model is a more gradual one. Instead of a sudden

‘turning point’ there is a slower transition to the neo-classical model.

What evidence is there that wages lagged behind productivity? The

actual record of German wage and productivity growth is set out in

Figure 1.3. There are some caveats to note. The Wgures used for these

calculations are taken from Walter HoVmann’s Das Wachstum der

deutschen Wirtschaft. The wage series used is derived from a combin-

ation of diVerent sources, and this means that it must be treated with

some caution.15

Productivity

Profits

Wages

Labour surplus model Neo-classical model

Fig. 1.2. The modified Lewis Model: changing factor

shares during industrialization

15 More reliable wage data becomes available from the 1870s and 1880s onwards, Bry(1960), Desai (1968), and Hohls (1995). However, the onset of German industrialization is


The second caveat is that wages may lag behind productivity for

other reasons. If the terms of trade move against the industrial sector,

then factor returns will not keep pace with productivity.16 In the period

after 1895, the German export sector had to gain market share from

Britain, and this led to downwards pressure on industrial prices at a

time when world agricultural prices were recovering from the low point

generally dated to the 1850s, and Figure 1.3 shows that the gap between the two seriesopened mainly in the 1860s and early 1870s. It would also be interesting to compare realwages and the return on capital in the industrial sector, but HoVmann’s proWt Wgures havecome in for heavy criticism, Fremdling (1991).

16 The diVerence is that, with adverse movements of the terms of trade, factor returnsin the industrial sector are depressed in general, so both proWts and wages are held down,but under the Lewis Model only wages are held back.

1850

80

90

100

200

300

400

1856 1862 1868 1874 1880 1886 1892 1898 1904 1910

Real wages

NVA/head

Fig. 1.3. Indices of productivity and real wages in the advanced sectors of the

German economy (industry, mining, and transport), 1850–1913

Log scale; base is average for the years 1850–9 (¼100)

Source: Calculated from Hoffmann (1965). The series are averages for the three sectorscalculated from Hoffmann’s figures for employment, net value added, and wages. Thereal wage figures are produced using Hoffmann’s deflator for private consumption.


of the mid-1890s. This shift in the terms of trade meant that the im-

pressive productivity performance of the German industrial sector in

the years just before the First World War was not fully reXected in real

wage gains.

Figure 1.3 shows that, according to HoVmann’s Wgures, real wages

did not keep pace with productivity in the modern sectors. Between

1850/9 and 1904/13 productivity rose by 202%; real wages rose by

119%. The rate of increase of productivity was 70% higher than the

rate of increase of real wages. The Wgure shows that the gap opened

substantially in the 1860s and early 1870s. It then closed in the 1880s

before widening again in the 1890s and 1900s.

The fact that so much of the widening of the gap occurred in the

1860s, which is a period for which there are few reliable wage statistics,

means that these results need to be interpreted with caution. What can

be said is that these are Wgures which are consistent with the modiWedversion of the Lewis Model as shown in Figure 1.2. Wages did not keep

up with productivity; most of the beneWts of industrialization went to

the owners of capital.

3. Towards a More General Theory of Inequality

and Industrialization :

The Weber–Lewis–Kuznets Hypothes is

The Lewis Model is not a complete model of inequality and industri-

alization: it lacks several important components. It does not explain how

the labour surplus is created, and how this phase comes to an end. It

provides a possible explanation of rising urban inequality, but to explain

the general movement of inequality in an industrializing society it is

necessary to widen the scope of the enquiry to include consideration of

rural inequality and the relationship between the two sectors.

Lewis’s view of the labour surplus is essentially a static one: less-

developed countries are overpopulated relative to economic opportun-

ities; there is underemployment both in agriculture and in an informal

urban sector.17 His views are, in some respects, very similar to those of

Max Weber, writing about conditions in east-Elbian agriculture in the

17 In societies which do not have comprehensive unemployment insurance there israrely much formal unemployment. Labour surplus conditions are more likely to Wndexpression in high levels of underemployment.


late nineteenth century. Lewis wrote: ‘in over populated countries the

code of ethical behaviour so shapes itself that it becomes good form for

each person to oVer as much employment as he can’.18 Weber held that

‘the old economic order asks: ‘‘how can I give, on this piece of land,

work and sustenance to the greatest number’’ ’.19 Both these statements

describe a system of values which gives rise to behaviour which departs

from the proWt-maximizing norm.

Lewis did not try to explain why this changed, but Weber did. The

most important factor was the changing attitudes and values of the

land-owners. These were obliged by their desire to maintain their pos-

ition in society to match the consumption patterns of the rising urban

bourgeoisie, and this in turn led to the adoption of entrepreneurial,

proWt-maximizing behaviour.

Under capitalism, entrepreneurial land-owners would employ the

cheapest labour, use the best technology, and prefer market relation-

ships to paternalistic systems of mutual obligation. Thus, while the ‘old

economic order’ provided employment for surplus labour, this surplus

was released when capitalist values came to predominate.

In Weber’s view, therefore, it was the permeation of capitalist values

throughout society which led to the creation and release of the rural

labour surplus. This was an inevitable consequence of industrialization.

While Weber’s views on the exact mechanism which brings about

social change in rural societies during industrialization can be disputed,

his basic point that there are institutional arrangements in traditional

agriculture whose erosion during industrialization can create a period of

heavy migration into urban labour markets is more in tune with modern

analysis. This could arise for a number of reasons. Traditional marriage

patterns may have held back demographic increase. There may be insti-

tutionalized underemployment in agriculture, supported by systems

favouring revenue sharing or informal arrangements for mutual insur-

ance in hard times. Institutions which were appropriate to an isolated

agricultural sector with limited access to urban markets may hold back

the use of new technology: improved transportation may bring modern

inputs and a consequent increase in productivity, making it possible to

18 Lewis (1954), 142.19 The quotation is from a lecture Weber gave in the United States, Weber (1948), 367.

Weber’s major work on east-Elbian agriculture formed part of a survey by the Verein furSozialpolitik, Weber (1893). Other writings on this subject include an article whichappeared in 1894, Weber (1894, trans. 1979) and various speeches and addresses, whichcan be found in Weber (1993).


shift large quantities of labour into the industrial sector while maintain-

ing or increasing food production.20

A more general view of the relationship between industrialization and

inequality in all sectors of society was provided by Simon Kuznets,

writing, like Lewis, in the 1950s.21 He was inXuenced by the work of

the Soviet economist Prokopovitch, who had used nineteenth-century

Prussian tax data to show a rise in inequality during industrialization.

As a result of Prokopovitch’s work, Kuznets was also aware of a gap

between rural and urban inequality in nineteenth-century Prussia. He

found a similar pattern in studies of inequality in the United States and

India.22 This led him to a general statement of the relationship between

inequality and economic development.

His hypothesis was that the shifting balance between a relatively

egalitarian, but poor, rural sector, and a relatively unequal, but richer,

urban sector would produce a curve, an ‘inverted U’, as inequality Wrstrises with industrialization, and then declines (Figure 1.4).

Migration has an important role in the mechanism which, according

to Kuznets, produced his curve. It was migration between the two

sectors which produced the shifting sectoral balance. Moreover, this

eVect would be intensiWed if the initial eVect of industrialization was to

raise the size of the rural–urban wage gap. Modern studies have shown

that the movement of the rural–urban gap is an important factor in the

overall income distribution: large gaps between rural and urban earn-

ings are associated with high levels of overall inequality.23 Poorer rural

workers and peasants account for a disproportionate number of the total

numbers on low incomes even in advanced countries, so a worsening of

the relative position of agriculture will tend to worsen the overall distri-

bution of income.

This factor in itself could explain part of the increase in inequality

predicted by Kuznets. If modern industrial centres are to grow rapidly

they will need to attract labour from rural areas. Initially this may come

from nearby regions, where there may be surplus labour. But, when this

surplus is exhausted, sources of labour must be found in further-oV

20 The view that many of the traditional institutions of agriculture in underdevelopedcountries are eVective solutions to the problems of remoteness, uncertain yields, andlimited access to markets is put forward in Dasgupta (1993).

21 Kuznets (1955).22 Ibid. 7; the German Wgures came from Prokopovitch (1926).23 Bourguignon and Morrison (1998), Bourgignon (1990); Basu (1997) discusses the

theory behind this.


regions. This may require a substantial increase in the gap between

rural and urban earnings to create the necessary incentives for would-be

migrants.

Although Lewis and Kuznets had diVerent mechanisms in mind,

there are similarities between their views. The downward phase of the

Kuznets Curve is brought about by two factors: the declining import-

ance of the poorer rural sector in the overall distribution of income

(which reduces the importance of the gap between rural and urban

incomes as a contribution to general inequality) and a tendency for

inequality within the urban sector to start high and fall over time. This

last point can also be derived from the Lewis Model.

Putting the views of these three authors together produces a general

theory of the relationship between economic development and inequal-

ity: industrialization and the spread of modern values and relationships

unleashes a process of social change in agriculture, which in turn leads

to the transfer of surplus labour to the urban sector; labour surplus

conditions produce an unequal distribution of income in the urban

sector; the increased importance of the unequal urban sector leads to a

general rise in inequality; the ending of the labour surplus period

creates conditions which will eventually lead to a fall in inequality.

Inequality

Gross Domestic Product per capita

Fig. 1.4. The Kuznets Curve: an inverted U-shaped relationship between

inequality and economic growth during the early stages of industrialization


4. Crit ic ism of the Lewis Model and Alternative V i ews

The Lewis Model has been widely used in development economics, and

it is still one of the mainstays of this branch of economic analysis, but it

has had its critics.

1. In the development economics literature the aspect of the Lewis

Model which has attracted most criticism is the view that agriculture

has only a passive role in economic development, and does not progress

technologically in the same way as the urban or industrial sector.24 One

of the most serious charges against the Lewis Model is that it has

contributed to the tendency of modern developing countries to under-

invest in agriculture.

2. Lewis makes a restrictive assumption about savings: ‘only capital-

ists save’. While this assumption may be justiWed for some developing

countries where banks and other savings institutions are not well estab-

lished, the experience of other industrializing countries shows that,

given the right institutional framework, even people who are not par-

ticularly well oV will save, and that these savings can be an important

component of total funds available for investment. A more realistic

assumption might be that ‘only capitalists invest in industry’: meaning

that, in the early stages of industrialization, only a minority have the

specialized knowledge of the new technology to be able to take advan-

tage of the opportunities it oVers. This, in turn, points to the import-

ance of institutional development, in making it possible for the savings

of those who lack technical knowledge to be channelled into industrial

investments: the German banking system performed such a function.

3. Lewis has been accused of overemphasizing the importance of

agriculture as a source of surplus labour. Surveys of modern developing

economies show that the main area of underemployment is in fact the

informal urban sector. Whether this also applies to nineteenth-century

Germany is another matter. The phenomenon of high rates of urbaniza-

tion, in countries which are still in the early stages of industrialization,

is one which is found in post-1945 developing countries and not in

nineteenth-century Europe. But an informal urban sector did exist:

Tagelohner, workers without permanent employment dependent on

occasional day-labouring jobs, were a signiWcant part of the urban

labour market.

24 Industrialization and urban growth produces change in agriculture in nearby regions,as was pointed out by Von Thunen (1826).


4. In addition, there has been criticism of the concept of the labour

surplus itself. When, in 1979, Lewis was awarded the Nobel prize for

economics, it was shared with the Chicago agricultural economist Theo-

dore Schultz. Both had made a range of important contributions to

development economics, but on this subject they were in fundamental

disagreement: Schultz denied the existence of a labour surplus in trad-

itional agriculture, certainly in Lewis’s sense of a large body of labour

with zero marginal productivity.25

The essence of Schultz’s critique was, Wrst, to deny that resource allo-

cation was ineYcient in traditional agriculture, or that there was a prefer-

ence for idleness or a bias against thrift. From this it followed that labour

would be fully occupied to the point where marginal cost was equal to

marginal product. Secondly, he made use of empirical studies to show

that underemployment was not prevalent, and also demonstrated, using

the eVects of the 1918–19 inXuenza epidemic in India, that an exogenous

removal of agricultural labour did aVect food production.26

In Schultz’s view peasants were ‘poor but eYcient’.27 If investment

was low it was because the return on capital was also low, or the

discount rate was high. It was expensive for small subsistence farmers

to search for innovations; this, rather than resistance to change,

explained low technical progress. Peasant ineYciency was a myth im-

posed by Western agricultural experts who failed to understand the

seasonal nature of much work in traditional agriculture.

Although Lewis may well be the better known, or at least the more

cited, author, Schultz’s critique has been highly inXuential. The view

that peasants Wnd eYcient solutions to the problems they face is now a

commonplace. Moreover, Schultz’s stress on the importance of the

introduction of new factors—new productive knowledge, better educa-

tion, human capital in the broadest sense—appears prescient in the

light of subsequent developments in growth theory.

Nineteenth-century German agriculture oVers an excellent example

of the introduction of new, non-traditional factors exactly as described

by Schultz: new scientiWc knowledge, heavy investments in education,

and specialist training in agricultural sciences.

But before setting up Lewis and Schultz as polar opposites it should

be noted that in some respects they are not so far apart. Schultz thought

25 Possible deWnitions of the labour surplus are discussed in Reynolds (1975), 11.26 Major wars have also led to the sudden removal of large numbers from the labour

market; Prest (1948) gives some interesting examples of cases where this led to a fall ofproduction in less-developed economies, and some others where it did not.

27 Schultz (1964), 40.


that the introduction of new factors could produce a relatively ‘cheap’

rise in productivity: the transformation of traditional agriculture could

be quite rapid. Now there is a clear distinction between Lewis’s view of

a static rural labour surplus, which pre-dates industrialization, and

Schultz’s more dynamic version, but the practical implications are simi-

lar: quite large amounts of labour can be released from agriculture while

food production is maintained or increased.28

Moreover, it is worth considering what the source of these new

inputs might be. Government investments in agricultural science and

education in nineteenth-century Germany were not the whole story.

Industrialization brought new agricultural inputs: basic slag from steel-

making; ammonium sulphate from gas production. Industrial progress

stimulated scientiWc research as well as beneWting from it, and this

created possibilities for agricultural advances in turn. Industrialization

raised the demand for agricultural products, some of which, such as

sugar beet, also required improved technical knowledge. Industrial

growth led to improved communications, bringing to rural society

better access to markets and to productive knowledge. Institutional

developments, such as improved banking systems, also aVected agricul-

ture. So, the transformation of traditional agriculture was, at least in

part, a consequence of industrialization.

Although Schultz and Weber came from diVerent traditions and

were writing in diVerent centuries, there is a similarity in their views in

that the creation of an agricultural labour surplus is a dynamic process,

which comes about as a result of industrialization and the eVect that

this has on agriculture. Weber’s primary stress was on the changes in

social attitudes but he also considered the eVect of changing technology,

in particular the introduction of sugar beet. This increased the seasonal-

ity of agricultural work, leading to the employment of seasonal migrant

labour, which in turn brought about a weakening of social ties.

5. The Lewis Model and the H i storiography

of Imperial Germany

Historical studies are mostly bounded in scope and in time, and this

one is no exception: July 1914 makes a convenient date for the ending

28 Ibid. 24–5. This eVect is greatly strengthened if, as is often the case, industrializationis accompanied by increased population growth, an issue which will be examined inChapter 5.


of a study of conditions inside Imperial Germany. The year 1913 is the

last one for which a full set of data on peacetime conditions is available.

But the temptation to see the events of 1914–18, not just the war but

the revolutionary outbreaks which accompanied its ending, as the cul-

mination of forces and developments already under way within the

social structure of the Kaiserreich needs to be resisted. The sense that

the narrative structure needs a natural full-stop, not an accidental one,

pervades much that has been written about Imperial Germany. The

massive disruption of German political and social development which

followed is felt to need deep-seated explanations, not just the impact of

chance events. Hence the appeal of Sonderweg arguments which trace

the evolution of Xaws rooted in the character of the Bismarckian state

and the social or cultural background to its political travails.

The current consensus is not one to accept the Kehrite view of the

absolute primacy of internal politics in foreign policy without reserva-

tion. Instead the view which prevails is one which stresses the inter-

action of external and internal politics: the way that interest group

politics hemmed and hampered policy-making in both Welds; the con-

stricting eVect of unresolved constitutional problems on government

action; the eVect of a drift to radicalism and chauvinism on the tenor of

debate and the ability of political actors to see issues as resolvable, given

goodwill and a willingness to engage in rational analysis, rather than

unbridgeable divisions, calling for recourse to apocalyptical solutions.

The predominant view sees the intertwined problems of the Kaiserreich

as a Gordian knot which, by implication, only a radical discontinuity

could solve. Historians who take a positive and optimistic view, who

stress the real prospects for peaceful evolution without violent confron-

tation, have had a poor reception from their colleagues.29

The value of economic models to historians lies in their ability to

draw together diVerent pieces of evidence and present a coherent pic-

ture of social or economic evolution. The common features of the ex-

perience of many societies engaged in the process of industrialization

has naturally excited the interest of theorists and led to the construction

of theories which seek to explain these consistencies. While some his-

torians may distrust these attempts to impose a common structure on

historical events, the alternative approach, a history without theory,

leads to the particularization of every historical event, and an undue

29 See, for example, the reaction to Rauh (1973) and (1977).


stress on the unique features of every society, ignoring many points of

similarity.

Migration is a particularly good example of this tendency to see

events without considering the broader context. It is a subject which

has rightly stimulated historical analysis. The movement of individuals

or families from one social and economic context to a diVerent one is a

dramatic historical event which brings about the disruption of trad-

itional relations, and leads to the creation of new ones. But it also forms

part of a larger picture, playing a role in the determination of wage

levels and other factor returns in the economy as a whole. This in turn

has an eVect on attitudes and motivations at the level of the individual.

For a society in the upwards phase of the Kuznets Curve, a drift to

radicalism is not inevitable, but the context is one which is, at the very

least, conducive to alienation and a sense of social exclusion. The diver-

gent paths of factor returns predicted by the Lewis Model, when trans-

lated into the actual experiences of workers exposed to the movements

of urban labour markets under labour surplus conditions, meant that a

generation of workers had poor living conditions, low life expectancy,

and insecure jobs and saw only slow improvements in these circum-

stances, despite rapid economic growth.

But these were not German peculiarities; they were the result of

general forces unleashed by the introduction of modern technology.

And as Germany began to climb out of the trap created by the labour

surplus phase of economic development, conditions in German cities

did improve, albeit slowly. This in turn created a more favourable

background for social reconciliation and political progress. The Lewis

Model suggests that the ending of the labour surplus phase may well

represent an opportunity for a wider process of political and social

modernization.

6. Standards of Comparison for the H i storiography

of the Kaiserreich

A major concern of historians has been the extent to which Imperial

Germany was an unusual state or society in the context of late nine-

teenth-century Europe. The view that there was something exceptional

about Germany takes diVerent forms: Wolfgang Mommsen has placed

the emphasis on Xaws in the political system; for Hans-Ulrich Wehler it

is the Janusgesicht of incomplete modernization, which raises broader


questions of social and economic development. But one feature of these

studies is that the standard of comparison is taken from the experience

of the states of Western Europe: Wehler has written about choices

between ‘Modernisierung und Tradition . . . okzidentalen Gemeinsam-

keiten und deutschen Sonderkeiten’.30

The implication of the Weber–Lewis–Kuznets hypothesis is that this

comparison may not be appropriate. In the pioneer industrial states the

process of industrialization was more drawn out, the erosion of pre-

industrial, or pre-capitalist, values and institutions was slower.31 There

may have been a labour surplus period, when labour supplies were

abundant due to population growth or changes within agriculture, but

this was less acute and spread over a longer period. The growth of cities

was not so rapid. Patterns of inequality changed slowly, if at all. Polit-

ical and social institutions could evolve in situations where there was

less pressure for sudden change.32

In the British case, the existence of labour surplus conditions has

been detected in the operation of the labour market in the period up to

1850.33 This was the result of a relatively drawn-out process: compara-

tively slow industrial growth combined with demographic pressure and

the Wnal stages of the transition to large-scale agriculture which had

started in the sixteenth century. But the German experience was a

diVerent one. Although there had been pockets of proto-industrial ac-

tivity in the eighteenth century, the onset of rapid economic growth

was delayed until the ‘triple uniWcation’ of the mid-nineteenth century:

economic uniWcation with the creation of the Zollverein in 1834; polit-

ical and monetary uniWcation with the founding of the German Empire

in 1871; and, crucially for a country with limited access to sea transport,

the creation of a uniWed market as a result of the expansion of the

railway network in the 1850s and 1860s.

30 Modernisation or Tradition . . . Convergence to Western Norms or German Exceptional-ism, Wehler (1995), 1252; the use of the adjective okzidentalen obscures which Westernstate is being used as the standard of comparison. Since the strictures of Blackbourn andEley (1984) appeared, German historians of the Sonderweg persuasion have been lessspeciWc on this point, although it is often Britain that is the implied model.

31 In the British case, the switch to large-scale capitalist agriculture came well beforethe onset of industrialization.

32 The main focus of comparison is obviously Britain, but Belgium and the Nether-lands were also advanced industrial or commercial societies by 1830. Recent scholarshiphas also shown that the process of French industrial growth was a relatively drawn-outone.

33 As argued in Feinstein (1998), 651.


German economic development was, like that other of Continental

countries, ‘jerked’ forwards by the application of British techology.34 In

the German case, the jerk was made even more abrupt by the overcom-

ing of barriers to internal market integration which had hitherto held

back economic development. The result was that German industrializa-

tion from 1850 onwards was unusually fast. The process was telescoped

in a way which was new in the experience of other European countries.

This created problems of adjustment and adaptation which were unpre-

cedented in their intensity.

If so, then the standard of comparison for Imperial Germany should

not be taken from Western Europe but, in the Wrst instance, from other

late-industrializing European countries—Russia, Italy, and Spain, for

example—and secondly, from other non-European countries which

have gone through the process of industrialization. There is a large

range of scholarship which has been devoted to the comparative study

of developing economies, and the diVerent social, economic, and polit-

ical problems associated with or resulting from the development pro-

cess. It is the basic contention of this book that the tools developed

within this Weld, the Weld of development studies, can be used to under-

stand and elucidate the problems of Imperial Germany. This will, it is

hoped, shed a new light on the diYculties, and the achievements, of the

Kaiserreich.

34 Pollard (1981), 84. The transforming nature of the new British industrial technologyis clear from the regional studies presented in chapter 3 of Pollard’s study.


2

Sources of Inequality in Rural

Germany

1. Introduction

The Kuznets Curve—the hypothesis that inequality will initially rise

and then fall with industrialization—is predicated on the basis of a

series of assumptions made by Kuznets in his original article. One of

these is the assertion that, initially, inequality will be higher in the

urban than the rural sector.

However, consideration of the historical record, as well as the current

position in developing countries, shows that there are a wide range of

possible levels of inequality in the rural sector in underdeveloped or

newly industrializing countries. Factors such as climate, crop specializa-

tion, the eYcacy of land reform, the legacy of colonialism or feudalism

can produce very diVerent outcomes. The contemporary situation

shows wide divergences. On the one hand there are the ‘plantation

economies’ of central and southern America, which typically have large

inequalities in land ownership and rural incomes; at the other extreme,

the rural sectors of many newly industrializing East Asian countries are

characterized by small-scale intensive rice production, with a conse-

quent narrowing of disparities in income and wealth.1

Thus, a study of the evolution of inequality in an industrializing

society needs to start with a consideration of the state of rural society:

the allocation of property rights, the distribution of land holding and

land ownership, the position of rural labourers. Even if the predictions

of the Lewis Model concerning conditions in urban labour markets are

correct, and this leads to high inequality in the urban sector, there

are some societies in which the level of rural inequality is so high that

industrialization produces little or no increase in the overall level of

inequality.

1 Works which have stressed the importance of such geographical factors include Bray(1986) and Engerman and SokoloV (1997).

2. The Agrarian Reform Movement in

N ineteenth-Century Germany :

General Considerations

Historical analysis of the nineteenth-century agrarian reforms has

tended to look for statistical indicators to judge the impact of the

reforms: Wgures for peasant (or smallholder) land losses, information on

the numbers of landless labourers and their incomes, the extent of

peasant indebtedness, the overall structure of land holding before and

after the reforms, the balance between rented land and land in owner-

occupation. These are important issues, and it is right that they should

be the main focus of research, but the use of statistical indicators is

meaningless without ‘benchmarks’, standards by which to judge the

diVerence between success and failure.

Much of the modern historical analysis of the agrarian reforms has

been inXuenced by the ‘incomplete’ or ‘absent’ revolution thesis of the

Sonderweg school of German historians: the view that the deWning char-

acteristic of German history has been a succession of ‘revolutions from

above’, which necessarily lacked the capacity to achieve thoroughgoing

social transformations of the type supposedly achieved in other coun-

tries which experienced ‘revolutions from below’. Implicitly, therefore,

the standards of comparison for the agrarian reforms are the revolution-

ary land settlements brought about by the French or Russian revolu-

tions (referring in the latter case to the pre-collectivization period).

These are possible points of comparison, but not the only ones.

There is now a large number of studies in the Weld of development

economics on land reform and its problems. This has a number of

lessons for historians. Perhaps the most important of these is the sheer

diYculty of land reform: it is quite rare to Wnd examples of land reform

programmes which have been even moderately successful. Agrarian

reforms have too often been either ineVective or else led to results

which were inimical to the development process.2

Land reform is a social and political process which places an unusual

strain on the capacity of the political system to deliver an unbiased

outcome: one which is not inXuenced either by the social composition

of the bureaucracy or by the ability of special interest groups to aVect

political decisions. It is a test of the ‘encompassing’ nature of the polit-

ical process: the extent to which decision-makers put the long-term

2 Modern works that make this point include Low (1996) and Ghimure (2001).

Sources of Inequality in Rural Germany 23

interests of society, considered as a whole, ahead of considerations of

short-term expediency (which militate against oVending powerful social

groups) or the favouring of sectional interests.3

Moreover, it is a process which involves balancing a number of

diVerent objectives. The history of land reform shows that there are

many pitfalls. These include the following.

(i) The eVect of expropriation on property rights

Radical land reform without compensation can achieve a dramatic alter-

ation of rural property rights. However, expropriation on such a scale

will have adverse eVects on investment in other sectors: investors will

wonder if similar violations of property rights are possible in other

areas. This is one consideration which has led most governments to pay

compensation to landlords and others adversely aVected by land

reforms. The record of recent years shows that some compensation has

nearly always been paid: the main exceptions being land reforms

following the establishment of Communist governments which have

been coupled with more general restrictions on private property rights.

In the German case, compensation was paid by the peasants them-

selves, either in the form of monetary compensation or land given up to

landlords, in order to secure the elimination of labour services and

other onerous duties or restrictions. Arguably, the various German

states could have provided more of these payments themselves, but the

limited size of state budgets in the early nineteenth century probably

precluded this option.

(ii) The eVect on production

The objectives of land reform are Wrst to reduce rural inequality, and

secondly to promote increased agricultural productivity, the use of new

technology, and rising food production. Otherwise, inadequate agricul-

tural production may act as a brake on industrial development, produ-

cing problems such as the ‘scissors crisis’ of the Soviet economy in the

NEP period.4 Land reform should not create a rural sector which is an

obstacle to general economic development. It should not consist entirely

of subsistence farmers unable to produce a marketable surplus.

3 The concepts of ‘encompassing’ and ‘sectional’ interest groups are taken from Olson(1982).

4 Nove (1969), 93–6.

24 Sources of Inequality in Rural Germany

The impressive productivity performance of the German agricultural

sector in the second half of the nineteenth century showed that the

reforms produced an eYcient market-orientated sector, which was well

able to take advantage of the opportunities presented by expanding

urban demand.

(iii) The importance of rural credit

A major theme which connects to this last point is the availability of

rural credit. In cases where land reform has had an adverse eVect on

production, the lack of credit has often been an important factor. More-

over, if peasants are obliged to pay compensation to former landlords,

then some means of Wnance will be required for this. In cases where

rural Wnance has been unobtainable or prohibitively expensive, there has

often been a return to landlord control, as indebted peasants have been

forced to sell land to local money-lenders or back to former landlords.5

This is probably the area where the Prussian reforms, in particular,

were most deWcient. In the early nineteenth century, state-sponsored

rural credit institutions, such as the Landschaften or rural mortgage

banks, lent mainly to larger estates. Smaller farmers had limited access

to credit, and this contributed to the extent of peasant land losses. By

contrast, in areas such as the Kingdom of Saxony, where more eVort

was made to give small and medium-sized farmers access to credit, the

post-reform land distribution was more egalitarian.

(iv) The importance of market access

As the objective is to create an agricultural sector which is both egalitar-

ian and also market-orientated, it follows that the costs of market access

will be an important factor. For peasant farmers, with relatively small

marketable surpluses, high transport costs and other obstacles can make

the use of the market mechanism an unattractive proposition. The con-

sequences are, Wrst, that peasant farmers revert to subsistence agricul-

ture, which makes it diYcult to pay for modern inputs or maintain

payments on debts, and secondly, that supplies of food and other

materials from the rural sector are limited.

5 In the Philippines there have been 11 consecutive land reform programmes, but thelack of rural credit and other follow-up services has meant that land has tended to passback into the hands of landlords and agricultural companies, Ghimure (2001), 13.


Inevitably, diVerent areas of Germany had very diVerent experiences.

In the western regions the proximity of expanding urban markets made

life easier for peasant farmers who could intensify livestock production

or specialize in perishable high-value products for local markets. In the

east transport costs were high, which meant that the main marketable

crops would be cereals.6 This inevitably favoured the larger estates,

which tended to be more orientated towards cereal production. But the

elimination of internal tariVs and the creation of a single German

market, together with the expansion of the railway network, did im-

prove opportunities for farmers in some of the remoter areas.7

(v) The importance of contingency

But even the best-designed land reform is still subject to a range of

inXuences which are outside the control of the political authorities.

Two of the most important are demographic pressure, and the move-

ment of farm prices. Under circumstances where there is acute demo-

graphic pressure in rural areas, and the supply of fertile land is limited,

it will be extremely diYcult to avoid a rise in the landless population,

even with a highly egalitarian land distribution. The alternative is to

permit the subdivision of holdings to an extent which would substan-

tially reduce the marketable surplus, which would jeopardize other rural

policy objectives.

Another outside factor is the movement of farm prices. If land

reform leaves the peasant sector with an increased level of indebtedness

as a result both of compensation payments to landlords, and also invest-

ments consequent on the shift to a more market-orientated agricultural

system, then the real burden of these debts will be raised if prices fall,

and reduced if prices rise. Some notably successful land reform pro-

grammes, such as those of post-1945 Japan and eighteenth-century

Denmark, have been considerably assisted by subsequent bursts of

inXation which reduced the real value of peasant debts.

6 Which had a relatively high value-to-weight ratio; see Brinkmann (1935), 82–3 for thetransport costs of various commodities; this uses German data collected by H. Settegast inthe pre-1914 period.

7 East Prussia was an area which contained a large number of small or medium sizedholdings, and which was able to expand livestock production substantially in the latenineteenth century, moving away from wool production into more intensive livestocksystems.


Figure 2.1 illustrates this point by comparing the movement of agri-

cultural prices in Prussia after 1816, and in Japan after 1946. The lower

line gives the movement of rye prices in Prussia (1815 boundaries);

superimposed on this is a second line giving the movement of rice

prices in Japan in the years following the land reform introduced by the

American occupation authorities in the years immediately after the

Second World War. The Japanese reforms were mainly carried out in

1946–7. In the years which followed there was a burst of inXation which

carried rice prices to levels two or three times those of the reform

period. A more gradual increase in subsequent years took prices up to

an index of 500 by the late 1960s.

By contrast, the years which followed the Prussian reforms saw a fall

in prices to levels which were around a third of the 1816–18 average by

1817

40

60

80

100

200

400

600

1820 1823 1826 1829 1832 1835 1838

Prussia

Japan

Fig. 2.1. A comparison of agricultural price movements in Prussia

(rye prices 1816–40) and Japan (rice prices 1946–69)

Three-year moving averages as indices (base for Prussia is 1816–18, for Japan 1946–8).Japanese index is superimposed (1946 corresponds to 1816).

Source: Japanese prices are from various issues of the Japanese statistical yearbook,1950–70; the index is a chained index, using whatever was reported, so that the 1946–7prices are wholesale prices (Tokyo), 1947–50 prices are black market prices (Tokyo), andthe 1950–70 prices are producer prices (all Japan). The German index uses the averagePrussian wholesale prices reported in ZdKPSB (1907), 84–92.


the mid-1820s. The reforms were carried out under circumstances

which were considerably less favourable for peasant farmers than those

in post-1945 Japan.

The value to historians of the broader context provided by modern

writings on land reform is that it emphasizes the point that agrarian

policy needs to serve a range of objectives, and it must make a positive

contribution to the general objective of social and economic develop-

ment. Land reform cannot be judged according to a simple ‘black–

white’ dichotomy, where a resulting egalitarian land distribution is

‘white’ and an inegalitarian one is ‘black’. These other objectives also

need to be taken into account.

3. A Comparison of Agrarian Reforms in Germanyand Denmark

(i) The Danish experience

The consideration of the Danish land reforms makes an interesting

point of comparison for the later German reforms. Agrarian reform in

Denmark was essentially an eighteenth-century phenomenon, which

was largely completed by 1810. Most of the legislation was passed in

the period 1768–81 (although the abolition of serfdom came earlier in

1701). There are a number of points of interest which can be compared

and contrasted with the German experience.8

First, agrarian reform in Denmark was to a large extent a product of

‘enlightened’ state action. The state took the initiative and the reforms

were designed to serve state interests. These included the improvement

of state Wnances (which led the government to favour the peasantry who

were not exempt from taxation over the nobility who were). There was

also a conscious attempt to create a free independent peasantry; the

abolition of serfdom in 1701, for example, was connected with a deci-

sion to move to a conscript army. Neither the French revolution nor

industrialization had any part in the decision to introduce the reforms.

Secondly, despite this the reforms did not operate against aristocratic

interests in general. The nobility did give up their feudal rights but

8 The main sources used for this section are Baack (1977), Bjorn (1977), and Skrubbel-trang (1954); Tracey (1982) describes the evolution of Danish agriculture in the post-reform period.


they were well compensated. Many estates chose to sell land to their

tenants, and it was this which increased the proportion of freehold land

held by peasants from the pre-reform level of under 10% to 40% in

1810. Between 1786 and 1807 226 out of 800 estates sold a majority of

their land. However, the nobility did achieve an increase in rents on the

land that remained in their possession, and the price of land rose by

93% between 1781–5 and 1796–1800. As land-owners, both nobility

and peasantry gained from this increase.

The main impetus to the reform came from the realization that im-

proved agricultural techniques, coupled with the enclosure and consoli-

dation of holdings, could bring about a large increase in output.

Experiments by progressive landlords in the 1760s showed that these

measures, which included the abolition of labour service and the

granting of copyhold tenancies, could lead to a doubling of production

(as occurred on the pioneering BernstorV estate between 1764 and

1783). The ‘increase in the cake’ was large enough to give everyone a

share (grain production rose 40% 1770–1803). What the government

did was to make sure that a substantial slice of the beneWts went to the

peasantry. In other states, notably Britain, the gains from agricultural

improvement went largely to landlords and large-scale tenant farmers.

Thirdly, and this emphasizes the voluntary, co-operative nature of the

reforms, some aspects were not the result of legislation. Thus the aboli-

tion of labour service or Hoveri was not decreed. Yet by 1807 this only

applied to a quarter of Danish farmland, having previously been fairly

universal. The reason was that it had become an ineYcient system which

obliged the peasantry to keep large numbers of horses instead of cattle.

So the reforms made it possible to raise cattle numbers, from 280,000 in

1774, to 560,000 in 1804 and 834,000 in 1837. At the same time the

number of horses fell (from 400,000 in 1804 to 325,000 in 1837).

Equally, the process of enclosure and consolidation was not regarded

by the peasantry with hostility. Indeed, where they were freeholders or

copyholders they often initiated such schemes themselves, and in areas

where peasant freehold was common, enclosure was earlier.9 This

points to the importance of the prior evolution of tenant rights. Where

these had become hereditary and Wrmly entrenched, with legal support,

the peasantry could bargain successfully with land-owners to secure

enclosure and the commutation of feudal duties even if the state stood

aside.

9 Bjorn (1977).


A fourth point is the importance of Wnance. Here the state did estab-

lish a Royal Credit Bank which made loans to peasant farmers for the

purchase of freehold land or copyhold tenures. But the impact of this

bank was not large compared to other sources of Wnance. It lent 442,000

Rigsdalers for this purpose in 1786–98 (compared to 1,058,000 for land-

lord improvements). In the overall Wnancing of peasant land purchase

this was only a small share. In the 1790s it is thought that around 11⁄2million Rigsdalers were lent by banks for this purpose and another 1 mil-

lion by private investors. In 1797–1807 the Eukkase (or widow’s pension

fund) lent 2,245,000 million for the purchase of freeholds or copyholds.10

As far as the Wnancing of the cost of enclosure and other improve-

ments is concerned, this seems to have been mainly done by the land-

lords themselves. Tenants were required to repay the cost of enclosure

at an interest rate of 4%, which would have been a very reasonable rate

for agrarian credit in a pre-industrial society. However, landlords also

beneWted from the increase in rents, which may explain why they were

prepared to fund the costs of enclosure on favourable terms.

A Wfth point is that although the Danish reforms had an egalitarian

aspect, this did not mean that all beneWted equally. There were large

numbers of cottagers who did not gain freehold or copyhold land, either

through purchase or the enclosure of the common Welds. These bore the

burden of such labour service as remained. It is probably wrong to blame

the reforms for the creation of a ‘rural proletariat’ (which had probably

always existed in some form or another), but the fact that some groups

beneWted substantially while others did not was bound to increase social

divisions (at least between these groups). This was probably inevitable. No

eighteenth- or nineteenth-century government (and few since) had the

power to push through a completely egalitarian land redistribution. The

reforms had to take account of existing interests and work with these.

What the state could (and did) do was to alter the terms on which the

peasants bargained with the landlords. It forced the granting of longer

leases, regulated the conditions under which landlords could evict, and

obliged them to compensate tenants for improvements. The acquisition

of peasant holdings by large estates was prevented (as was the splitting

of holdings once consolidated).11 As one of the Danish ministers

involved, Count von Eggers, wrote in an open letter to Stein in 1807:

‘it is not enough . . . just to dissolve personal bondage . . . One must

secure them (the peasants) sustenance even as one gives them back their

10 Baack (1977), 37. 11 Tracey (1982).


freedom; one must ensure that they remain proprietors and do not

become day-labourers.’12

Finally, the special conditions of Danish agriculture in the late eight-

eenth and early nineteenth century should be mentioned. In the Wrst place,

even before the reforms Danish agriculture had not been dominated by

‘demesne’ farming (large estates typically using labour services provided in

lieu of rent rather than wage labour). Large, directly farmed estates had

occupied only 13% of the agricultural area before the reforms; this was

reduced to 9% in 1835. Secondly, the reforms took place at a time of rising

agricultural prosperity. Between the 1730s and the 1790s land prices rose

sevenfold. Between 1771–80 and 1791–1800 rye prices rose 26%, oats

prices rose 32%, and barley 45%. This made it easier to provide adequate

landlord compensation without overburdening the peasantry. Thirdly,

Denmark was well-situated for agricultural exports. By 1803, 12% of grain

production was exported. With good access to foreign markets, in particu-

lar the expanding British market, there were favourable conditions for the

establishment of a market-orientated peasantry. Danish geography was

also well suited to an era when water transport was the only practical

means of moving bulk cargoes such as grain. Finally, the debts that the

peasants incurred as a result of the reforms were wiped out in a burst of

inXation in 1807–10. This gave the reforms a redistributional aspect which

was certainly unintended.13

(ii) Agrarian reform in central and north-west Germany

Comparison with the Danish example shows that the German reforms,

in particular those introduced in Prussia in 1807–11, diVered in several

respects. One of the principal objectives of the centrepiece of the

German reforms, the edict of 1811, was the abolition of labour services,

an issue which was not addressed by legislation in Denmark. Peasant

emancipation in the German case was, as Clapham put it, ‘two-

edged’:14 the peasant was freed from restrictions and gained much in

the way of rights, but at the same time the previous policy of the

Prussian government—Bauernschutz, the deliberate protection of

the peasantry against landlord demands—was also given up.15

12 Baack (1977), 39.13 A similar burst of inXation contributed to the success of the land reform imposed on

Japan by the US occupation authorities in 1945.14 Clapham (1921, repr. 1966), 43.15 Knapp (1891), 35–6.


Labour services were not transformed into monetary obligations, but

rather redeemed by the cession of land to the landlord by the peasant (a

third of the total holding for tenants with hereditary rights, a half for

the others). Thus, the reforms had a built-in tendency to expand the

proportion of the total area farmed by large estates. Peasant land losses

to landlords were, however, balanced by gains as a result of enclosure.

The process still left many peasant holdings below the minimum

needed to maintain a self-suYcient rural enterprise (Spannfahige—

capable of maintaining a team of horses or oxen—is the term used in

the oYcial statistics).16

The enclosure movement in Germany started before the reforms and

continued long after. The result was that enclosure and emancipation

tended to be separate movements, with the consequence that the surge

in production which followed enclosure could not be used to provide

compensation for emancipation.

However, despite these diVerences, the fact is that in a large slice of

Germany, to the west of the Elbe and the north of the Main (and includ-

ing the Kingdom of Saxony), the reforms worked very much as they had

in Denmark: that is they produced a class of medium-sized peasant

holdings, mostly owner-occupied although the majority did rent some

land, free of labour service and other ‘feudal’ obligations, with the

common Welds largely enclosed and some consolidation of holdings.17

There were a number of reasons for this. In many areas there was, as

in Denmark, a deliberate policy bias in favour of the peasantry. The

Kingdom of Saxony is perhaps the best example of this.18 In the Rhine-

land and other western regions, French inXuence in 1806–13 may have

had some eVect, but was probably too brief to have made a major

impact.19 The availability of credit was an important factor. In Saxony

the Sachsische Landesrentenbank lent 86 million Thalers to peasants for

the purchase of labour service duties and other obligations in 1834–59.20

16 According to a Prussian survey of 1859, there were 344,701 Spannfahige holdingsand 604,501 Nichtspannfahige (holdings below the minimum size needed to maintain ateam of draught animals). The number of Spannfahige had fallen from 351,607 in 1816.However, the same survey also showed that in many areas the Spannfahige holdings hadmade net gains during the reform period; gains from the enclosure of common Weldsoutweighing losses through compensation for landlords, Meitzen (1868), 500.

17 In 1895 86% of agricultural land was owner-occupied, Rauchberg (1900), 580.18 Conze (1969).19 Dovring (1956), 144 emphasizes the impact of the French Revolution, but German

historians have tended not to support this point.20 Wehler (1995).


In general the western peasant farmer was more likely to get access to

credit from Landeskreditkassen and Rentenbanken than his counterpart

east of the Elbe, where the dominant credit institutions, the Landschaf-

ten, lent mainly to the large estates.21

The legal framework in the large parts of north-western and central

Germany that came under Prussian administration was similar to that

in the east, where the reforms led to the transfer of substantial amounts

of land to the landlords, due to the fact that peasants were obliged to

oVer land in return for the abolition of labour services. The fact that

this did not happen suggests that labour services were either already

rare, or naturally declining, or that the peasants were able to oVer

monetary compensation instead.22 Local conditions could play a part.

Thus in the Magdeburg region the prosperity brought about by the

cultivation of sugar beet created a class of medium-sized peasant

farmers who often had shares in the sugar reWneries. Labour services

and payments in kind were ended by 1845, which was relatively early

for central Germany, and the resulting debts seem to have been paid oV

quite easily.23

In general the circumstances under which the reforms were intro-

duced seem to have been at least as important as the actual legislation

(and possibly more so). What these areas had in common with Denmark

was a low level of demesne farming before the reforms, well-entrenched

peasant rights which were often hereditary, and favourable economic

opportunities thanks to the proximity of growing urban markets which

also provided employment opportunities for surplus rural labour. It also

should be remembered that the reforms of 1807–11 applied only to

Prussia and to Spannfahige peasants (who could provide a team of

draught animals); only in the 1850s were the reforms extended to peas-

ants who were Nichtspannfahige and to important non-Prussian territor-

ies (the main reforms in Saxony date from this period). This coincided

with the period of economic growth brought about by the railway

boom. Almost certainly this was a favourable conjuncture. It meant that

labourers who did not receive land as a result of the reforms could Wnd

employment outside agriculture.24

21 Tcherkinsky (1921), 78. There were other sources of Wnance, but the Landschaftenwere the main sources of mortgage lending, ibid. 82.

22 This is the view of Conze (1969).23 Plaul (1986).24 Hempel (1957) describes how the Heuerlingen (the hired labourers of north-western

Germany) became increasingly divided between agricultural and non-agricultural employ-ment (having previously combined agricultural work and cottage industry).


(iii) The reforms in Prussia east of the Elbe

This is the area which has received most attention from historians.

There is a widely accepted view that something went wrong with the

reforms in east-Elbian agriculture: ‘the beneWts of the agrarian reforms

went largely to the feudal aristocracy’, as one historian has written.25

The aristocracy gained substantially from the surrender of peasant land

as compensation for the loss of labour services, and also obtained 86%

of the land made available by the enclosure of the common Welds.

This view has been fairly well established since the writings of Theo-

dor von der Goltz and Adolph Meitzen in the nineteenth century. In

1868 Meitzen analysed the movement of the land distribution in the

original Prussian provinces over the 1816–59 period, and showed the

impact on peasant land-holdings of the loss of land given up to land-

lords. However, his analysis also showed that the peasantry had had

some compensation from gains made as a result of the overall increase

in the land under cultivation. Writing slightly later, in 1875, von der

Goltz examined the position of landless agricultural labourers, which

appeared to have worsened as a result of the reforms (although this

might also have been caused by demographic pressure).26

Subsequent writers, such as Max Weber and Gustav Knapp, en-

dorsed the views put forward by Meitzen and von der Goltz, and the

supposed failure of the land reforms in the Prussian east became a

generally accepted part of ‘received opinion’ on the broader question of

LandXucht: the depopulation of the rural east as a result of migration to

cities in central and western Germany.

There were Wve main reasons why the impact of the land reforms in

the east diVered from the eVect in the rest of Germany. The Wrst was

the remoteness of many eastern regions. In the period before the estab-

lishment of a railway network, marketing costs were extremely high for

any farm which did not have easy access to water transport. While the

east had its share of navigable rivers, such as the Oder and the Vistula,

there were large areas between these rivers which had poor communi-

cations, and little access to the expanding urban markets of western

Germany. In these circumstances, the establishment of a market-

orientated peasant agriculture faced almost insuperable obstacles.

25 Koselleck (1967), 498. The aristocratic estates made substantial gains from enclosure,adding over 100 hectares on average according to one calculation, Berthold et al. (1990),i. 234.

26 Meitzen (1868), von der Goltz (1875).


The poor state of communications east of the Elbe is demonstrated

by the Wgures in Table 2.1, taken from a survey in 1867. The Wgures are

expressed in relation to land area. An average square mile of territory

east of the Elbe had only 44% of the total length of improved roads in a

typical square mile in the provinces west of the Elbe. It had 46% of the

railway mileage, and 78% of the total mileage of canals and navigable

waterways, for an area of comparable size in the rest of Prussia.

The chances of having reasonable access to improved roads or a railway

were correspondingly low.27

Secondly, access to credit was limited. Only about a tenth of the

rural population was able to obtain credit, and interest rates were in the

range 12–30%,28 much higher than the rate of 4% charged to Danish

peasants, and the similar rate which aristocratic borrowers paid on loans

from the Landschaften.29

27 The importance of having good communications is well demonstrated in contempor-ary Argentina, where the lack of good roads in rural areas is an obstacle to the develop-ment of small farms. Only large estates can aVord the high cost of getting produce tomarkets.

28 Koselleck (1967), 507.29 Tcherkinsky (1921), 43. Lutge also stresses the role of credit, Lutge (1963), 236.

Table 2.1. Communications in Prussian provinces east and west of

the Elbe, 1867

Miles of improved

road (Chausee)

Miles of

railway line

Miles of canals and

navigable rivers

East and West Prussia 53.4 9.0 14.2

Pomerania 63.0 10.8 13.4

Posen 67.2 10.7 13.0

Brandenburg 62.8 18.1 28.4

Silesia 82.9 24.3 9.0

Average east of the Elbe 64.6 14.4 15.7

Prussian Saxony 105.9 26.4 20.7

Westphalia 159.6 30.1 14.0

Rhineland 178.9 37.0 24.2

Average rest of Prussia 148.0 31.3 20.1

Note: All figures are relative to the total land area (excluding water) in square miles (unitis pre-1870 Prussian mile).Source: Calculated from figures in Meitzen (1863), i. 65–6, iii. 291.


Thirdly, the main areas of industrial expansion were not in eastern

Germany but in the centre and the west. In the vicinity of these west-

ern cities, small-scale agricultural enterprises could prosper, supplying

perishable goods, such as fruit and vegetables, and also livestock pro-

duce. The economics of livestock production were much better suited

to peasant holdings than the extensive arable systems which character-

ized eastern Germany.30

Fourthly, the east was an area where the Prussian monarchy’s policy

of Bauernschutz—the protection and entrenchment of peasant rights—

had had less eVect. The position of the peasants before emancipation

was less favourable, and their bargaining position was consequently

weaker. This was an especially important factor in the parts of the east

which were not part of Prussia (such as Mecklenburg-Schwerin and

Mecklenburg-Strelitz) or where Prussian rule was more recently estab-

lished (such as Pomerania).31

Finally, the Gutsherrschaft administrative system, common in the

east, gave the aristocracy an important role in local government and

the administration of justice. It is probable that this had some eVect

on the agrarian reform process, either because the land-owners gained

control over the operation of the reforms, or because the government

was unwilling or unable to confront powerful local interests (a factor

which would have applied a fortiori in areas where a German aristocracy

ruled over a Polish peasantry).

In assessing the reforms, historians have tended to divide between

those who see this Wfth factor as the most important inXuence, and

those who recognize that the other factors meant that, even if the

government had been prepared to overrule aristocratic objections, the

conditions in eastern Germany in the early nineteenth century were not

favourable for small-scale, market-orientated, peasant farming.32 In

which case, a more radical reallocation of rural property rights would

have been damaging, holding back productivity gains, and producing a

subsistence sector unable to produce a marketable surplus.

For, considering the broader context of the contribution of the agrar-

ian reforms to German economic development, it is clear that there are

30 Von Thunen (1826) was the Wrst to analyse the eVect of urban proximity on agricul-tural systems; Levy (1911) explains the advantages that smallholdings specializing inlivestock produce had over larger units.

31 For a more detailed discussion of the importance of the pre-reform legacy see, interalia, Conze (1969), Lutge (1963), Harnish (1986), and Perkins (1986).

32 Schissler (1986) stresses the way the aristocracy controlled the reform process east ofthe Elbe.


strong reasons to regard the eastern reforms as successful, despite the

failure to match the egalitarian outcome in western Germany. Free

labour, paid in cash or in kind, was, on some estimates, three times

more productive than labour contributed under the pre-reform

system.33 There was a large increase in investment: the new demesne

farms required three times as much capital as the pre-reform hold-

ings.34 And uncultivated land was reduced from 40.3% of the total

agricultural area in 1816 to 19.0% in 1864. This, combined with

increased yields (there was a 43% increase in the yields of the major

crops 1805–63), meant that in the Wrst half of the nineteenth century

the Prussian arable sector was sustaining growth rates of 1.0–1.3% per

annum.35 In the context of an economy which, in this period, was

largely pre-industrial (there were pockets of industrialization in Saxony

and the Ruhr but no general upwards shift in the rate of growth before

the 1850s), it appears that arable farming has a good claim to be

regarded as a ‘leading sector’ in this period.36

The new, post-reform, ‘demesne-farming’ of east-Elbian agriculture

was, in many respects, a highly Xexible contractual system. The land-

lord provided land to the peasant in return for labour services which

were employed in the production of marketable produce on the de-

mesne farm. In the pure form, the peasant received recompense only in

the form of land. But in practice the system showed more variation.

The Inste of eastern agriculture received some land and a share in the

harvest. The Deputate or Kossaten received a smallholding plus Wxed

payment (in kind or later in money).37 This Xexibility allowed the

system to adjust for the requirements of the production system. Thus

peasants were often given responsibility for the care of animals and the

upkeep of buildings, becoming in eVect small contractors, which

33 Tipton (1976), 24. 34 Koselleck (1967), 507.35 Finck von Finckenstein (1960), 99–100 provides Wgures for yields and the cultivated

area. The implied growth rate of production is easily calculated from these. The range of1.0 to 1.3% per annum reXects the eVect of choosing diVerent base years.

36 HoVmann (1963), 100–4, produced much higher Wgures for the growth of agricul-tural production in Germany as a whole (1.9% per annum in 1800–1860/5) and derivedfrom these an estimate of the increase in agricultural productivity of 1.3% per annum overthis period. These Wgures were, however, based on estimates of livestock productionwhich were much less well founded than those provided for the Prussian arable sector byFinck von Finckenstein. HoVmann also assumed that non-Prussian areas matched thePrussian increase in arable production, which is unlikely, given that the Prussian increasewas made possible by the incorporation of previously uncultivated areas in the east, anunder-utilized resource not available to other German states.

37 Finck von Finckenstein (1960), 197 gives payments to a Deputat around 1800.


reduced monitoring costs for the landlord. Granting harvest shares gave

peasants an interest in the yield from the demesne Welds.38

The economics of the system shows that it was particularly well

suited to the settlement of areas where land was relatively abundant,

labour comparatively scarce, and markets either a long way away or

inherently risky. The peasant household had no need to market any of

its production beyond a small amount for local trade and barter.

Marketing costs were borne entirely by the landlord.

(iv) The extent of peasant land losses in Prussia

The extent of peasant land losses has remained one of the most serious

charges against the Prussian land reforms. Knapp claimed that the peas-

ants ‘lost’ a million hectares as a result of the reforms. This Wgure is

almost certainly an exaggeration. The situation was confused by the

point that ‘peasant’ holdings, which in the literature of the period are

generally deWned as holdings of at least 2 hectares, made both gains and

losses as a result of the reforms (losing land given up as compensation

for relief from labour services and gaining from the enclosure of the

previously uncultivated areas). The most careful examination of the net

gains and losses is probably that by Saalfeld, which is given in Table 2.2.

So, in total, peasant losses were not so large when it is considered that

labour services were removed from 16.95 million Morgen (the net loss of

5.5% of the aVected land is far less than the half to one third provided

for in the legislation), but this has to be set alongside a substantial

expansion of the total area under cultivation (the total arable area east of

the Elbe rose by 61% 1800–61).39 Thus, of the land made available by

the division of the commons only 14% went to the peasants; most of the

rest went to the large estates. It is estimated that there was an overall

increase in demesne land of 18%: 13.5% was due to the reforms and

4.5% due to purchases by the estates. Moreover, peasant losses were not

evenly distributed. In Westphalia peasant land-holdings rose by 9.9% in

area, and by 1.8% in Prussian Saxony, but falls of 13.2% and 12.8%

were recorded for Pomerania and Silesia (respectively).40

38 It has been suggested that labour service was highly ineYcient compared to ‘freelabour’ by a factor of 1 : 2 or 1 : 3 but this would have created a massive incentive to moveto wage labour; the long survival of demesne farming suggests that any ineYciencies werenot on this scale.

39 Dickler (1975), 273.40 All Wgures are from Saalfeld (1963).


The losses have to be set against the removal of labour services. Of

the spannfahigen Bauern 84% had had their service obligations removed

(a total of 6 million days) by 1848 and of the 23.5 million Handdienst-

tage (days of labour service without animals) owed in 1816, 17 million

had also been relieved. To get some perspective on this, it implies

(assuming that the Wgure of 17 million Morgen for the aVected area is

correct) that a holding of 20 hectare (or 80 Morgen) might have owed

on average 80 days labour service without draught animals, and 28 with

draught animals: a total of 108 days.41 It would have been a substantial

gain to be free of this.

The process by which the larger estates gained at the expense of

peasant holdings was a complex one, in which free purchase played an

important role. There was a high rate of turnover in the land market in

the early part of the century, both for estates and peasant holdings.

Thus 24% of Prussian peasant land changed hands in 1816–59, but

most of these transfers (which totalled 8.2 million Morgen) were from

one peasant to another.42 Transfers to smaller or larger holdings were

only 21% of the total. Equally, the larger estates changed ownership

41 In practice it would have been more likely to be one or the other; so perhaps 50–60days with animals and no Handsdienst, or 120–50 days’ Handsdienst on its own; calculationsuse Wgures from Koselleck (1967), 498; note, the estimates are much lower than the 5–6days per week quoted by Harnisch (1986), but these levels could have been reached insome regions.

42 Saalfeld (1963).

Table 2.2. Peasant gains and losses, Prussia 1816–59, in million

Morgen

in Morgen (m.) as % of all peasant

land 1816

Losses: to smallholdings 1.29 3.8

to large estates 0.47 1.4

total 1.76 5.2

Gains: through enclosure 0.83 2.5

Net losses 0.93 2.7

Note: ‘Peasant’ holdings are defined as between 2 and 100 hectares; a ‘Morgen’ is,according to the dictionary (Langenscheidt) ‘a variable unit of between 0.6 and 0.9acres’ , but most commentators seem to assume that 4 Morgen ¼ 1 hectare.Source: Saalfeld (1963).


quite frequently. Out of 888 estates in East Prussia in 1806 only 378

were in the same family in 1829.43

There were two main reasons for this. The Wrst was the cost of the

reforms and associated investments which led to build-up of debt. The

second was the fall in agricultural prices in the 1820s.

The reforms led to a considerable expansion of the area under culti-

vation, but this was expensive. New houses had to be built for the

labour needed to replace that previously provided by the tenants. The

cost of enclosing (and where necessary draining) the common Welds had

to be met. The peasants had to pay compensation where this had been

agreed, meet their share of the enclosure costs, and replace livestock

which were the landlord’s property.

Rather more is known about the debts owed by the large estates

which tended to be recorded (and periodically surveyed by the Prussian

government) but the position of the peasantry was also serious.44

Whereas the estates could borrow from the Landschaften at 4% the

peasantry had to pay considerably more, if they were oVered credit at

all. Certainly the nobility had problems: a third of estates in Pomerania

went into receivership during the period 1817–29.45 But the fact that

they, and those investors who subsequently bought up the bankrupt

estates, could borrow may have contributed to the gains made by the

large estates during this period.

If this is right then it was the economic success of the reforms which

contributed to their ‘social’ failure. For it was the increase in the area

under cultivation, and the consequent reduction in the uncultivated

area from 40% of the Prussian land area in 1816 to 19% in 1864, that

was the reform’s greatest achievement: ‘this increase of cultivated area

and total production was only made possible by the new energies re-

leased by the reform process’.46

(v) The regional pattern of land losses and transfers in Prussia

The impact of the reforms varied considerably in the diVerent Prussian

regions. In a sense this was an intriguing ‘natural experiment’: a similar

set of legal changes had diVerent eVects according to circumstances.

43 Koselleck (1967), 512.44 Meitzen (1885) is a rare attempt to survey debt on peasant holdings, which showed

that debts could be as high relative to property values as on larger estates.45 Koselleck (1967), 508.46 Lutge (1963), 228.


Government surveys showed that previously existing diVerences in

agricultural structure, and in systems of agricultural tenure, aVected the

way the reforms operated in the diVerent regions. Table 2.3 gives some

Wgures for the most important reforms. The Wrst column gives the total

number of holdings which were completely released from all services.

The largest numbers of these were in Silesia and Prussian Saxony. The

next two columns give Wgures for the two of the most important cat-

egories of labour service: Spanndienst (work on the demesne farm with a

team of draught animals provided by the peasant household) and Hand-

dienst (work on the demesne farm without the obligation to provide

draught animals). The province where the amount of relief from Spann-

dienst was highest was Posen; the province with the highest amount of

relief from Handdienst was Silesia.

The Wnal column gives the area of the holdings whose land was

‘reformed’ (neu reguliert) through the enclosure of common Welds or theconsolidation of strips. These areas were largest in the four eastern

provinces: East and West Prussia, Posen, Pomerania, and Brandenburg.

The other provinces had little land in this category.

In general, the reforms are shown to have had little eVect in the

western provinces of the Rhineland and Westphalia. In these areas

feudal obligations were already in decline, as a result of developments

in the pre-reform era (including the period of Napoleonic reform).

Enclosure and the consolidation of holdings either took place in the

Table 2.3. The impact of the agrarian reforms in Prussia, 1816–48

Number of

holdings

released

from services

Given up

days of

Spanndienst

(000s)

Given up

days of

Handdienst

(000s)

Area of

reformed

holdings

(000 Morgen)

East and West Prussia 8490 375.1 568.2 1125.7

Pomerania 13,015 818.0 1465.1 1208.3

Posen 15,002 1985.9 4344.4 1388.0

Brandenburg 39,830 1145.1 2632.2 1231.3

Silesia 95,014 195.1 7547.5 205.3

Prussian Saxony 93,185 33.4 252.8 0.2

Westphalia 18,326 0.0 59.5 0.0

Rhineland 6789 0.0 0.0 0.0

Source: Figures from Meitzen (1868), iv. 302–3.


pre-reform era or else were carried out without direct involvement by

the state or its agencies, which would exclude the reformed holdings

from these statistics. Another province where there was little eVect

from the reforms was Prussia (the province of Prussia was later divided

into East and West Prussia), apart from the area recorded in the fourth

column.

Turning to the other side of the reform process, the provision of

compensation to landlords for the loss of labour services and other

feudal duties, this shows a similar variation between diVerent regions

(Table 2.4). Large lump sum compensation payments were made in

three regions: Brandenburg, Saxony, and Silesia. In Posen, compensa-

tion was more likely to take the form of an annual monetary payment.

The heaviest payments in kind were made in Brandenburg.

The Wnal column gives the amount of land given up as compensation,

and this reXects both the extent of the reforms, and the amount of

compensation paid in other ways. Thus, the low level of land losses in

Saxony and Silesia can be connected to the relatively high level of

capital sum payments in these provinces. In Posen, land losses were not

so high, but there were substantial annual payments. The provinces

Table 2.4. The costs of agrarian reforms in Prussia, 1816–48:

compensation payments by peasants to landlords

Total compensation paid, in the form of:

Capital sums

(000 Thaler)

Annual

payments

in money

(0000

Thaler)

Annual

payments

in kind

(Scheffel

of rye)

Land

given up

(000

Morgen)

East and West Prussia 423.9 153.2 15,300 196.9

Pomerania 1403.4 140.2 36,900 590.6

Posen 104.7 492.6 10,400 206.9

Brandenburg 4172.2 266.4 108,500 405.9

Silesia 4192.5 338.9 51,900 118.4

Prussian Saxony 5932.1 134.0 23,700 14.0

Westphalia 1965.6 68.7 2100 0.3

Rhineland 350.3 5.9 600 0.1

Source: As for Table 2.3.


with the largest losses were Pomerania and Brandenburg. In the former,

payments under the other categories were relatively low, and so it seems

that giving up land was the preferred method of compensation in this

province. In Brandenburg, the overall level of compensation appears to

have been relatively high, with high payments in all categories.

What this comparison shows is the variation in the reform process: a

typical reform in Posen might involve relief from an obligation to pro-

vide Spanndienst, labour service with draught animals, and the compen-

sation would be in the form of increased monetary payments; in Silesia,

it was more likely to be relief from Handdienst, with compensation in

the form of a lump sum payment. Concentrating on the extent of land

losses and gains by peasant households obscures these other dimensions

of the reform process.

Although the reforms did involve some loss of land by peasant

farmers, this has to be set against the gains that were made. The loss of

grazing land may have been outweighed by the ability to devote more

time to the holding (due to relief from labour services). The new hold-

ing was likely to be enclosed, which made it easier to introduce

improved husbandry techniques. Cattle could be kept instead of horses

or oxen to work the landlord’s Welds. For the landless labourers, how-

ever, the loss of common grazing was probably not outweighed by other

gains. The main social cost of the reforms was the rise in the number of

landless labourers, and the increase in the numbers of peasant holdings

which were not large enough to be self-suYcient.47

47 Berthold (1978).

Table 2.5. Consolidation of holdings in Prussian Saxony

Altengotten Großengotten

Number of farms 848 1241

Total area 6101 Morgen 6803 Morgen

Number of plots before consolidation 18920 16100

Average size 0.32 Morgen 0.42 Morgen

Number of plots after consolidation 913 1584

Average size 6.68 Morgen 4.29 Morgen

Source: Conrad (1891), 666.


(vi) Consolidation and the reforms in southern Germany

In southern Germany the reforms started late and in one important

respect were unsuccessful: there was relatively little consolidation of

holdings. Fragmentation of holdings had been a fairly general conse-

quence of the open Welds system practised over much of Germany in

the eighteenth century. It was greatly increased by the system of part-

ible inheritance or Realteilung practised in the south (and other areas).

In those areas of central Germany where the reforms were fully imple-

mented there was considerable consolidation. Thus in two districts of

Prussian Saxony, Altengotten and großengotten, the average numbers

of separate plots were reduced as follows:

In the article from which these Wgures are drawn Conrad gives seven

examples of districts where consolidation had not taken place by the time

the article was written (it was published in 1891). In these the number of

separate plots per holding ranges from 21 to 59. Most are in the south,

though some are in central Germany (the Koblenz area, for example).

In general consolidation made slow progress in the Wrst half of the

century, and eVective legislation was not passed until the 1850s in

Prussia, Baden, and Hesse, and the 1860s in Bavaria and Wurttemberg

(further laws were passed in these states in 1886).48 Despite this there

had already been some consolidation in the north-west. Even after laws

had been passed there was much less consolidation in the south. In the

south-west in particular, where peasant rights were strongly protected,

there was opposition to the division of the common Welds.

The system which evolved in the south was characterized by a rela-

tively high population density, quite high reliance on non-agricultural

earnings to supplement the income from the holding, few large estates

but many smallholdings, and continuous division of holdings (not

necessarily on inheritance—children could be given a share on reaching

their majority). It was probably not very eYcient in economic terms.

Germany came to import livestock products from Denmark and Hol-

land which, arguably, could have been produced on southern farms.

But it provided a reserve of labour which was available for industrial

use when needed.49 And in downturns the rural economy could re-

absorb surplus labour, at least for a time.

Whether this system produced more or less discontent is hard to say.

Contemporary observers thought it gave the peasants a higher social

48 Haushofer (1957), 58. 49 Catt (1986).


status.50 But, on the other hand, it was a vulnerable system, with a

tendency to overpopulation, and not one which was conducive to agri-

cultural progress. The dependence on non-agricultural earnings worked

well when there were local sources of employment, but less well when

craft earnings came under pressure due to competition from large-scale

industry. In practice only relatively few of those with smallholdings

found industrial employment: 27.5% of those with possession of hold-

ings of less than 2 hectares in Baden in 1907, and 23.4% in Wurttem-

berg, against 55.8% in Westphalia and 40.6% in Anhalt.51

(vii) Concluding comments

The comparison with Denmark, and the consideration of the reform

process in the diVerent parts of Germany, makes it plain that local

circumstances and the timing of the reforms could have important

eVects on the outcome. Past policies also had an eVect. The Wrm estab-

lishment of peasant rights before the reforms certainly aVected the

subsequent distribution of land. Thus in Mecklenburg and the former

Swedish Pomerania there had been little protection of peasant rights

and these areas became dominated by large estates: 75% of agricultural

land in the Stralsund region was in holdings of over 100 hectares in

1882. In Prussia east of the Elbe the proportion was 43%.52 So the

Prussian policy of peasant protection did have some eVect. But in view

of the cost of the reforms, the paucity of rural credit institutions, and

the fact that labour was becoming less scarce, it was probably inevitable

that the large estates were going to make gains, especially in the remote

areas east of the Elbe.

4. German Agriculture in Comparison

with Other European Countries

As already noted, the implicit assumption behind the ‘failed revolution’

hypothesis is that there was the possibility of a genuinely revolutionary

outcome, which would have resulted in a more complete modernization

of nineteenth-century German politics and society. For political

50 For example, Knapp (1887), 309. 51 Dillwitz (1975), 110–11.52 Perkin (1986); Dillwitz (1973); Conze (1969) refers to the annihilation of the peas-

antry in Mecklenburg and Pomerania in the eighteenth century.


developments the implied point of comparison is usually Britain, whose

path of gradual progress to a widened franchise and a fully democratic

system of constitutional government contrasts with Germany’s halting

and uncertain moves in the same direction. For agriculture, the most

relevant comparison is the land settlement brought about by the French

revolution.

But it should not be assumed from this that all was well with rural

France in the nineteenth century. While the revolutionary settlement had

created a large number of small and medium-sized holdings, the level of

owner-occupation was lower than in Germany. In 1882, only 60% of

French agricultural land was in owner-occupation, 27% was rented, and

13% was in metayage, or share-cropping. By contrast, in Germany in

1895, over 86% of land was owner-occupied, just 12% was rented, and

only 1% was in share-cropping (this adds in Deputatland, which was a

form of share farming although applied to animal husbandry).53

A second problem was the failure of yields and productivity in

France to rise as much as in Germany. According to Bairoch, around

1800 French wheat yields were 85% of the German level; by 1910 they

were only 71% of the German Wgure. His estimates of agricultural

productivity show a rise for Germany of 371% 1800–1910, for France

the increase is 172%.54

There are various measures which can be used to compare the struc-

ture of German agriculture after the reforms with other countries. One

is the division of the agricultural labour force between employed and

self-employed workers. Labourers who were either landless, or who had

insuYcient land to provide full-time employment on their own holding,

would be obliged to take work on other farms or outside agriculture.

These would then be classiWed as employed, in contrast to those whose

own holding was large enough to support themselves and their family

without working oV the holding. The proportion of the agricultural

labour force classiWed as employed is thus an approximate measure of

53 German Wgures from Rauchberg (1900), 579; French from Auge-Laribe (1950), 93.The German Wgure was just below that of South Korea in 1980, 92% owner-occupied,which had increased from 35% in 1950, Wgures from Asian Productivity Organisation(1994), 37. The South Korean example is generally considered the best-managed landreform in recent years. By contrast, despite various eVorts at land reform, Pakistan stillhas 36% of land in tenancy and 18% in share-cropping, ibid. 70.

54 Figures from Bairoch (1989); Bairoch’s measure of agricultural productivity is cal-ories produced per male agricultural worker, which is not ideal but probably the onlymeasure which can be constructed for this period given the deWciencies of the datasources.


the proportion that had access to holdings of a viable size. Table 2.6

provides a comparison drawn from national censuses carried out around

1900. Limitations in the data sources mean that the analysis has to be

limited to the breakdown of the male labour force.

This table shows that the German reforms produced an agrarian

sector which predominantly consisted of self-employed owner-

occupiers: the classic land reform objective. The percentage of the agri-

cultural labour force which was classiWed as being employed was low by

comparison with other European countries. The French level was

notably higher. The high level of employment in England and Wales is

not so surprising, given a historical process which had resulted in an

unusually inegalitarian allocation of rural property rights. However, a

number of other countries, not usually considered to be particularly

unequal, had lower levels of self-employment than Germany. Quite

evidently, the Danish land reforms did not produce a position in which

all had equal access to land.55

55 The extent of ‘landlessness’ is more diYcult to compare. Many employed workers inGermany, and elsewhere in Europe, had the use of a smallholding to grow potatoes andother vegetables. The Wgures given in Table 2.5 are not comparable with modern surveysof landlessness: see Jazairy et al. (1992), 406–7 for a summary of these.

Table 2.6. Percentages of the male agricultural labour force classified

as being employed (including servants), excluding family members,

various censuses circa 1900

Date of census % classified as employed

Austria-Hungary 1900 28.3

Norway 1900 29.3

Finland 1910 30.7

Germany 1907 36.7

Belgium 1901 36.8

Italy 1910 46.8

France 1901 47.2

Sweden 1900 47.6

Switzerland 1900 51.5

Denmark 1900 60.1

Netherlands 1899 67.6

England and Wales 1901 73.0

Source: Figures from Flora et al. (1983), adjusted to exclude family members andunclassified workers.


Comparable Wgures on the distribution of agricultural land by size of

holding are only available for a few countries. Table 2.7 provides some

Wgures for England and Wales, Germany, and France. The Wgures have

been adjusted to allow for diVerences in the classiWcations by size class.

The British and German classiWcations use broadly comparable meth-

odology: German holdings were classiWed by total agricultural area, the

British by area under crops and grass. The French agricultural censuses

classiWed holdings by total area, which tended to increase the import-

ance of larger holdings, which would often include a disproportionate

amount of woodland, moor, or other uncultivated areas.

Comparison of the English and German Wgures shows that the distri-

bution of land by size of holding was much more unequal in England:

there was a smaller smallholding sector and the area in large holdings

was proportionately greater. Typically, English agricultural land was

concentrated in medium- or large-scale tenanted holdings of 100 acres

or more. Comparison with France is more diYcult, due to the diVerent

classiWcation methods used. An English statistician who examined this

issue in 1887, P. G. Craigie, thought that, whereas three-quarters of the

English agricultural area was in holdings of over 100 acres (roughly

Table 2.7. A comparison of the percentage distribution of land by

size of holding

England and

Wales 1885

Germany 1882 France 1882

Area in holdings of less

than 20 hectares

14.4 45.0 38.2

Area in holdings of

between 20 and 40 hectares

13.0 13.1 12.9

Area in holdings of

over 40 hectares

72.6 41.9 48.9

Area in holdings of

between 40 and 100 hectares

34.0 17.5 n.a.

Area in holdings of

over 100 hectares

38.6 24.4 n.a.

Note: England and Wales: classified by area under crops and grass; Germany: classifiedby total agricultural area; France: classified by total area.Source: British figures from Ministry of Agriculture, Fisheries and Food (1968), 20;French and German figures from Tracey (1982). To produce comparable figures, thedensity functions were interpolated using a log-log interpolation method.


equivalent to 40 hectares), only a third of the equivalent French area

was in this class.56 This is a considerably lower proportion than the

48.9% share given in Table 2.7. If this is right, then the distribution of

agricultural area in France may have been rather more egalitarian than

in Germany, contrary to the impression given by the French distribu-

tion by total land area, shown in Table 2.2. Either way, the diVerence

was probably not that great.

However, the objective of agrarian reform is not just to produce a

relatively equal distribution of land-holding: it is also to produce a more

egalitarian pattern of land ownership. Comparison of the distribution of

land according to the size of the owned farm or estate is even more

diYcult, as there are no oYcial Wgures available for this period.

In England in the early 1870s there was a Local Government survey

of land ownership, sometimes referred to as the ‘New Doomsday

Book’.57 This survey showed that land ownership in England and

Wales was highly concentrated: 74% of privately owned land was in

estates of 300 acres or more. This is a distribution by size of total land

ownership, not ownership of agricultural land. Making some allowance

for this suggests that between 60% and 65% of agricultural land was in

estates of 300 acres or more (the lower Wgure assumes that all non-

agricultural land was in larger estates), and between 65% and 70% in

estates of 100 hectares or more. This contrasts with the farm census

Wgures which show that 28% of agricultural land was in holdings of 300acres or more in 1875 (or 38% in holdings of over 100 hectares). Thus,

the distribution of land by ownership was much more inegalitarian

than the distribution by holding.

The high level of owner-occupation in Germany meant that the

gap between the distribution by ownership and by holding was nothing

like as large as in Britain. The distribution by ownership may well

have been more egalitarian than the distribution by holding size. This

could arise if larger holdings rented from more than one landlord, or

if they included some rented land as well as some which was owner-

occupied.

The best source of information on the structure of ownership in

German agriculture is a survey carried out by the Reich statistical oYce

in 1938. This produced a careful cross-tabulation of the distribution of

56 Craigie (1887), 110; Levy (1911) is also a useful source on the structure of Englishagriculture in comparison with other countries.

57 The original returns were summarized in Bateman (1879), but this study makes useof the adjusted results published in Brodrick (1881).


land by size of estate and by size of holding (something which is

not available for British agriculture). The results are summarized in

Table 2.8.

This table shows a fairly close correspondence between the size dis-

tribution by holding size and the pattern of ownership. There are some

regions where the ownership distribution was signiWcantly more equal

than the holding distribution: Brunswick, Silesia, Westphalia, and

Hanover are the most notable. There are two regions where the owner-

ship distribution was signiWcantly less egalitarian: Mecklenburg and

Anhalt. But in general the two series do not diVer by much.

Table 2.8. The distribution of land by size of holding and by size of

estate, by province, 1938

% of land in

individual

ownership

% of agricultural

land in holdings of

over 100 hectares

% of all privately

owned land in estates

of over 100 hectares

East Prussia 72.0 32.2 32.0

Brandenburg 71.1 35.7 31.1

Pomerania 71.8 46.5 40.0

Silesia 71.7 38.1 26.7

Prussian Saxony 72.1 25.9 25.7

Schleswig-Holstein 83.8 16.8 14.4

Hanover 77.1 19.7 10.1

Westphalia 81.1 19.1 5.7

Hesse-Nassau 48.2 6.4 6.7

Rhineland 59.9 10.5 4.7

Bavaria 72.1 5.7 4.5

Saxony 72.4 17.4 13.8

Wurttemberg 63.9 1.1 3.9

Baden 61.5 10.9 5.9

Thuringia 62.3 9.8 11.3

Hesse 62.7 9.9 5.1

Mecklenburg 82.8 35.0 48.2

Oldenburg 84.4 11.8 19.3

Brunswick 56.2 13.2 2.9

Anhalt 55.5 26.1 35.2

All Germany 70.0 22.3 18.7

Source: SDR (1938).


The drawing of conclusions for the pre-1914 era from a survey

undertaken in 1938 is obviously something which needs to be done

with some care. The crisis which aVected German agriculture in the

1920s would have altered the relationship between the two series to

some extent. One consequence could have been an increase in the own-

ership of land by institutions, both public sector and private (as a

result of the repossession of mortgaged estates, for example). The Wrstcolumn of Table 2.8 shows that, in some regions, the percentage of land

in private ownership had fallen to under 60% (though this does

not appear to have much eVect on the relationship between the two

distributions).

With these caveats in mind, a hedged statement is advisable: on the

evidence currently available, there is no reason to anticipate that the

ownership distribution in the German regions before 1914 would have

been markedly less egalitarian than the holding distribution, with the

exception of regions such as Mecklenburg where the 1938 survey

showed this to be the case. In other words the holding distribution is a

good guide to inequality in land ownership in pre-1914 Germany

(which follows from the high level of owner-occupation in German

agriculture).

Looking at the map of German agriculture in 1895 (Figure 2.2), the

only areas with a level of inequality in land-holding which came up to

the English level (38.6% of land in holdings of over 100 hectares) were

those east of the Elbe. Even in this area there were several regions

where larger holdings were relatively rare: in the East Prussian region

of Gumbinnen holdings with over 100 hectares accounted for 33.1% of

the total agricultural area, in the Silesian Regierungsbezirke of Oppeln

and Liegnitz they occupied 31.2 and 28.4% respectively.

But when allowance is made for the substantial gap in Britain be-

tween inequality in land ownership and in land-holding, the diVerence

between the two countries becomes a massive one. The English level of

65–70% of land in the ownership of estates with at least 100 hectares

was way above the most probable level in most regions east of the Elbe.

It was equalled only in parts of Pomerania (the Stralsund Regierungsbe-

zirk and possibly Koslin as well) and, most probably, in Mecklenburg.

In Posen and Bromberg, the English level could have been reached if

the ownership distribution had been substantially less equal than the

holding distribution.

The image of east-Elbian agriculture as one dominated by large

estates, on the English pattern, is to a large degree a false one. The


typical farm in Brandenburg, Silesia, East Prussia, and the Danzig

region of West Prussia was more likely to be an owner-occupied hold-

ing of around 30–50 hectares. Even where larger estates predominated,

they were very diVerent from the English model: an average Junker

estate might consist of around 250 hectares farmed ‘in hand’; the

equivalent English aristocratic estate in the 1890s would be almost en-

tirely let out to tenants, and considerably larger.

In Germany 14% of land was in some form of tenancy in 1895; the

Wgure for England and Wales in 1891 was 85%.58 Tenanted percentages

in German regions came nowhere near the English Wgures. Figure 2.3

gives the percentage of the total agricultural area which was rented, by

region. The map shows that the highest percentages were in Alsace-

Lorraine and in Mecklenburg. The lowest was in Bavaria. In general,

the rented percentage was not particularly high east of the Elbe, apart

58 Rauchberg (1900), 579; Ministry of Agriculture, Fisheries and Food (1968), 25.

<5%

5−10%

10−15%

15−20%

20−25%

25−30%

30−35%

35−40%

40−45%

45−50%

50−55%

55−60%

60−65%

>65%

Fig. 2.2. Percentage of land in holdings of over 100 hectares, 1895


from Mecklenburg. Levels in the other eastern provinces were below

those in central and western regions such as Hanover, Prussian Saxony,

and the Rhineland.

In most of Germany, especially in the west and the south, large

estates were a rarity. The typical farm was small, 10–20 hectares, and

owner-occupied. There was little employed labour. The rural sector

was therefore, by the standards of contemporary European countries, a

relatively egalitarian one. The low proportion of landless labourers in

the rural population as a whole, and the high level of owner-occupation,

mean that the structure of nineteenth-century German agriculture

compared well with the situation in many less-developed economies

today.59

5. The Kuznets Curve and German Agriculture

The Kuznets Curve is a hypothesis about the development of inequality

in industrializing economies based on a series of assumptions about

inequality in diVerent sectors. However, the analysis of particular coun-

tries shows that historical circumstances and geographical conditions

matter a lot and can lead to large diVerences: Do these favour large

estates or plantations? Has there been an attempt at land reform? Was

this successful or unsuccessful?

In the German case, the agrarian reforms produced a relatively egali-

tarian rural sector, which meant that the starting point for the Kuznets

Curve was a low one.60 If inequality in the industrial sector turned out

to be high, then a shift into this sector from the more egalitarian rural

sector would tend to push up overall inequality. The structure of

German agriculture also meant that the experience of rural–urban

migration produced a much more dramatic change in social circum-

stances than in Britain. In Britain the tripartite division of industrial

society—owner or shareholder at the top, salaried manager or foreman

in the middle, wage-earning worker at the bottom—was not dissimilar

to the tripartite division of rural society—large land-owner who rented

59 Prosterman and Riedinger (1987), 41 gives the current position in developing econ-omies.

60 The Prussian rural distribution was an egalitarian one by modern standards. Thisissue is further discussed in Chapter 9, which includes estimates of income inequalityderived from tax statistics.


out land at the top, tenant farmer in the middle, landless rural labourer

at the bottom.

The position in rural Germany was a very diVerent one. Migrants

from the rural areas in the south and the west came from areas where

peasant proprietorship was the dominant system. If they were the sons

and daughters of peasant farmers they would have had little experience

of wage labour. If they came from the families of the relatively small

number of landless labourers, the social context would still have been a

dissimilar one (working alongside neighbours who did own some land)

to that provided by an urban factory. Even in the east, the position of

the agricultural labourer was not that of a wage-earning labourer paid

by the hour. There were a variety of contractual arrangements, most of

which gave a degree of autonomy to the labourer and his family, at least

as far as the organization of the day’s work was concerned.61

61 The system was one which required the worker to supply speciWed hours of labourat harvest time and other peak periods, but which left the organization of duties such ascow-keeping up to the labourer. Outside peak periods, the worker and his family were

>5%

5−10%

10−15%

20−25%

15−20%

25−30%

30−35%

Fig. 2.3. Rented land as a percentage of the total agricultural area, 1882


For these reasons, the structure of German agriculture had an eVect

on the impact of industrialization and migration: the relative change

was a greater one. Industrialization brought about a transformation of

society which was more rapid than in early-industrializing economies

such as Britain. It was also a more profound change than in Britain.

expected to occupy themselves on a smallholding provided by the land-owner. See Weber(1893) and Chapter 6 below for further details.


3

The Pattern of Migration,

1870–1913

1. Introduction

The discussion in the Wrst two chapters has made it clear that there are

particular reasons to be concerned with the process of migration during

the early stages of industrialization. First, migration is the conduit

whereby a rural labour surplus holds down wages in urban labour

markets during the labour surplus phase of the Lewis Model. Secondly,

the transfer of population from a relatively egalitarian rural sector to a

more unequal urban sector is an integral part of the Kuznets process,

pushing up overall inequality. Both these factors make it important to

consider the factors which aVected the pace of migration, and which

eventually led to a change in the pattern of migration as the labour

surplus phase came to an end.1

The important feature of the labour surplus phase is the movement

of population from predominantly agricultural regions to the expanding

urban centres, resulting in a high level of net losses in the former areas,

and large gains in the latter. But this feature was not the only type of

migration in industrializing societies. It was superimposed on a pattern

which already existed in pre-industrial societies: of seasonal migration

connected with the harvest demands and other types of work inXuencedby weather conditions. It also merged into a more modern pattern,

characteristic of mature industrial societies, of migration between urban

centres, responding to shifts in labour demand by industry. But neither

of these other streams has the implications predicted by the Lewis

Model.

Other migration streams which have diVerent implications are those

which cross international boundaries. Where there is a large outward

1 Internal migration is not the only source of downward pressure on labour markets.Kindleberger (1967), 20 argues that in late nineteenth-century America there were laboursurplus conditions created by heavy immigration. In post-1945 West Germany there was asimilar eVect as a result of the inXux of refugees, ibid. 34.

Xow of emigrants this will transfer a portion of the rural labour surplus

to foreign labour markets, thus acting as a ‘safety valve’, relieving pres-

sure on over-supplied urban markets in the country of origin. But

where there is large-scale immigration, then surplus conditions else-

where can intensify the ‘labour surplus’ phase in the host country.

EVectively, the labour market in the receiving country is integrated

with that of the country of origin.

Those parts of the migration process which have tended to attract

most attention are moves which result in a permanent gain or loss of

population. In the aggregate this is net migration: a surplus or deWcit is

created when moves in are deducted from moves out. But in most soci-

eties net internal migration is relatively small compared to gross internal

migration. There are many moves which are temporary or reversible.

There are life-cycle moves: young men (less often women) who move to

cities intending to return, perhaps hoping to have accumulated enough

savings to buy a smallholding or some other business in their region of

birth. There are seasonal moves: farmers who take winter jobs in cities

but return for the harvest; other agricultural workers who move with the

harvest, starting in the warmer southern regions, and moving north and

east as the harvest season progresses. There are certain skilled occupa-

tions which have always been associated with high mobility: stone-

masons and other skilled building workers moved with major

construction projects in medieval Europe, such as bridges or cathedrals.

Railway construction had a similar eVect in the nineteenth century,

attracting large numbers of workers, skilled and unskilled.2

Most of the available statistical sources tend to measure net migration

better than gross migration. Census returns generally record place of

residence at the time of the census and place of birth: it is rare to Wnd

any evidence on places of residence between these two points. This

means that many temporary moves between these dates are unrecorded.

Germany is relatively unusual in that, thanks to the system of

MeldepXicht, compulsory registration after a change of residence, there

is some evidence on gross migration.3 This makes it possible to examine

the relationship between the labour surplus transfers and the general

pattern of mobility.

2 Von der Goltz (1875) includes the results of a survey of rural migration in the early1870s, which shows both the traditional pattern of seasonal migration, and the impact ofrailway construction.

3 The MeldepXicht system and its value as a migration source is described in Meyer(1938).

The Pattern of Migration, 1870–1913 57

2. German M igration and Urbanization

in International Comparison

Most of this study is concerned with the eVect of internal migration in

Germany. But it is worth considering whether the German experience

diVered from that of other European countries, and if so, in what way?

It is not easy to achieve comparability of migration statistics between

countries. The drawing of the regional boundaries aVects migration

records. A country with a large number of regional units will tend to

have more recorded migration than one with a relatively small number

of larger units. An accurate comparison needs to be based on a similar

set of regional units for all countries.

In the late nineteenth century, the American statistician, Adna

Weber, attempted such a comparison. His results are given in Table 3.1.

These show that, compared to all other European countries, the German

states of Prussia and Saxony had the highest rates of internal migration.

In many other countries there were relatively few internal moves. This is

particularly striking for the Scandinavian countries, which had high

rates of external migration, but much less internal migration.4

4 Norway and Sweden had the highest rates of emigration after Ireland, Hatton andWilliamson (1998), 33.

Table 3.1. Percentage of the population born outside

the country where enumerated

Date of census % born elsewhere

Saxony 1885 30.9

Prussia 1890 30.3

England and Wales 1891 28.4

Denmark 1890 22.1

Austria 1890 19.8

France 1891 18.7

Switzerland 1888 17.9

Netherlands 1889 13.0

Norway 1875 12.8

Sweden 1880 11.5

Hungary 1890 10.8

Source: Weber (1899), 249.

58 The Pattern of Migration, 1870–1913

Another approach is to compare rates of urbanization. If cities are

growing fast, so that the proportion of the population living in cities is

rising, then this can come about either through higher rates of natural

increase or through migration. In practice, in most European countries,

the Wrst factor was not important: the rate of natural increase in cities

was generally below that in rural areas. So, rising urbanization was a

consequence of migration.

Table 3.2 presents some Wgures for the proportion of the population

living in cities in various European countries. These are countries for

which results are available both for the period around 1870 and again

for the period just before the First World War. The results for the

Netherlands in 1909 are given in brackets as there are no comparable

Wgures for the earlier period. The table is in two parts: Part A gives the

percentage of the population living in cities of over 100,000 inhabitants;

Part B gives the proportion living in towns and cities with over 20,000inhabitants.

The table shows that by 1910 Germany was the third most urbanized

country in Europe after Britain and the Netherlands. There is a sign-

iWcant gap between Germany and the fourth-placed country on both

measures. In 1870, by contrast, Germany was behind France and Bel-

gium on both measures, and around the same level as Italy, Denmark,

and Ireland (the relative position depending on the measure used).

The increase in urbanization between the two dates (measured as the

annual gain in percentage points) is larger in Germany than in any

other country. The German gain during the period 1871–1910 was 16.5

percentage points using the ‘over 100,000’ measure and 22.2 using the

‘over 20,000’ measure. The next-placed country was England and

Wales with gains of 12.1 and 18.6 respectively (1871–1911). The gain

for France was just 5.7 percentage points on the ‘over 100,000’ meas-

ure, and 9.3 percentage points for the ‘over 20,000’ measure.

Urbanization in nineteenth-century Germany proceeded at a much

faster pace than in other European countries. The pressure on urban

services and amenities was correspondingly high, as was the pressure on

urban labour markets in the Lewis Model framework. Where high urban

growth resulted from high inward migration, this also meant that a

higher portion of the population was exposed to the experience of social

dislocation resulting from moves from a rural society still dominated by

traditional attitudes and values to a very diVerent urban environment.

Table 3.3 provides Wgures for a more detailed comparison of

Germany with England and Wales over the whole of the nineteenth


Table 3.2. Urbanization in various European countries, c. 1870

and c. 1910

A: Living in cities >100,000

Date of

census

% Date of

census

%

England and Wales 1871 25.8 England and Wales 1911 37.9

Scotland 1871 22.7 Scotland 1911 30.1

Denmark 1870 10.2 (Netherlands) (1909) (29.7)

France 1872 9.1 Germany 1910 21.3

Belgium 1866 8.1 Denmark 1911 16.8

Ireland 1871 7.8 Ireland 1911 15.8

Italy 1861 5.4 France 1910 14.8

Germany 1871 4.8 Belgium 1910 11.0

Austria 1869 3.7 Italy 1911 10.9

Sweden 1870 3.3 Austria 1910 10.6

Norway 1875 0 Norway 1910 10.3

Switzerland 1870 0 Sweden 1910 9.2

Finland 1870 0 Switzerland 1910 8.6

Netherlands not available Finland 1910 4.7

B: Living in towns and cities >20,000

Date of

census

% Date of

census

%

England and Wales 1871 42.0 England and Wales 1911 60.6

Scotland 1871 31.8 Scotland 1911 49.5

Belgium 1866 15.4 Germany 1910 34.7

France 1872 14.8 Belgium 1910 28.8

Germany 1871 12.5 Italy 1911 28.2

Ireland 1871 11.6 Denmark 1911 27.0

Italy 1861 11.1 France 1910 24.1

Denmark 1870 10.2 Ireland 1911 21.5

Norway 1875 8.5 Switzerland 1910 21.1

Austria 1869 6.5 Norway 1910 18.1

Switzerland 1870 6.4 Austria 1910 16.9

Sweden 1870 5.8 Sweden 1910 16.1

Finland 1870 1.6 Finland 1910 9.3

(Netherlands) (1909) (40.4)

Source: Data from Flora et al. (1996), 249–80; the French and Belgian results useddiVerent-size classes and were adjusted to make them comparable by extrapolating thedensity function.


century. This shows that in 1801 England was much more urbanized

than Germany in 1871. Interpolating between the census years, it

appears that Germany reached the English 1801 Wgure of 23.0% in

1893. Germany then moved to 34.7% in 1910. Using the same ap-

proach on the results for England and Wales shows that this Wgure was

attained around 1853. So, while it took Germany 17 years to get from

23.0% to 34.7%, it took England and Wales 52 years to make the same

move. Urbanization was three times as fast in Germany than in England

during the equivalent period.5

Looking at results for individual cities, it is clear that one reason why

urbanization was so fast in Germany was that all the main cities grew

fast. The annual rates of growth of the Wve largest cities in 1871–1900

were:

Table 3.3. Percentage of the population living in towns and

cities >20,000

Germany England and Wales

Date of census % Date of census %

1801 23.0

1811 24.4

1821 25.9

1831 28.5

1841 30.7

1851 33.6

1861 38.2

1871 12.5 1871 42.0

1880 16.1 1881 47.9

1890 21.0 1891 53.7

1900 28.8 1901 58.2

1910 34.7 1911 60.6

Note: A comparison with France shows that the move from 4.8% of the population livingin cities of over 100,000 to 14.8% took 59 years in France (1852–1911) but 26 years inGermany (1871–97).Source: Data from Flora et al. (1996), 262, 278.

5 There is no simple relationship between urbanization and economic growth, but thiscomparison provides another piece of evidence (if more were needed) indicating that therate of growth in Britain during the industrial revolution was relatively slow.


Berlin 3.70%

Hamburg 2.85%

Munich 3.52%

Leipzig 3.66%

Dresden 3.22%6

In Britain there were cities which matched these rates of growth:

Liverpool grew at 3.30% per annum in 1801–31; Glasgow at 3.03% in

1801–41. But there were other cities which grew less fast: Birmingham

at 2.4% in 1801–31, and London at 1.95% in 1801–41. The slow growth

of London, which was already a large city in the eighteenth century, is

particularly important in holding back the overall rate of increase.7

Comparison with cities outside Western Europe shows that there

were American cities which matched or surpassed German rates of

growth: New York grew at 3.9% per annum in 1825–71 and Chicago

went from a population of 80,000 in 1855 to one of 420,000 in 1876, an

annual growth rate of 8.2%. Massive immigration made these rates of

growth possible.

But the German rate of urbanization was unusually fast by the stand-

ards of nineteenth-century Europe: faster than any other European

country had experienced. This meant that internal migration was also

relatively high. The natural rate of population increase for Berlin was

only 0.95% per annum in 1841–1900; it required high rates of inward

migration to raise the annual growth rate to 3.7%.

3. Measuring Gross and Net M igration

As mentioned in the Introduction, there was an obligation on the

German population to register a change of residence. This created a

source of data on migration which is particularly valuable in that it

gives information about total moves. The obligation did not apply to

temporary changes and there may also have been some evasion. In 1902

the Berlin statistical oYce published adjusted Wgures which took

account of under-recording, the extent of this being estimated from a

comparison with the census results. No adjustment to the amount of

6 Figures from Schrott (1903), who provides a valuable adjustment to the populationWgures which takes account of boundary changes.

7 British Wgures taken from the 1879 edition of the Encyclopaedia Britannica, articles onvarious cities.


inward migration was considered necessary, but the numbers of out-

ward moves were adjusted upwards by 17%. However, it is possible

that the Berlin police were unusually vigilant. Others may have been

more lax. It is noticeable, for example, that recorded gross migration is

signiWcantly lower for Bavarian cities than in other parts of Germany.8

Keeping these reservations in mind, these Wgures still show a society

which had high levels of mobility. From 1898 onwards the available city

statistics were collected and published in the Statistisches Jahrbuch

Deutscher Stadte. By 1907 this included Wgures for 48 cities, with a total

population of 11.3 million. These cities recorded a total number of

moves in of 2.06 million and moves out of 1.81 million. Total moves

amounted to 346 per 1,000 head of population.9 There will, however,

be some double counting in this (moves from one city to another will be

counted twice). Moreover, some individuals will have left and come

back in the same year, thus counting as two moves.

The adjusted series produced by the Berlin statistical oYce runs

from 1841 to 1902. For this study the series has been extended to 1912

using the same adjustment procedure, and the results of the 1905 and

1910 censuses. Figure 3.1 shows the results.

There are four periods which can be distinguished. The Wrst runs

from 1841 to the late 1850s. In this period overall mobility is relatively

low. On average total moves are 141 per 1,000 head of population in

1841–59. Net gains are 1.2% per annum on average, but these Xuctuateconsiderably. In the second period, from 1860 to the early 1870s, overall

mobility is much higher and rises to a peak of over 270 moves per 1,000

in 1871–3. Net gains are also high, averaging 3.4% in 1860–73,

although these do still Xuctuate.10

It is worth taking note of the timing of this phase: it coincides with

the onset of faster industrial growth brought about by the boom in

railway construction. It is a period in which agricultural prices were

tending to rise. The increase in mobility was not connected to any

agricultural crisis.

8 Berlin Wgures from SJB (1903), 158; Wgures for all cities were published annually inSJDS.

9 SJDS (1908), 39; all the main cities are included with the exception of Munich.Total moves includes all moves to and from each city; moves within the city boundariesare not included.

10 This is partly due to the inXuence of the war years 1866 and 1870; in both theseyears net migration is low, presumably as a result of the conscription of many potentialmigrants.


In the third period, from 1874 to 1890, mobility is below the 1871–3

peak although there is an upward trend. Net gains are substantial,

although, at 2.2% per annum, these are lower than in the previous

period. From 1890 onwards net migration falls, while gross migration

rises, eventually rising above the level of the 1871–3 peak. By 1910–12

Berlin was no longer gaining population through migration.11 So the

Wnal period is one of high mobility, a Xexible pre-1914 labour market,

but not one of high net migration. Whether this is typical of Germany

as a whole is a question which will be considered later.

What was the reason for this high mobility? There was a seasonal

element, which leads one to conclude that some migrants did return

home to help with farmwork in the summer. Figure 3.2 gives a monthly

breakdown of migration in and out of Berlin in 1901–2.12 There was a

11 The chart shows a net loss; however, this is partly due to a shift in population toareas which were not included within the pre-1914 city boundaries: Charlottenburg,Neukolln, Schoneberg, Spandau, Lichtenberg, and Wilmersdorf. Including these in a‘Greater Berlin’ unit (and allowing for some under-recording of moves out) alters theoverall result to one of balance: moves out equalled moves in.

12 Data from SJB (1903), 159; the moves out have been adjusted upwards by thestatistical oYce.

1841−100

0

100

200

300

400

1849 1857 1865 1873 1881 1889 1897 1905

Net moves

Total moves

Fig. 3.1. Net and gross migration, Berlin 1841–1912: net moves and total

moves per 1,000 head of population

Source: The original series was in SJB (1903), 158; the extension uses annual data fromSJDS, adjusted for the census results for 1905 and 1910.


substantial move out in March, at the start of the sowing season, and a

smaller outwards movement in the summer months. In all Berlin lost

0.76% of the total population in the period February–July and gained

0.70% in the remaining months. However, this is probably a serious

underestimate of the true movement; many of those who returned for

just a few weeks in the summer may not have needed to register a

change of address. In general, however, this is not a major contribution

to overall mobility. As the chart shows, there were a large number of

moves in both directions in most months. Monthly net migration is

small relative to gross migration, as is annual net migration.

Results from other cities also show a seasonal pattern. Figure 3.3

provides a similar analysis of moves to and from Cologne in 1912–13.13

Here the seasonal pattern is a rather diVerent one: there is net outwards

migration in the June–August period, equivalent to 0.5% of the total

population, and gains in the remaining months amounting to 2.9% of

the total population. Again, monthly net migration is small relative to

gross migration.

13 SJK (1914), 30; the moves out are not adjusted, so these may be underestimated.

Jan.

−2

−1

0

1

2

3

Mar. May July Sept. Nov.

Net moves

Moves in

Moves out

Fig. 3.2. Net moves, moves out and moves in, as percentages of the total

population, Berlin 1901–2 (averages of the two years)

Source: Data from SJB (1903), 159.


It is also worth considering what aspects of migration are obscured or

downgraded by the tendency to concentrate on net migration. Two

studies are of interest. The Wrst is an analysis by the Berlin statistical

oYce of the occupations of migrants in 1910. In this year there were

263,991 registered moves into Berlin and 206,552 moves out, an appar-

ent gain of 56,839. Table 3.4a compares the breakdown by occupation

of this Wgure with the distribution of total moves (in and out).

In this year Berlin took in numbers of industrial workers, and these

accounted for a third of net moves, but only a quarter of all moves. By

contrast, the professional categories, oYcials, students, and workers in

trade and commerce (a highly mobile group according to census

returns) all occupy a larger share of total moves than net moves.

The Cologne statistical oYce classiWed migrants in 1913 according to

their last place of residence or intended destination. This again can be

used to compare net and gross migration. The results (in Table 3.4b)

show that net migration Wgures tend to be biased towards moves over

longer distances, while the importance of moves within the locality, in

this case the Regierungsbezirk Koln, is underestimated.

Net moves

Moves in

Moves out

Jan.−2

−1

0

1

2

Mar. May July Sept. Nov.

Fig. 3.3. Net moves, moves out and moves in, as percentages of the total

population, Cologne 1911–12 (averages of the two years)


To summarize, using Wgures for net migration concentrates attention

on the movement of migrants over substantial distances into urban

industrial jobs, but there are other short-distance moves which are also

an important part of internal migration in the period: a student going to

a large city to study; a doctor or oYcial moving to a diVerent post; a

small trader trying his or her luck in a series of diVerent localities. Most

of the rest of this study will be concerned with net migration, or moves

revealed by the census comparison of place of residence with place of

Table 3.4a. A comparison of the distribution by occupation of net

moves and total moves, Berlin 1910

% share in

net moves

% share in

total moves

Occupied in agriculture and Wshery 1.15 0.78

Occupied in industry 33.62 26.54

Occupied in trade and commerce 15.74 20.01

Domestic servants 25.60 23.09

General labourers 16.63 11.98

Professional, state oVicials, and military personnel 1.67 7.40

Pensioners and rentiers 1.14 1.31

Students 0.65 4.20

UnclassiWed 3.80 4.71

Note: Unoccupied family members are included, classiWed according to the occupation ofthe head of the household.Source: Calculated from SJB (1908–11), 215, which classiWed migrants into 43occupational categories. These have been amalgamated for use in Table 3.4a.

Table 3.4b. A comparison of the distribution of net moves and total

moves, by last place of residence or intended destination, Cologne 1913

% share in

net moves

% share in

total moves

Rest of Regierungsbezirk Koln 18.4 29.4

Rest of Rhineland province 43.5 38.8

Hesse-Nassau and Westphalia 12.7 12.1

Rest of Germany 25.5 19.7

Source: Calculated from SJK (1913), 27.


birth. But it should be borne in mind that these moves represent just

the visible part of a much larger iceberg.

4. Sources of M igrants

Where did migrants come from? Did this change over time? The analy-

sis of census data provides one way to answer these questions. By

looking at cohorts, census returns sorted by the date of arrival in the

city, it is possible to compare the origins of migrants in diVerent

periods. Two points should be noted in advance. The Wrst, following

from the discussion in the previous section, is that this will only capture

permanent moves, or at least moves which had not been reversed by the

time of the census. The second is that the analysis will be aVected if

there are variations in death rates according to migrant origins. If

migrants from the rural east, for example, live in poorer quality hous-

ing, and have lower life expectancy in consequence, then there will be a

tendency for these to be under-represented in earlier cohorts.

It is possible to make such an analysis by cohort of the 1910 Berlin

census. This is shown in Table 3.5. In 1910 the Berlin census asked for

the last place of residence prior to migration. This makes it possible to

compare those migrants who moved directly from their place of birth to

those who made an intermediate move.

The Wrst point to note is the decline in the numbers of moves from

the areas nearest to Berlin, especially the provinces of Brandenburg,

Prussian Saxony, and Anhalt. These provided over half the migrants in

the pre-1870 period, but only a third after 1900. There is a shift in

favour of the eastern provinces in the period to 1890, but thereafter the

east’s share declines. By 1906–10 migrants were drawn much more

evenly. There were rises in the share contributed by the west and

north-west, and by the other central provinces, and there was also a rise

in the foreign contingent.

There is a problem here: return migration could have the eVect of

boosting the east’s share over time, as eastern migrants were less likely

to return or make further moves. Thus, the eastern share will tend to

rise for cohorts furthest from the census date. Fortunately, there are

other ways to establish the trend of net migration and these will be

used in a later section.

Looking at the balance between migrants who moved from their

place of birth, and those who had already made at least one move, it is


clear that there is a considerable change. Over 70% of the pre-1870

cohort were on their Wrst move. By 1906–10 this had dropped to below

30%. There is also a shift in the source of migrants, from smaller towns

and rural areas to cities (the census data makes it possible to group

migrants from 87 major towns and cities, the Berlin suburbs not

Table 3.5. Analysis of the 1910 Berlin population by arrival cohort:

percentage distribution of those born outside the city by last place of

residence prior to migration (both sexes)

Date of arrival in the city

1906–

10

1901–

5

1896–

1900

1891–

5

1886–

90

1881–

5

1876–

80

1871–

5

pre-

1870

All moves:

Regbez. Potsdam 16.5 16.0 16.4 15.2 14.2 16.4 17.3 15.8 22.4

Other Brandenburg/

Pr. Saxony/Anhalt

16.7 17.9 19.5 19.3 19.5 22.3 23.5 23.4 28.1

Other Centre 8.3 7.0 6.2 5.4 4.8 5.1 4.6 3.7 3.4

total centre 41.5 40.8 42.0 39.9 38.5 43.8 45.5 42.9 53.8

East 37.5 45.5 46.7 49.2 53.1 48.4 47.5 51.3 41.4

West and

north-west

10.6 6.9 5.8 5.9 4.4 4.5 4.0 3.0 2.9

South and

south-west

3.1 2.1 1.7 1.8 1.4 1.2 1.2 0.9 0.5

Foreign 7.3 4.8 3.7 3.2 2.6 2.1 1.8 1.9 1.4

Moved from place of birth:

total 40.6 44.4 48.2 49.7 53.0 55.8 60.7 63.7 71.1

From towns and

rural areas

29.2 34.4 39.4 41.1 44.8 47.7 52.8 56.5 61.9

From cities 8.7 8.1 7.3 7.3 7.2 7.3 7.1 6.4 8.6

Moved from somewhere other than place of birth:

total 59.4 55.6 51.8 50.3 47.0 44.2 39.3 36.3 28.9

From towns and

rural areas

36.9 35.0 33.8 32.3 30.5 28.8 25.7 24.2 19.3

From cities 18.0 17.7 15.8 16.1 15.0 14.1 12.6 10.9 8.9

Source: These calculations were made using material from microWlm tapes kindly madeavailable by Professor Dudley Baines, which are now in the library of the London Schoolof Economics. These contain the full published record of the Berlin censuses 1861–1910.The results were summarized using the regional classiWcation set out in Appendix 2C.The data is drawn from table V.3 of the 1910 census, Berlin Statistisches Amt (1913–16),i. 32–9.


incorporated in the city, and the industrial regions of Zabrze and Bor-

beck). But it should be noted that these urban areas only contributed

27% of migrants in 1906–10; the majority still came from rural areas

and small towns.

Comparison with the results of earlier censuses is made diYcult by

the fact that these only recorded place of birth, not last place of resi-

dence. It is possible to take encouragement from the Wnding that most

of those who arrived in the earlier years had come from their place of

birth, but this is still an obvious source of discrepancy between the two

data sources. Table 3.6 provides an analysis of the population resident

in 1890, by place of birth and date of arrival. This shows a steady rise

in the proportion coming from the eastern provinces. In the 1870s and

1880s about half of all migrants came from this area. At the same time

there was decline in the proportion coming from the area around

Berlin, which is particularly marked in the case of the nearby Kreise:Teltow, Niederbarmin, and Osthavelland.

At this stage it would be premature to attempt to draw Wrm conclu-

sions, but the pattern which is emerging suggests that the German

Table 3.6. Cohort analysis of the Berlin population 1890: percentage

distribution of those born outside the city by place of birth

(both sexes)

Date of arrival in the city

1881–

90

1871–

80

1861–

70

1851–

60

1841–

50

1831–

40

1821–

30

pre-

1820

Nearby Kreise 6.3 6.4 7.4 10.2 12.3 16.4 20.0 22.6

Rest of Regbez.

Potsdam

9.4 11.2 13.5 16.1 17.8 20.7 21.5 20.9

Other Brandenburg/

Pr. Saxony/Anhalt

20.7 23.1 26.6 30.8 32.0 30.3 28.1 24.8

Other Centre 3.5 2.6 2.3 2.5 3.2 3.0 4.2 5.7

TOTAL CENTRE 39.9 43.3 49.8 59.6 65.3 70.4 73.8 74.0

East 49.3 50.3 45.0 34.9 28.6 24.2 20.4 15.2

West and north-west 6.2 3.9 3.5 4.1 4.3 3.8 3.3 5.1

South and south-west 1.6 0.8 0.5 0.5 0.8 0.7 0.9 2.4

Foreign 3.1 1.7 1.2 0.9 1.0 0.9 1.3 4.9

Source: Calculated from data in Berliner Statistisches Amt (1893), table V.2; the regionalclassiWcation is set out in Appendix 2C.


experience of migration during the period of industrialization can be

divided into three periods:

1. Early (up to the 1860s): low mobility, most migrants drawn from

the regions nearest to the cities.

2. Middle (from the 1860s to the 1890s): higher mobility, and a

major shift of population out of the rural east.

3. Late (after the mid-1890s): a highly mobile, Xexible labour

market, but one less dominated by the absorption of a labour

surplus from the rural east.

The cohort analysis made by the Berlin statistical oYce was also used

to examine the extent to which migrants from diVerent regions made

further moves after arrival. The basic requirement was an assessment of

the probable death rate of a given cohort. After this had been deducted,

a comparison of cohort numbers at diVerent census dates gave the

percentages of those who remained and those who moved again. This

was not an estimate of return migration as such, but return migration

formed a part of the total of those moving. Ideally it would have been

desirable to adjust for diVerences in likely death rates, but this infor-

mation was not available, and the statistical oYce applied the same

death rate to all groups (though diVerent rates were applied to each

sex). Some error would have resulted from this. If migrants from the

east, having lower wages and poorer living conditions than others, had

higher death rates then there will be a tendency to overestimate the

numbers moving: some of those assumed to have moved having, in fact,

died.14

Table 3.7 provides a summary of this analysis, using the 1875, 1880,and 1890 censuses. The table shows that those who moved over short

distances, coming from the adjacent areas of Brandenburg and Prussian

Saxony, were less likely to make further moves. Of those who moved

from more distant areas, those coming from the eastern provinces have

much lower rates of further migration when compared to the rest of

Germany. Two other points which emerge are the high rates of further

migration for foreigners and the lower rates for women compared to

men.

14 The view that the poor had higher death rates is well supported: see Spree (1988) forexample. There is no evidence that eastern migrants had lower wages; nevertheless, givenlower wage levels in the region of origin it can be assumed that reservation wage levelswould be lower for eastern migrants facing similar costs of migration, and it is thereforeprobable that wages would have tended to be lower.


This may appear a surprising result. The likelihood of a temporary

move could be expected to be inversely related to distance: individuals

might make short moves intending to return, to take advantage of un-

usual short-term employment opportunities, but over long distances the

cost of the move would mean that most would be intended to be per-

manent. Instead the reverse seems to be the case.

It might be that short-distance migrants are better informed, and

therefore less likely to make mistakes, but examination of Figure 3.4,

which gives the rate of further moves in 1880–90 by region, shows that

there is no simple relationship with distance: the more distant provinces

such as East Prussia, Alsace-Lorraine, and the Rhineland show very

diVerent results.

Figure 3.4 suggests that there are two factors which aVect the like-

lihood of further moves. The Wrst is what might be termed ‘cultural

aYnity’: migrants from Alsace-Lorraine and from the Catholic areas of

southern Germany were less likely to make permanent moves to Berlin.

The same point applies to foreigners. The second is the income level of

the region of origin. Migrants from the richer western and north-western

provinces (Schleswig-Holstein, Hanover, Rhineland, and Westphalia)

were more likely to make further moves. So this analysis shows that there

was a fundamental diVerence between migration from the east and

Table 3.7. Estimates of the numbers making further moves after

arrival, Berlin 1875–80 and 1880–90

% of each migrant cohort who move away from Berlin

Arrived pre-1875,

moved in 1875–80

Arrived pre-1880,

moved in 1880–90

men women men women

Arrived from:

Brandenburg/

Pr. Saxony

25.3 15.4 26.0 21.8

East 27.6 16.8 25.4 19.0

Rest of Germany 43.6 23.4 46.4 32.1

Foreigners 51.0 35.6 53.3 44.7

Source: Calculated from data in Berlin Statistisches Amt (1893), 44.


migration from other areas: the Wrst was a permanent shift of a popula-

tion surplus in the sense that it represents the rectiWcation of an imbal-

ance between the available labour force and perceived relative

opportunities in the region of origin; the second was part of a general

pattern of rising mobility.

The east was the principal source of surplus labour for the Berlin

labour market in the period. But there were other streams of long-

distance migration, seasonal as well as permanent. Figures 3.5 and 3.6

illustrate this by making use of a survey of local authorities in the 1860s

(at the Kreis level or equivalent). The administrators were asked about

migration to or from their district. Many replied in general terms

(‘much movement at harvest time’ or such like), but some identiWed

particular destinations or sources, and sometimes gave reasons as well.

>700

600−700

550−600

500−550

450−500

400−450

350−400

300−350

250−300

200−250

Berlin

Fig. 3.4. Estimated numbers making further moves in 1880–90 by region:

numbers moving again per 1,000 of those arrived in Berlin pre-1880 from each

region (men only)



The survey was not a comprehensive one, but it does provide a useful

picture of movements in Germany at the time.15

Figure 3.5 gives some examples of long-distance migration connected

with agricultural activities. It shows that, even in the 1860s, Germany

was attracting foreign seasonal labour at harvest time, although in

this period the main sources were Bohemia, Sweden, and Italy. There

were also harvest-related movements inside Germany though these were

mainly over relatively short distances. Mostly the movements were from

15 The authorities were not asked about the means of transportation used by thesemigrants. Some of these movements were suYciently short to have been possible on footor using horse-drawn transport. Others were so long as to imply the use of the railwaynetwork.

Italy - Harvest and brick-making

Bohemia- Harvest

Eisenach

Holland- Harvest

Sweden - Harvest

Peat-cutting

Sugar beet

Peat-cutting

Fig. 3.5. Examples of long-distance migration connected with agricultural

activities

Source: Examples taken from survey replies given in von der Goltz (1875). There wassubstantial harvest-related migration within Germany over shorter distances; Bertold et al.(1990), 268 has a map of these movements.


east to west, though there were some exceptions, such as the seasonal

migration from the Eisenach region to eastern Silesia at harvest time.

The sugar beet areas of central Germany were attracting seasonal mi-

grants, although the authorities indicated that peat-cutting in these

regions was also an important reason for migration to these areas.

Figure 3.6 gives some examples of migration connected with industrial

or urban activities. It shows, Wrst, those regions which speciWcally men-

tioned outward migration to Berlin or the Ruhr. The latter area was

mainly drawing migrants from western or central Germany in this period

(large-scale migration from the east came later). Berlin was attracting

migrants from eastern Brandenburg, and parts of Silesia and Posen.

These were, it appears, relatively concentrated streams, from a limited

number of areas, compared to the more widespread eVect of railway

construction. The Wgure shows the districts reporting major movements

Areas reportingmigration to railwayconstruction

Fig. 3.6. Examples of long-distance migration connected with industrial or

urban activities: areas reporting moves to Berlin or the Ruhr; areas reporting

migration to take part in railway construction

Source: As for Figure 3.5.


to railway construction. These were in predominantly rural areas in the

south, the centre, and in the east. Comparison with the previous Wgure

suggests that railway construction may well have Wtted in with the pre-

industrial pattern of seasonal migration at harvest time: young people

from regions accustomed to provide harvest labour to other areas went to

the railways instead. By contrast, moves to cities represented a more

fundamental change in lifestyle and social surroundings.

The transition from a pre-industrial to an industrial pattern of migra-

tion was not a sudden change for the supplying regions: there were

several stages. Regions might start with a tradition of harvest-time out-

migration, then move on to temporary migration connected with

railway building or other major construction projects. It was then a

possibility that a portion of those who made temporary moves for these

reasons might decide to make permanent moves. This in turn would

establish migrant cores, which could then expand as those making the

initial moves were joined by friends or other family members.

5. D i fferences in M igration by Sex

Many of the statistics on migration are available for both sexes. In this

study these are amalgamated unless there is some evident diVerence

which might be of interest, such as that shown in Table 3.6. Otherwise

every table would be twice as long.

There are indeed diVerences in the migration record between the

sexes. In general men moved more. In 1901–11 in Berlin there were

annually, on average, 301 moves per 1,000 for the male population, but

only 222 moves per 1,000 for the female population (total moves to and

from Berlin, Wgures adjusted by the statistical oYce). But this higher

rate of gross migration did not mean that net migration was necessarily

less for women. In fact in 1891–1912 the female population of Berlin

gained proportionately more through migration: an average annual gain

of 0.46% against one of 0.39% for the male population. Looking at the

country as a whole, and dividing it into 17 fairly large regional group-

ings, the 1907 census found that 14.7% of men were living outside their

region of birth, and 12.4% of women. This is a signiWcant diVerence,

but one which shows that women were by no means immobile, even if

they moved less often than men.16

16 Calculations from the analysis of the 1907 Occupational Census published in SJDS(1909).


Male migration was more volatile. Figure 3.7 compares annual

net migration for each sex for Berlin during the period 1891–1912.

The magnitude of the swings in male migration is clearly greater,

and evidently related to economic conditions. For example, the so-called

Elektroboom of 1898–1900, which was particularly marked in Berlin, pro-

duced a larger male inXow, and the collapse of this boom in 1901–2 led to

a net outXow of men, but not to a net outXow of women.

Looking at the historical record, there is evidence to show that

female mobility rose more slowly than male. Figures 3.8 and 3.9 make

use of the cohort analysis of the Berlin population in 1890 and 1910

given in Tables 3.5 and 3.6. In the Wrst of these, the proportion of

migrants who came from the central provinces is compared for both

sexes; in the second the analysis is repeated for the east. Two sets of

lines are shown. As mentioned in the discussion in section 4 above, the

census results are not comparable: the 1910 census classiWes by last

place of residence prior to migration; the 1890 census classiWes by place

of birth. So there are separate lines (A and B) for each census.

The Wgures show that in the Wrst half of the nineteenth century there

was a marked diVerence between the two sexes: female migrants were

much more likely to be drawn from the central provinces, and less

likely to come from the east. Female migration was predominantly

191119071903189918951891

30

20

10

0

−10

−20

−30

Men

Women

Fig. 3.7. Net migration of men and women, Berlin 1891–1912, per 1,000

head of population of each sex (moves out adjusted by the statistical office)

Source: Calculated from data in SJB (1913), 158.


short-distance. As migration from the east rose, female migration lagged

behind male migration, reaching the 30% mark about 20 years later, for

example. But by the 1860s the gap had narrowed considerably. And

from 1870 onwards there is no tendency for more female migrants to

come from the central regions, and the eastern proportion is rather

higher for female migration than for male.

In general, the diVerences between male and female migration are not

so large after 1870, at least when net migration is the main focus of

interest. The main diVerences are in the pre-1870 period, and in the

fact that men moved more often, and thus contributed a higher propor-

tion of total moves. Young men made more frequent moves than young

women, but women were nearly as likely to make permanent changes of

residence.

1905189518851875186518551845183518251815

80

70

60

50

40

30

Men (A)

Women (A)

Women (B)

Men (B)

Fig. 3.8. Percentage of migrants drawn from the centre, by arrival cohort,

Berlin 1815–1910

A 1890 census results classified by place of birth

B 1910 census results classified by last place of residence

Source: Figures given in Tables 3.5 and 3.6.


6. German Emigration

The Lewis Model deals with the operation of an economy seen in

isolation, without considering the geographical context: How easy is

migration across international frontiers? What are the conditions in

neighbouring countries? Are wages in these countries lower or higher?

In the model, the urban wage is determined by conditions in domestic

agriculture: by the existence, or non-existence, of a labour surplus in

this sector. But if there is migration from neighbouring countries with

lower wages, then the whole wage structure will be inXuenced by condi-

tions in these countries. And if there is substantial out-migration then

1905189518851875186518551845183518251815

60

50

40

30

20

10

Men (A)

Men (B)

Women (B)

Women (A)

Fig. 3.9. Percentage of migrants drawn from the east, by arrival cohort, Berlin

1815–1910

A 1890 census results classified by place of birth

B 1910 census results classified by last place of residence

Source: Figures given in Tables 3.5 and 3.6.


the internal labour market will be aVected by developments in the

recipient countries.

German external migration underwent a major structural change

around 1895. Comparison of census results with recorded births and

deaths for the territories included in the post-1871 German empire

show a total net loss of 2.34 million in 1841–71 (an annual average of

78,100) and 2.46 million in 1871–95 (an average of 102,500). But after1895 this Xow came to an end. Between 1895 and 1910 the net loss was

just 14,000, an annual average of less than a thousand.

Calculations based on census returns may underestimate the extent of

the change. By 1910 Germany was admitting large numbers of foreign-

ers on temporary permits. Holders of these permits were required to

leave the country by December, and then re-apply if they wished to

continue to reside in Germany. As German censuses were carried out

in December, those who complied with this regulation would have been

omitted from the census total.17 A census taken in some other month

would have shown a larger population, and the calculated rate of migra-

tion into Germany would have been greater. Germany may well have

already been a net gainer through migration.18

If so then this was an unusual development. Of the other European

countries, only Belgium was a net gainer before 1914. Britain did not

start to gain population through migration until the 1970s. This change

had the eVect of reinforcing the links between Germany and the other

labour markets of central and eastern Europe, and decoupling Germany

from the converging labour markets of the ‘Atlantic economy’.

Discussion of German emigration in the nineteenth century is dom-

inated by German emigration to the United States. Partly this is due to

the nature of the statistical sources. There are records of German over-

seas migration, both from the ports of departure and from the recipient

countries, but there is very little on movements across land frontiers. In

1912 the German statistical oYce made a survey of foreign censuses, to

establish the numbers of Germans living outside the Reich. The Wgures

are given in Table 3.8. The statistics are not fully comparable: some

countries recorded all those born in Germany; others only included

those who still retained German citizenship. Some countries which lack

suitable census data are excluded: Brazil is the most important of these.

17 Those who chose not to comply with the regulation would have had a good reason toevade the census.

18 Return migration was also important: see Schniedewind (1994) for a study of this.Other studies of German migration include Bade (1980 and 1991) and Marschalck (1973).


The pre-eminence of the United States as an emigrant destination is

evident. But quite large numbers went to neighbouring states, although

these in return provided migrants to Germany, reducing net migration.

The most substantial Xow of net migration out of Germany to a

European destination was to Switzerland (there were 68,000 Swiss

living in Germany in 1910).

The United States accounted for 91% of recorded overseas migration

in 1871–1913. Figure 3.10 shows that there were years when this fell to

80% or below. Typically these were years when migration to the

United States was low, but there is no evidence that migrants chose

other destinations when conditions were unfavourable for migration to

the United States (migration to the United States was not negatively

correlated with Xows to other destinations).

Towards the end of the period there is some evidence that German

migrants were looking for alternative destinations. There was a rise in

migration to Canada and Argentina. However, the Xows were not large

compared to the movements to the United States in the period up to

1895. By contrast, after 1900 British emigration switched to the

Empire, which took 65% of those leaving in 1911–13 (up from 28% in

1891–1900).19

Comparison of migrant destinations with the regional origins of mi-

grants is possible from German Wgures from 1899 onwards. Figure 3.11

gives migration to the United States as a percentage of all recorded

migration to non-European destinations, 1900–9. The main areas from

which there was migration to alternative destinations were a group of

central regions—Saxony, Prussian Saxony, and Berlin-Brandenburg.

19 British Wgures from Thomas (1973), 57; the German Wgures are calculations usingWgures for migrant destinations from SJDR.

Table 3.8. Germans living in foreign countries 1910–11

United Statesy 2,501,000 Britain and

Ireland y56,000 Argentina 27,000

Switzerland 220,000 Netherlands 38,000 Luxembourg 22,000

Austria-Hungary 126,000 Denmark 35,000 Chile 11,000

France 90,000 Australiay 33,000 Italy 11,000

Belgium 57,000

yCountries who recorded all those born in Germany, all others recorded only Germancitizens.Source: Kaiserlich Statistisches Amt (1916), 3*.


This group includes some important industrial centres. At the other

end of the scale, in the agricultural regions of eastern and southern

Germany there was little migration to destinations other than the

United States. The result is that there is a positive correlation between

the percentage occupied in agriculture in each region, and the percent-

age going to the United States (a coeYcient of 0.56), indicating that the

Xow to the United States was more ‘agricultural’ than the Xow to other

destinations.20

100

80

60

40

20

01871 1876 1881 1886 1891 1896 1901 1906 1911

Asia

Africa

Australia

Other North andSouth America

Brazil

United States

Fig. 3.10. Percentage distribution of recorded migration to non-European

destinations (from German sources) 1871–1913

Note: The German figures initially covered only migration from German ports andAntwerp. Later this was expanded to include migration from French ports. Migrationfrom British ports was not included, which affected the figures for migration to Australia.Source: Data from SJDR (1885), table 5, p. 23 and subsequent issues.

20 The negative correlation between the percentage going to Brazil (the next most im-portant destination) and the percentage occupied in agriculture is even stronger (a coeY-cient of �0.72). This Wnding that the ‘more advanced’ regions sent migrants to the ‘lessadvanced’ destination is intriguing although perhaps not of any great signiWcance. It has aparallel in the better-known division between northern and southern Italy (the south wasmore likely to send migrants to the United States), Hatton and Williamson (1998), ch. 6.


However, the basic conclusion has to be that the story of German

emigration is, in aggregate, the story of German migration to the

United States. There was no other Xow which was large enough to have

a signiWcant eVect on labour market conditions inside Germany. The

record of migration to the United States is presented in Figure 3.12.

Only moves from Germany to the United States are given; there are no

Wgures for return migration before 1909.

The chart includes two series for German migration to the United

States, one from German sources and the other from US immigration

records. The two sources agree relatively well. Between 1890 and 1900there is little diVerence. Before 1890 the discrepancy is probably due to

gaps in the German records. After 1900, the diVerence is less easy to

explain. The US authorities may have classiWed a number of German-

speaking immigrants from Russia as Germans. This could explain why

<86%

86−88%

88−90%

90−92%

92−94%

94−96%

96−98%

>98%

Fig. 3.11. Migration to the United States as a percentage of all migration to

non-European destinations 1900–9, by region

Source: Calculated from data in SJDR (1901–10).


the discrepancy rose after 1900. Some travellers, intending to stay for a

limited period, may also be classiWed as immigrants in the US Wgures.21

The migration Wgures show three main peaks: in the early 1850s, in

the late 1860s and early 1870s, and in the period 1881–93. After this

German emigration to the United States remained at a low level. These

changes can be linked to the relative movement of the German and US

economies: emigration was high when conditions were favourable in the

United States or unfavourable in Germany. But after 1895 this relation-

ship becomes much weaker: the collapse of the Elektroboom in 1900–1

produced only a modest rise in emigration, for example.22

1910190018901880187018601850184018301820

300

200

100

0

Emigration:US data

US−Germanwage ratio

Emigration:German data

Fig. 3.12. Annual German emigration to the United States (in 1,000s) and the

US-German wage ratio

Source: Migration figures from USDepartment of Commerce (1975), and SJDR, annualissues; wage ratio calculated fromWilliamson (1995), 164–7. The source for the German wagedata is a series for unskilled building wages in Rostock, Nuremburg, and Berlin, originallycollected by Robert Kuczynski, also published in Kuczynski (1945) and Bry (1960).

21 There are large discrepancies in statistics of German migration 1900–13 for all themajor non-European destinations. These are discussed in Burgdorfer (1926). One problemwas German emigration through British ports, but there is no obvious reason why thisshould only have caused diYculty after 1900.

22 Grant (2002d ) contains a detailed analysis of the decline of German emigration tothe United States.


Figure 3.12 also gives the US–German wage ratio, calculated from

Williamson’s series in purchasing power parities. There was a major

disturbance to this series in the 1860s, as a result of the American Civil

War. Once allowance is made for this, it is clear that there is no evident

tendency for wages in the two countries to converge. Germany was an

exception to the general tendency for wages in the Atlantic economies

to move closer. One reason for this was that declining emigration to the

United States, combined with increasing immigration from eastern and

central Europe, had the eVect of strengthening links with other over-

supplied labour markets in these regions, and weakening those with the

converging Atlantic economies.23

German emigration to the United States declined for a number of

reasons, due to changes within Germany and within the United States.

One factor was increased competition within the United States from

sources of ‘new immigration’: countries in southern and eastern Europe

which supplied an increased proportion of US immigrants from the

1890s onwards.24 Another was a reduction in the availability of land in

the United States which suited German farming skills as the frontier of

settlement moved into more arid western regions. But a further possible

reason is that Germany was moving out of the labour surplus phase

when migration, internal and external, was dominated by the release of

surplus agricultural labour.

German emigration had a strong agricultural character. Figures 3.13and 3.14 show the variation in the intensity of migration by region, for

two periods: 1871–9 and 1890–9. The regional rates of emigration (per

head of resident population) are expressed as percentages of the national

average. The variation between regions is striking: in 1871–9 the aver-

age annual rate of emigration was 0.44 per 1,000 in East Prussia, and

4.53 in West Prussia, ten times as large. In the 1870s the highest rates

of migration were in the rural east, in West Prussia, Mecklenburg, and

Pomerania. There were also high rates in the north-west, and in

the south-west, which had been an area of high emigration earlier in the

23 Important studies on convergence in the international economy of the late nine-teenth century include O’Rourke et al. (1994 and 1996) and Taylor and Williamson(1997). Although Germany contributed large absolute numbers of migrants in the yearsbefore 1895, the numbers were not as high relative to the size of the German populationas for Ireland and some Scandinavian countries. This meant that the convergence processwas not a strong one even in the years when emigration was high.

24 In the literature on US immigration these are the ‘new immigration’ sources, todistinguish them from the traditional sources: Britain, Ireland, Germany, and the Scandi-navian countries.


century. The pattern was not greatly altered by the 1890s: the highest

rates were then in Posen, West Prussia, and Pomerania, the

Mecklenburg rate having dropped.

There is a relationship between migration intensity and the percent-

age of the population engaged in agriculture. Rates of migration tended

to be below average in the more industrial regions, as can be seen from

the maps: Westphalia, Saxony, Silesia, the Rhineland, and Berlin-

Brandenburg all had low rates of emigration. Table 3.9 gives correlation

coeYcients for various periods, which show that this association

between high emigration and high rates of agricultural employment was

at its highest in 1885–94, and thereafter tended to fall.

Although German migrants came largely from agricultural regions

the statistical sources show that they were not predominantly occupied

as farmers or farmworkers immediately prior to departure. Table 3.10

250−300

200−250

150−200

125−150

100−125

80−100

60−80

40−60

20−40

<20

>300

Fig. 3.13. Regional emigration rates as percentages of the national average,

1871–9

Source: Calculated from data in SJDR (1881), table 5, pp. 14–15 and subsequent issues.These include all recorded migration to non-European destinations. The earlier figuresomit migration through French ports, which may affect the rates from some westernprovinces.


compares the previous occupations of German immigrants with those of

other groups. This shows that relatively few German immigrants had

previously been occupied in agriculture.25 The proportions were, how-

ever, rather higher than for the other ‘old migrant’ groups.

25 Marschalck (1973), 79–80 gives the Wgures from German sources; see alsoMarschalck and Kollmann (1973), 537.

Table 3.9. Correlation between rates of emigration by region

and the percentage occupied in agriculture in each region, for

diVerent periods, 1871–1913

1871–4 0.54 1885–9 0.59 1900–4 0.53

1875–9 0.47 1890–4 0.61 1905–9 0.45

1880–4 0.51 1895–9 0.48 1910–13 0.32

Source: Annual data from SJDR for recorded immigration by province.

250−300

200−250

150−200

125−150

100−125

80−100

60−80

40−60

20−40

<20

>300

Fig. 3.14. Regional emigration rates as percentages of the national average,

1890–9

Source: Calculated from annual data in SJDR.


However, this may be misleading. If there is surplus labour in agricul-

ture, then the Wrst to move will be the younger sons of farmers, and the

children of landless labourers. These will be unable to Wnd work in

agriculture, and may well start by taking non-agricultural jobs in their

region of birth. But if they are determined to work in agriculture

somewhere, then the realization that only emigration to a country where

land is readily available oVers this possibility will eventually make them

decide to move. However, the records will show that these are people

whose occupations prior to migration were outside agriculture, in a rural

handcraft sector or as unskilled non-agricultural labourers.

The recorded occupations of migrants after arrival in the United States

also show that relatively few found work in agriculture immediately. But it

does appear from the US census returns that German migrants had a

stronger preference for farming than some other groups, notably the

British. Table 3.11 gives Wgures for the percentage of the male labour

force who were classiWed as farmers by the United States. As this shows,

by 1910 German migrants were twice as likely as British migrants to be

farmers, having started from a position of equality in 1870.

The table also shows that only a relatively small proportion of the

German-born population were farmers, a rather lower Wgure than the

Table 3.10. Occupations of immigrants to the United States prior

to departure (as percentages of migrants 1899–1910 excluding those

without an occupation)

Irish Scandinavian German English

Professional 1.5 1.3 3.9 11.3

Farmers 0.3 2.5 3.0 2.4

Skilled workers 13.4 21.5 35.0 53.9

Unskilled: farm workers 4.8 7.6 18.0 2.5

Unskilled: non-agricultural 80.1 67.0 40.2 29.8

Note: The table gives results for the four largest ‘old migrant’ groups. Amongst the‘new migrant’ groups the Poles, Slovaks, and Croats had higher proportionspreviously occupied in agriculture than the Germans (‘nationalities’ are as classiWedby the US immigration authorities, which means that Polish-speaking migrants aretreated as a separate group).Source: Calculated from US Immigration Commission (1910–11), 100–76; thedistinction between skilled and unskilled is largely that used by the immigrationauthorities; the subsequent division into categories (industrial, handicrafts, etc.)follows the author’s classiWcation.


one for the whole US labour force. Even if the 4.9% of the 1910

German-born workforce who were farm-workers is added to the

numbers for farmers, the total is only 22.3%, which is well below the

equivalent Wgure for the United States as a whole (35.1%). Most

German migrants were in non-agricultural jobs.

However, once again, this comparison may well lead to an under-

estimate of the importance of agricultural opportunities to German

migrants. A higher proportion of immigrants may have become farmers

eventually. A typical immigrant life-cycle might include an initial

period of non-agricultural work to pay oV debts incurred during the

passage and to accumulate capital before applying for a Homestead land

grant. Some were obliged, by the terms under which they had travelled,

to take non-agricultural jobs to pay oV debts to steamship companies

and their agents.

The sudden decline in German emigration around 1895 Wts with the

picture given earlier, of a ‘labour surplus’ phase of migration out of the

rural east which came to an end in the late 1890s. Migrants who wanted

to remain in Germany went to the major cities. Those who were more

adventurous, or determined to go to a country where land was available,

or had contacts with Germans already in the United States, went across

the Atlantic. This trans-Atlantic ‘safety valve’ reduced the pressure on

urban labour markets quite considerably in 1880–5. In 1880 around

55% of those leaving the eastern provinces of Germany went abroad.

By the early 1900s this had changed: in 1900–5 less than 10% of those

leaving these regions left the country.26

Table 3.11. Farmers as a percentage of the US male labour force

1870–1910

1870 1880 1890 1900 1910

As a % of the total US labour force 27.8 28.3 27.1 22.1 19.5

As a % of those born in Germany 21.5 24.3 21.2 20.6 18.4

As a % of those born in Britain 20.2 19.8 14.7 13.1 9.0

German–British ratio 1.06 1.23 1.44 1.57 2.04

Source: Census results from US Department of Commerce (1975) and Hutchinson (1956);the 1870 Wgures are estimates using the movement of the total German-born populationengaged in agriculture. (Changing census deWnitions may account for some of the jumpsin the Wgures, but the relative movements should be unaVected.)

26 Calculated indirectly from a comparison of natural population movements, recordedpopulation, and registered emigration by region using Wgures published annually in the


7. Immigration into Germany 1870–1913

At the same time as German emigration was declining there was an

increase in immigration into Germany. The German population census

showed a substantial increase in the numbers of foreigners living in

Germany, from 433,000 in 1890, to 779,000 in 1900 and 1,260,000 in

1910. This indicates that the net annual rate of immigration was around

41,000–46,000 in the 1890s and 57,000–64,000 in the 1900s.27 But the

census was taken in the month of December, and this would tend to

aVect the numbers since migrants on short-term permits were expected

to return home in this month. The occupational census of 1907, taken

in June, gave the numbers of foreigners living in Germany as 1,342,000,

which was greater than the Wgures from the December censuses for

1905 (1,030,000) and 1910 (1,260,000). But even this may have been

below the peak, which is likely to have come in the harvest months of

August and September.

The total numbers who came to Germany on temporary permits

were considerable. Russian Wgures show that, on average, 755,000 tem-

porary eight-month passports were issued each year between 1910 and

1913 to those wishing to work in Germany and their families. Over the

same period, the German Wgures show that 484,000 work permits were

issued, on average, each year to Russians (including Poles and Germans

of Russian nationality), and 267,000 to other nationalities. Italian emi-

gration Wgures show that, on average, 69,000 Italians went to Germany

each year between 1910 and 1913, and the Germans issued 54,000 work

permits to Italians.28 So the numbers of workers, and their dependants,

coming to Germany on a temporary basis each year could have

exceeded one million.29 How long they stayed is a more diYcult ques-

tion. Some could have come for a few weeks, for a harvest job just over

Statistisches Jahrbuch des Deutschen Reichs. The resulting Wgures are approximate. Therewas some tendency to under-record external migration, but this has to be set against thefact that return migration was not recorded at all.

27 To get this estimate a mortality rate had to be applied, which should be lower thanthe German average to allow for the diVerent age structure of the migrant population.The range was produced by using, Wrst, a mortality rate half the overall German rate, andthen one 85% the overall rate. Data from SJDR (1913).

28 These were mostly seasonal workers; comparison of the Italian emigration Wgureswith the German census Wgures indicates that 90–95% of Italian migrants in 1900–10went back each year. Ferenczi and Willcox (1929), 810 and 828–31; see also DelFabbro (1996).

29 There might well have been a signiWcant amount of unrecorded migration fromother areas, from Bohemia for example.


the frontier; others may have stayed for the full eight months allowed

by the Russian passport.30

What is also uncertain is the extent to which this was a new develop-

ment, rather than the imposition of a system of passports and permits

on a Xow which was already in existence. Russian Wgures show a steep

increase in the numbers of passports, from an annual Wgure of around

100,000 in 1900. German Wgures show a more moderate rise in the

number of permits: 733,000 of these were issued in Prussia alone in

1907 (all nationalities); 768,000 in the whole of Germany in 1913.31

Surveys in the 1890s suggested that the use of seasonal migrant labour

in German agriculture was already substantial.32 It may well be that it

was the extension of this to German industry which was the novelty.

Table 3.12 gives the numbers of foreigners living in Germany

according to the December censuses. The Wgures show that the largest

contribution came from Austria-Hungary. The numbers of Russians

was relatively small (more came in the summer on temporary permits).

The Xow of migrants from Austria-Hungary has attracted less attention,

but it was also an important inXuence on the German labour market.

Table 3.12. Numbers of foreigners, by country of birth, resident in

Germany 1900, 1905, and 1910

1900 1905 1910

Russia 47.0 106.6 137.7

Austria-Hungary 391.0 525.8 667.2

Switzerland 55.5 62.9 68.3

Italy 69.7 98.2 104.2

France 20.5 20.6 19.1

Spain and Portugal 0.9 1.3 2.0

Belgium, Netherlands, and Luxembourg 113.5 127.6 172.0

Scandinavia 38.9 41.1 39.2

Britain and Ireland 16.1 17.3 18.3

Balkans including Greece and Turkey 4.1 5.8 8.1

Others 21.8 23.2 24.0

totals 779.0 1030.4 1260.0

Source: Kaiserlich Statistisches Amt (1916).

30 For a description of the work permit system, see Bade (1984), also Nichtweiss (1959).31 Burgdorfer (1929), 379; there are no Wgures for before 1905.32 Weber (1893).


Attention has also tended to focus on the Xow of labour into agricul-

ture. The 1907 occupational census showed that there were 288,300

foreign-born agricultural workers, 3.3% of the labour force. There were

rather more foreign industrial workers (501,000) and these were 4.6%

of the industrial labour force. The industries most dependent on for-

eign labour were mining and metals (7.5% foreign), quarrying (8.8%),

and construction (7.4%).33

Looking at the regional distribution of workers in these three indus-

tries, it is also noticeable that the highest concentrations of foreign

workers were not in the east but more in the south-west (Figure 3.15).

The Wgure for Alsace-Lorraine (27%) is perhaps something of an

anomaly, but there are also high Wgures for Baden (14.3%) and Wurt-

temberg (12.0%). By contrast the distribution of foreign agricultural

>1

1−2

2−4

4−6

6−8

8−10

10−12

12−14

14−16

>20

Fig. 3.15. Foreign-born workers as a percentage of all those occupied in mining

and metals, quarrying, and construction, 1907

Source: 1907 occupational census, SDR (1908), n.f. 210.

33 SDR n.f. 210 (1908).


workers is more or less as expected (Figure 3.16). The highest Wgure

was the 11.3% of the Mecklenburg agricultural labour force which was

foreign-born. Next came Prussian Saxony-Anhalt with 8.1%.

Although the conventional view may be that the typical pre-1914

migrant was a Russian Pole working in eastern agriculture, this shows that

a typical migrant was just as likely to be someone from Austria-Hungary

working in a quarry or on a building site in south-western Germany.

8. M igration and Labour Surplus

The previous section has stressed the growing links between Germany

and the labour markets of eastern Europe, particularly Russia and

Austria-Hungary. In combination with the decline in emigration to the

United States, which weakened the connection between Germany and

<1

1−2

2−4

4−6

6−8

8−10

10−12

12−14

14−16

>20

Fig. 3.16. Foreign-born workers as a percentage of all those occupied in

agriculture, June 1907

Source: 1907 occupational census, SDR (1908), n.f. 210.


the converging Atlantic labour markets, this development reinforced the

‘Continental’ nature of the German labour market, insulating it from

the forces driving the process of convergence in the Atlantic economies.

It also meant that one source of elastic labour supplies—the labour

surplus in German agriculture—was supplemented by another—migra-

tion from neighbours to the east. This in turn held back the increase in

German wages.

This was an inevitable consequence of the eastern movement of the

process of industrialization in Europe. Both inside Germany and out-

side Germany industrialization led to increased mobility, traditional

institutions which promoted immobility were eroded, and improved

communications made migration attractive and aVordable. In the last

quarter of the nineteenth century these same forces were unleashed in

the countries to the east and the south of Germany. This in turn set oV

a new wave of migration to the United States, displacing potential

German emigrants, and also increased migration across Germany’s land

frontiers to the south and the east.

A theme which runs through the chapter is the identiWcation of

periods within the 1870–1913 timespan covered by this study, and the

characterization of these periods. Evidence presented points to the im-

portance of the years around 1895 as a dividing point. After this exter-

nal migration is greatly reduced, and the rate of migration out of the

east falls. There is also a shift in the characteristics of the typical

migrant, from being someone with no previous experience of urban life,

to a person who had already moved at least once, and who was more

likely to have come from another city.

The importance of the 1895 caesura is demonstrated by Wgures which

will be presented in Chapter 9. These show that the Kuznets Curve,

the relationship between income inequality and economic development,

peaks around these years. The years up to 1895 are characterized by

rising inequality and high net migration: the labour surplus phase of the

Lewis Model. After 1895 there was a highly mobile labour market, but

static or falling inequality; Germany was moving out of the labour

surplus phase.

Appendix 3A. A Note on Regional Units

The regions used in pre-1914 German statistics are a source of some

diYculty in that they vary considerably in size, and the boundaries


Table 3A.1. Regions and regional groupings

Regional

grouping

Regional unit used

in this study

Includes the following

regional units used

in German statistical

publications

East East Prussia East Prussia

West Prussia West Prussia

Pomerania Pomerania

Posen Posen

Silesia Silesia

Mecklenburg Mecklenburg-Schwerin

and -Strelitz

Centre Berlin/Brandenburg Berlin and Brandenburg

Pr. Saxony/Anhalt Provinz Saxony and Anhalt

Thuringia Saxony-Weimar,

-Altenburg, -Meiningen,

and -Coburg-Gotha,

both Reub and both

Schwarzburgs

Saxony Saxony (Kingdom of)

West and north-west Schleswig-Holstein Schleswig-Holstein,

Hamburg, and Lubeck

Hanover/Oldenburg/

Brunswick

Hanover, Oldenburg,

Brunswick, and Bremen

Westphalia Westphalia, Waldeck,

and both Lippes

Hesse-Nassau Hesse-Nassau

Rhineland Rhineland

Hesse Hesse

South and

south-west

Bavaria excl. Pfalz

Pfalz

Baden

Wurttemberg/

Hohenzollern

Alsace-Lorraine

Bavaria excl. Pfalz

Pfalz

Baden

Wurttemberg and

Hohenzollern

Alsace-Lorraine


reXect a historical process: Prussian expansion leading to eventual

German uniWcation, not any Napoleonic striving for geographical logic.

Therefore, all the statistics in this study are presented in tables or

Wgures which use regional units designed to eliminate the smaller

regions. On occasions, these are then grouped into four main regional

categories: east, centre, west and north-west, and south and south-west.

The basic system is shown in Table 3A.1. In addition, wherever pos-

sible, Oldenburg is split: the separate enclaves of Lubeck and Birken-

feld being included in Schleswig-Holstein and Rhineland respectively.

For some purposes, chieXy agricultural, the term ‘east-Elbian’ is used:

here Berlin/Brandenburg is added to the other eastern provinces.

For the tables and maps presented in this study a certain amount of

additional work was required to produce results which conformed to

these boundaries. The Wgures could not be taken directly from oYcial

publications.


4

Migration in Germany 1870–1913:

A Statistical Analysis

1. Introduction : A Note on Methodology

and Source Materials

The study of migration has been one of the most active areas of

German historical research in recent years. This has produced a

number of excellent local studies, making use of regional archives.

These have described the eVects of large-scale migration: the process of

assimilation, the problems faced by rapidly expanding cities, the process

whereby an industrial workforce was created from a stream of rural

migrants.1 The response of the German state and the upper echelons of

German society to these problems has been examined, and, at times,

found wanting. Studies have repeatedly shown pre-1914 Germany to be

a society of high inequality, and limited social mobility.2

The aim of this book is not to produce another local study. It is

instead to try to grasp the process in its entirety: to put migration in a

theoretical context which explains the origins of the problems described

by the regional studies, and to answer the question ‘Why is migration in

this period so important?’ The Lewis Model, the concept of a ‘labour

surplus’ phase in economic development, provides such a context, and

the application of this model is therefore a central theme.

The remit of the study is therefore a wide one. The aim is to test

certain hypotheses about the causes and consequences of migration.

The level is the national one: not what was happening in certain cities

or regions, but the eVect on Germany as a whole. And, wherever pos-

sible, hypotheses have been expressed in quantiWable terms.

1 Examples include Barfuss (1985), Bolsker-Schlicht (1997), Del Fabbro (1996),Jackson (1997), Kollmann (1960), Kosters-Kraft (1998), Lourens and Lucassen (1999),Oberpennig (1998), Reich (1998), Wenneman (1997), Crew (1979), Catt (1986), and thevarious studies in Hoerder and Nagler (1995), Robler (1995), and Bade (1987). Studies ofmigration with a wider remit include Matzerath (1985), Caumanns (1994), Kollmann(1959), Tipton (1976), and Hochstadt (1999).

2 Kaelble (1986).

This is not because this is regarded as the only way to approach the

subject, or as necessarily superior to other techniques of historical en-

quiry; it is because this is a type of study not currently available. The

idea is to produce a work which is complementary to the regional and

local studies: for example the collection published as People in Transit.3

Moreover, the main emphasis has been on the use of statistical sources

available in nineteenth-century government publications. This period

was one in which statistical enquiry was developing, both in scope, in the

areas covered by published statistical series, and in the use that was made

of the results. There was a belief that the proper use of statistics could

lead to a better understanding of social problems and, at times, produce

solutions. The classic case was the use of mortality tables to prove the

connection between breast-feeding and low infant mortality.4

The new German empire had an Imperial Statistical OYce, which

oversaw the collection of national statistics and the administration of

censuses. There were also state statistical oYces, which collected a lot

of the data and conducted enquiries of their own. The most important

was the Prussian Statistical Bureau, headed by Ernst Engel in the 1860s

and 1870s, but others also made interesting contributions: the Saxon

Statistical OYce made a series of studies of income distribution in the

1900s which were unusually comprehensive for the time. In addition,

all the major cities had statistical oYces, and typically published a

statistical yearbook. The most important of these was the Berlin Statis-

tical Bureau, whose director in the years before the First World War

was Robert Kuczynski.

The analysis of these materials used the techniques available at the

time: where causation was suspected a series was ranked by one vari-

able, and results given for a second variable. This could show a rough

correlation: if the eye running down the page could pick up a tendency

for the second variable to rise or fall. But regression analysis was not

used, and any relationship involving more than one explanatory variable

could not be studied.

The result is a vast amount of statistical material presented in a form

which is not easy to understand or digest. Taking one example: the

1907 Zeitschrift of the Prussian Statistical Bureau contained an article

by Dr M. Broesike on internal migration in Prussia which included

sixteen of these ‘visual correlation’ tables, some of which extended over

two or more pages, and contained as many as 489 observations (an

3 Hoerder and Nagler (1995). 4 Berlin, Statistisches Amt (1893), 82.

98 Migration 1870–1913: A Statistical Analysis

example is shown as Figure 4.1).5 There were also a number of other

tables containing basic statistical material. This was a barrier to compre-

hension greater, in most respects, than the obstacle presented by the

terminology of modern econometrics.

So the approach used in this study has been to seek, in the Wrst place,

to achieve a thorough knowledge of the materials collected and made

available by the various statistical oYces, and secondly to subject these

to examination using modern econometric analysis.

For the general reader the statistical results have been summarized in

a series of tables which are included in the main text of this chapter,

5 Broesike (1907).

Fig. 4.1. Part of a ‘visual correlation’ table from Broesike (1907)

Note: This is part of a table covering 489 Prussian Kreise. Broesike has classified them bythe dominant farm size; the first column gives the percentage of the total agricultural areain holdings of this size. He is looking for a relationship between this and rates of migrationfor 1895–1900 and 1900–5 (the two columns at the right). He also includes a classification(Kb or Gb) according to the next most important farm size class, and follows this with thepercentage of the total agricultural area in holdings of this size. So, the reader is asked tolook down the two right-hand columns, and consider if migration tends to rise or fall withthe dominant farm size, keeping in mind the possible effect that might flow from thefarm size sector ranked second.

Migration 1870–1913: A Statistical Analysis 99

and in subsequent chapters which also make use of econometric analy-

sis. These provide a ‘verbal exposition’ of the econometric results. A

more detailed analysis is then presented in appendices, which may be of

interest to those readers who are well versed in statistical matters. They

can, however, be skipped without losing any essential part of the argu-

ment presented in each chapter.

2. The Changing Nature of M igration , 1871–1913

The basic hypothesis considered in this study is that Germany went

through a ‘labour surplus’ phase of economic development, and that

this phase was coming to an end by the late 1890s. The shift to a more

modern pattern of migration post-1900 created conditions which were

favourable to political stabilization and the transition to a mature demo-

cratic society. The Lewis Model provides the theoretical basis for these

claims.

One of the best sources of Wgures for migration is the indirect calcu-

lation of net migration by region based on the comparison of census

returns. For each census year from 1871 to 1910, the Imperial Statis-

tical OYce compared the actual increase in population by region with

the recorded numbers of births and deaths since the last census (gener-

ally Wve years before). Deducting the recorded natural increase from the

actual increase produced a Wgure for net migration. This was done for

eighty-one regions and eight periods, thus producing a panel data set of

648 observations.

The reliability of these calculations will depend on the accurate

recording of births and deaths. By and large, this was not unsatisfactory

in Germany in the period. The main question raised by demographers

concerns the recording of stillbirths in the Wrst half of the century, but

this is not a problem as it does not aVect the overall rate of natural

increase (the issue is whether these were recorded as a birth followed

by a death or left unrecorded).6

One obvious problem is that the destination of migrants is not

recorded by this procedure. Other data sources will be examined in the

next section which do provide this information. It is, however, possible

to make the same calculation for Germany as a whole, which provides a

6 See the discussion of this in Knodel (1974).


Wgure for net external migration. This gives an idea of the balance

between external and internal migration.

In 1900 the Imperial Statistical OYce extended the analysis back to

1841. In this pre-uniWcation period censuses were organized by the

diVerent states, and did not always take place in the same year. In

addition, boundary changes imply that a calculation of net external

migration, for the territories included in the post-1871 state, will in-

clude a certain amount of internal migration within French and Danish

territories (pre-1871 and pre-1864 respectively).

For these reasons the pre-1871 Wgures should be treated with some

care, although they are probably reliable for Prussian territories. The

overall record 1841–1910 is given in Table 4.1.

The table uses four regional groupings. Of these two are consistent

sources of net outwards migration: east, and south and south-west. But

the trajectories are very diVerent. The east was not a major source of

net outwards migration prior to 1861. Indeed, parts of the east were

gaining through migration: East Prussia and West Prussia both made

gains in 1851–61. Thereafter, outwards migration rises until, in the

decade 1880–90, the region is losing 1% of its total population annually

Table 4.1. Net migration 1841–1910, in average annual rates per

100,000 of the resident population

East Centre West and

north-west

South and

south-west

All Germany

1841–51 �55 þ37 �273 �301 �167

1851–61 �65 �98 �331 �433 �243

1861–71 �342 þ22 �260 �284 �229

1871–5 �738 þ323 þ119 �365 �192

1875–80 �431 þ48 �99 �203 �174

1880–5 �959 þ35 �252 �559 �427

1885–90 �1038 þ490 þ167 �268 �137

1890–5 �661 þ99 þ50 �259 �177

1895–1900 �815 þ351 þ449 �60 þ35

1900–5 �662 þ316 þ285 �113 þ18

1905–10 �663 þ225 þ217 �192 �51

Source: Pre-1871 data from SDR n.f.150, pp. 190*–193*; post-1871 data from SDR a.f.25(2) Ubersicht 1B, pp. 50–2 (1871–5 Wgures) and subsequent census publications,combined in the regional units given in Appendix Table 3A.1.


through migration. This rate of out-migration then falls back in the

decade after 1900. In the south and south-west, by contrast, out-

migration was already substantial in the pre-1871 period. There is a

continuous outXow, with Xuctuations, until 1895, although the rates are

well below those in the east. After 1895, the rate of out-migration is

much reduced.

The experience of the two gaining regions is also diVerent. In the

pre-1871 period the west and north-west was also a major source of

out-migration. This was a period in which this region contributed heav-

ily to German migration to the United States. But after 1871 this

changed. There are periods of in-migration as well as out-migration.

After 1885 the region consistently absorbs migrants, post-1895 this is

on a larger scale. By contrast there is consistent migration into the

central region; the only period in which outwards net migration is

recorded is 1851–61. The rate of absorption Xuctuates, with peaks in

1871–5 and 1885–90, but it is more consistent after 1895.

The story in summary is that, in the early part of the century, there

was a substantial labour surplus in western Germany, north and south.

This was chieXy exported, although some was transferred to the centre

and east. In the rural east the amount of land under cultivation was still

rising so any local surpluses could be mainly absorbed within the

region. From the 1860s onwards industrial growth in the west and the

centre meant that more migrants could be found employment within

these regions, but at the same time the east emerged as a major source

of out-migration. Before 1895 the rate of progress of the industrial

regions was not suYcient to take in all the migrants out of the east and

the south, resulting in a continuing high rate of emigration for

Germany as a whole (Wnal column of Table 4.1).

Statistical analysis of the migration Wgures can be used to extract

additional information from the census results.7 Although the published

census returns do not distinguish between agricultural and non-

agricultural areas, the use of modern econometric techniques can pro-

duce estimated rates of net gain and loss for each period. These are

given in Table 4.2.

The table shows that there were four periods in which the net rate of

out-migration from agricultural areas exceeded 2.0% per annum, and

7 Results of the regressions used to make these calculations are given in Appendix 4B,which also contains a discussion of the statistical issues involved, and further commentson the pattern of migration revealed by this analysis.


these also produced annual rates of gain for the non-agricultural areas

of at least 1.0%.8 After 1900, the rate of migration drops to one which

produces a signiWcantly lower rate of gain for the non-agricultural

regions: the rate of gain for 1905–10 is the lowest of all.

This does not mean that the overall level of migration had necessarily

dropped; rather it had changed its character from one dominated by the

transfer of surplus labour out of agriculture, to a pattern where this was

one stream amongst others, although still important. One way of dem-

onstrating this change is to consider what percentage of the pattern of

net gains and losses can be explained purely by the extent to which the

region specializes in agriculture. In a ‘labour surplus’ period a large

proportion of migrant moves are from agricultural to non-agricultural

areas, and most of the variation in net gains and losses by region is due

to this factor. In other periods transfers between industrial regions due

to the relative performance of diVerent industries will be an important

inXuence and the role of rural–urban migration will be reduced. Table

4.3 gives Wgures produced by the statistical analysis of the census

results, which show how German internal migration changed over the

period covered by this study.

8 The rate of gain is lower relative to the size of the non-agricultural population for tworeasons. First, the total non-agricultural population was larger. Secondly, some of thoseleaving agriculture emigrated and so did not transfer to non-agricultural regions insideGermany.

Table 4.2. Results of statistical analysis of net migration by region,

1871–1910: rates of net gain or loss through migration for agricultural

and non-agricultural areas (annual rates of gain or loss as percentages

of the total population)

Estimated rate of net loss for

an entirely agricultural region

Estimated rate of net gain for an

entirely non-agricultural region

1871–5 �2.33 þ1.60

1875–80 �1.25 þ0.73

1880–5 �2.50 þ1.01

1885–90 �2.89 þ1.65

1890–5 �1.64 þ0.68

1895–1900 �2.48 þ1.34

1900–5 �1.58 þ0.77

1905–10 �1.55 þ0.61


The table shows that, before 1900, labour surplus migration was the

most important factor in internal German migration. Even in periods

when the rates of net gain and loss were reduced, such as 1875–80, the

importance of the ‘labour surplus’ component remained high. The peak

for labour surplus migration came in the 1880s: in this period over 50%

of the variation can be explained by this one factor: was the region

agricultural or not? The 1890s are a transitional period, with a less

pronounced agricultural character. By 1900–5 only a third of the pat-

tern of net gains and losses can be explained by the proportion occupied

in agriculture, and by 1905–10 this has dropped to just over a quarter.

The labour surplus period was over.

So, what happened in the 1890s that brought about this change? Was

there a static lump of surplus labour in agriculture which had simply

moved out by then? Or were there other changes which improved the

prospects of the agricultural sector?

Some Wgures on these issues are brought together in Figure 4.2. All

are taken from HoVmann. The two indices of arable producer prices and

imported cereal prices show the eVect that world market conditions had

on German producer prices. World market prices fell from the 1870s

onwards and this brought down German producer prices despite tariV

protection. But from the mid-1890s onwards world market prices began

to recover, and this in turn brought about a recovery of German arable

producer prices. The consequences of this can be seen from the move-

ment of the index of land prices. The index fell slowly from the late

Table 4.3. ‘Rural labour surplus’ migration as a

percentage of all migration (the percentage of the

variation in net gains and losses which can be

explained by the proportion occupied in agriculture)

1871–5 44.2

1875–80 43.5

1880–5 57.1

1885–90 63.9

1890–5 39.8

1895–1900 48.9

1900–5 33.1

1905–10 26.9

Note: These are the R2 statistics from the regressions given inAppendix Table 4B1.


1870s, when it was over 67, to a low point of 59.6 in 1895. From then

onwards it was rising, the level reached in 1913 representing a rise of

68% in eighteen years.

The uppermost line in Figure 4.2 gives industrial wages as a percent-

age of agricultural. In the early 1870s industrial wages were 40–50%

above agricultural wages, and dipped below this in the late 1870s.

Thereafter the ratio rose and remained in the range 170–180% from

the late 1880s onwards. The improved agricultural conditions post-

1895 did not lead to a fall in this ratio, but agricultural wages did keep

pace with industrial wages during this period.

187040

60

80

100

120

140

160

180

200

1875 1880 1885 1890 1895 1900 1905 1910

Wage ratio

Import prices

Land pricesArable prices

Fig. 4.2. Indicators of agricultural conditions, 1870–1913 (all indices have

1913 ¼100)

‘Wage ratio’ is the ratio of industrial to agricultural wages; ‘arable prices’ is an index ofarable producer prices; ‘import prices’ is an index of the price of imported cereals.

Source: Calculated from data in Hoffmann (1965), 492–5, 561–2, 569–70, and 610–11.Hoffman’s index of agricultural workers’ earnings is based on figures for cash payments,but figures given in Pohl (1908) also show gains for workers paid in kind or by harvestshares. Earnings of Insteleute with threshing shares in Oletzko (East Prussia) rose 73%between 1880 and 1903 for example.


The movement of this ratio can also be interpreted in the light of

the migration patterns considered in the previous chapter. In the mid-

century, the industrial regions were drawing migrants predominantly

from nearby rural areas. Consequently, the premium required to per-

suade migrants to come from these areas needed to be only a moderate

one. But by the late nineteenth century the premium had to be suY-

cient to persuade migrants to move over much longer distances. It is

noticeable that the premium did not fall back after 1900, even though

agriculture was then doing better. Industry still had need of long-

distance migrants.

3. Analysing Net M igration by Region , 1871–1910

Scrutiny of the pattern of net migration shown in Table 4.1 in the light

of the evidence on changing agricultural conditions given in Figure 4.2

shows that there is a possible connection between migration and agri-

cultural conditions: the improvement in land prices after 1895 appears

to have had some eVect, reducing the rate of migration out of the east

for example. This raises the possibility that a more precise modelling of

the eVect of changing agricultural conditions on migration might pro-

duce an equation which can be estimated econometrically.

The value of this does not lie only in giving precise values to rela-

tionships which can be guessed at from visual inspection—it also makes

it possible to compare diVerent causes of migration. There are two

issues which are of particular interest. The Wrst is the eVect of agricul-

tural structure, the extent to which the prevalence of large estates had

an eVect on migration. The second is the relationship between migra-

tion and demographic variables: how far was the regional pattern of

migration caused by diVerences in rates of population increase? Both

these issues will be examined in greater depth in subsequent chapters;

the aim here is to show the part they played in the overall pattern of

migration.

The full set of econometric results is given in Table 4B.2 in the

Appendix; Table 4.4 provides a summary of the results.

The analysis shows that demographic factors had an important inXu-

ence on the pattern of migration: regions with high birth rates tended to

have high rates of out-migration; regions with high death rates took in

migrants to replace these losses. The birth rate which matters most is

the one which aVects current levels of entry onto the regional labour


market. It is the lagged birth rate—the birth rate for 20 years before—

which is used in the estimated equations.

The estimates conWrm that the level of agricultural occupation was

an important inXuence on migration: agricultural regions tended to

shed population and non-agricultural regions made gains. But they also

show that agricultural structure mattered: the presence of large farms

and estates tended to increase out-migration. This question will be one

of the main themes of Chapter 6, but the result appears to conWrm a

point made by a number of contemporary authors, including Max

Weber.9

The eVect of changing agricultural conditions on migration can be

estimated as part of the overall procedure: eVectively this asks whether

migration out of agricultural areas was higher in bad periods for the

agricultural sector—when land prices were low or the rural–urban

wage gap was high. The analysis shows that times of low land prices

were indeed periods in which migration out of rural areas tended to be

9 It was one of the major themes of Weber’s study of the conditions of the agriculturalworkers east of the Elbe, Weber (1893).

Table 4.4. Summary of results of statistical analysis of net migration

for 81 regions, Wve-year time periods, 1871–1910

I. Secure statistical results

Factors which increase net outwards migration:

High birth rates in the past

Presence of large farms and estates

Increases in the rural–urban wage gap

Factors which decrease net outwards migration or lead to net gains

from migration:

High death rates in the current period

High rates of non-agricultural employment

II. Less secure statistical results

Factors which increase net outwards migration:

Low land prices

Note: The classiWcation of results as ‘secure’ or ‘less secure’ is based on the analysis anddiscussion given in the Appendix. The assessment is based on the usual statistical tests ofsigniWcance, plus the ‘robustness’ of the results—the extent to which a coeYcient isaVected by changes in the form of the estimated equation or the inclusion of additionalvariables.


higher, but this result is not a particularly secure one in statistical terms.

A stronger result is obtained when the eVect of changes in the rural–

urban wage gap is included in the estimated equations: upwards move-

ments in this gap led to increased out-migration; falls held back the Xow

out of rural areas.

How important are these variables in the explanation of the overall

pattern of migration? One way to approach this is to try to explain

migration between the four main regional groupings. An important

caveat should be noted: there is a lot of migration within these units

which is not covered by this analysis.

Table 4.5 compares the eVect of predicted migration using the coeY-

cients calculated for the two demographic variables with the actual

Wgures for net migration previously given in Table 4.1.10 The results

show that demographic factors contributed to out-migration from the

east, though the eVect is not large in relation to the total Xow. But the

10 The calculation is a simulation which allows for the eVect through the NETMIGkt�1

term.

Table 4.5. Demography and migration: the estimated eVect of

demographic variables on net migration by regional grouping (columns

headed a) compared with actual net migration, annual rates per 100,000

(columns headed b)


north-west

South and

south-west

a b a b a b a b

1871–5 �24 �738 �15 þ323 �14 þ119 0 �365

1875–80 �27 �431 �17 þ48 �19 �99 �1 �203

1880–5 �29 �959 �21 þ35 �27 �252 �16 �559

1885–90 �34 �1038 �28 þ490 �33 þ167 �26 �268

1890–5 �41 �661 �40 þ99 �43 þ50 �36 �259

1895–1900 �46 �815 �52 þ351 �52 þ449 �44 �60

1900–5 �48 �662 �58 þ316 �55 þ285 �43 �113

1905–10 �58 �663 �66 þ225 �60 þ217 �44 �192

Note: The equation used in these calculations was NETMIGjt ¼ 0:45NETMIGjt�1þ0:26 DEATHRTjt �0:23 BIRTHRTjt�4 (the j subscripts denoting the use of regionalgroupings); these coeYcients are from column a of Appendix Table 4B.2.


same eVect should have led to a similar level of out-migration from the

centre and the west, which in fact gained through migration. In the case

of the south and south-west, there was a very small eVect from the

demographic variables in the period of highest out-migration (1871–85)

and a larger eVect in the period of lower out-migration after 1895.

Some of the reasons for this can be seen from the comparison of the

various regional groupings given in Table 4.6. The east did have a high

birth rate, but it also had a high death rate (and more of the variation in

the death rate fed into migration); the overall rate of natural population

increase was not unusually high. The south had the lowest rates of

natural population increase, yet initially a high rate of out-migration.

The other factors show quite large variations which contribute rather

more to the pattern of migration. Table 4.7 provides standardized esti-

mates of the eVect of each factor on net migration over the whole period

1871–1910 (the standardization adjusts the calculated eVect so that the

national average is zero). The impact of the demographic variables is

small compared to the others. The most important eVect is the percent-

age of the population occupied in agriculture: the south and the east

lost population through migration because they were more agricultural

and because they had less industry. The south and the east diVered in

the importance of larger farms and large estates, and this apparently

caused a further rise in out-migration from the east. For the south, it

caused a small reduction. This also had diVerent eVects on the two

regions which gained through migration.

The eVect of demographic factors on migration will be considered

further in the next section and in the next chapter. In aggregate, it does

Table 4.6. Average values by regional grouping, 1871–1910


north-west

South and

south-west

Death rate, current period

(annual %)

2.63 2.45 2.22 2.62

Birth rate, 20 years previous

(annual %)

4.17 3.95 3.69 3.87

Percentage of land in holdings

of over 100 hectares (1895)

45.6 26.4 7.9 3.1

Percentage of the population

occupied in agriculture (1895)

49.3 25.4 31.3 46.6


not appear that out-migration from the eastern provinces to the rest of

Germany was driven to any signiWcant extent by higher rates of popula-

tion growth than those in other regions. But population growth was

high relative to the limited employment opportunities available in east-

ern agriculture. This drove the demographic surplus to seek employ-

ment in cities both inside the east and in other regions. It was the

combination of demographic increase with the relative decline of agri-

culture which made the east a major source of migrants in the ‘labour

surplus’ period.

But this uses a broad classiWcation. Results in subsequent chapters

show a diVerent picture for individual Kreise and regions. High rates of

population growth at this level were an important factor in migration.

4. Rural–Urban M igration from the 1890 Census

The analysis in the previous section used net migration Wgures calcu-

lated from census returns. This data gives no indication of the destin-

ation of the migrant Xows. But a comparison of place of residence with

place of birth from census returns can provide information of this

type.11

An individual census is a snapshot, describing a moment in time,

which tells us quite a lot about migration but which will leave out many

Table 4.7. Estimates of the eVect of diVerent factors on the pattern

of net migration by regional grouping (annual average net migration

per 100,000), 1871–1910


north-west

South and

south-west

Both demographic variables �3 �2 �3 þ9

Percentage of land in holdings of

over 100 hectares

�135 �31 þ70 þ96

Percentage of the population occupied

in agriculture

�290 þ331 þ178 �220

Note: A similar equation to that given in the previous footnote is used for each variable;the coeYcients used are �0:32 for LAND>100j and �0:168 for % AGOCCPjt.

11 This type of data has been used by Kollmann (1959) amongst others.


moves. Those recorded as living away from their place of birth will

include many who made other moves before reaching their current

place of residence. Some of those living in the region of their birth may

have left and then returned. In addition, there is no way of knowing

which moves are temporary and which are intended to be permanent.

Migration calculated from census records in this way is more than just

net migration (since a calculation of the Xow from one region to another

is possible in both directions) but it will also tend to understate the

overall level of mobility.

One advantage of this type of data is that it is possible to separate

diVerent types of migration. The 1890 census is particularly useful. The

published results provided an analysis of the population resident in

major cities according to place of birth. This can be used to examine

moves to the twenty-seven largest cities from areas outside these cities,

thus excluding moves between major cities, and moves between rural

areas. The analysis can consider whether the likelihood of a move was

aVected by conditions in the region of origin or the recipient city, identi-

fying factors which promoted or retarded migration between urban and

rural areas. The full set of econometric results is given in Appendix 4B,

Section (iii); Table 4.8 provides a summary of the Wndings.

There are a number of general Wndings which provide an overview of

the migration process in the period up to 1890. Migration was strongly

aVected by the distance between the recipient cities and the regions of

origin. If the recipient city lay within the region of origin this increased

the likelihood of a move. This indicates a strong preference for short

moves over long-distance ones. It also means that the proximity of

major cities would have an eVect on overall rates of migration out of

rural areas. In remote regions, the large distances needed to be travelled

to reach the various major cities would have a depressing eVect on

migration in general.

High wages in the recipient cities, and low levels of unemployment

or casual employment as revealed by the 1895 occupational census,

tended to increase migration: moves were more likely to cities with

stable or buoyant labour markets. There were certain industries which

attracted migrants: clothing and construction oVered easy entry into the

labour market for migrants who lacked previous experience of industrial

work. By contrast, cities with high levels of employment in textiles did

not attract migrants: the main period of rapid expansion of these cities

had been in the 1860s and 1870s, and they had less need of migrants

once this period had passed.


In the regions of origin, poor economic prospects, as indicated by a

low rate of formation of Aktiengesellschaften (joint stock companies), had

a positive eVect on outwards migration. This is interesting as it indi-

cates a ‘forward-looking’ element to migration: migrants responded to

the direction of economic change as well as current conditions.

When these Wndings are combined with the result of those given in

Table 4.4, they show how important economic factors could be in

aVecting the course of migration: migrants were more likely to move if

the urban–rural wage gap increased; they chose destinations with high

wages and good chances of employment; they left regions which were

lagging in terms of economic development.

Demography was also important, although the strongest and most

secure results are obtained for the recipient cities: high rates of natural

population increase in these had a negative eVect on inwards migration,

since such cities could Wll a greater proportion of available vacancies

from their own natural increase. By contrast, while it appears that high

rates of population increase in the regions of origin may have promoted

out-migration from these regions, this Wnding is much less secure.

Table 4.8. Summary of results of statistical analysis of

rural–urban moves from 1890 census


Characteristics of the regions of origin which increase outwards

migration:

Proximity to the recipient cities


Poor economic prospects

High rates of productivity growth in agriculture

Characteristics of the recipient cities which increase inwards migration:

High wages

Low rates of natural population increase

Low numbers of unemployed workers or workers without permanent jobs

Presence of the construction and clothing industries

Absence of the textile industry


Characteristics of the regions of origin which increase migration:

High rates of natural population increase

High levels of agricultural employment


Agricultural conditions are shown to have had an important eVect on

out-migration. The presence of large farms and estates increased the

likelihood of outwards migration in the regions of origin. In addition,

agricultural change, as measured by the rate of productivity increase

between 1880/4 and 1893/7, is associated with high rates of out-migra-

tion. This could be an indication that labour was shed as a direct conse-

quence of changes in agricultural technology—the use of mechanized

threshing, for example—or it could show that there was a general pro-

cess of agricultural transformation, involving the replacement of trad-

itional social and economic relations with modern capitalist ones, which

had the eVect of both raising productivity and increasing the likelihood

of migration. Alternatively, improvements in transportation may have

brought new inputs to remote areas, raising productivity, as well as

making it easier for potential migrants to leave.

5. Conclusions

In general the statistical results presented in this chapter and the appen-

dices conWrm the analysis presented in the previous chapter. There was

a period of intense rural–urban migration in the 1870s and 1880s, and

this was aVected by conditions within agriculture relative to those in

industry, and by changes within agriculture. Agricultural transform-

ation brought about the release of surplus labour onto urban labour

markets. After a period of transition in the 1890s, the pattern of migra-

tion changed to one which was no longer dominated by the movement

of surplus rural labour. The ending of this ‘labour surplus’ phase

created conditions which were favourable to the creation of a modern

mature industrial society, and to a process of social and economic rec-

onciliation which could have led to the emergence of democratic insti-

tutions and a reduction in social conXict.


Appendix 4A. Description of Regress ion Variables

and Sources

(i) Appendix Tables 4B.1 and 4B.2; Tables 4.2, 4.3, 4.4, 4.5, 4.6, and 4.7

in text

NETMIGkt

BIRTHRTkt

DEATHRTkt

Net migration, births, and deaths per 10,000 head

of population from census results, from SDR a.f.

20 and 52, n.f. 32, 68 and 150–1, VSDR 1897 iv,

1906 iv, and 1911 iv.

%AGOCCPkt The population occupied in agriculture as a per-

centage of the total occupied population, from the

occupational censuses, from SDR a.f. 4, n.f. 104–6

and 204–6: 1895 and 1907 results are also in Kaelble

and Hohls (1995).

LAND > 100k The percentage of the total agricultural area in

holdings of over 100 hectares, from 1895 Agricul-

tural Census, SDR n.f. 212, pp. 489–500.

LANDPRICEt HoVmann’s index of land prices, HoVmann

(1965), 569–70.

DAGINDWGAPt Change in the ratio of industrial and agricultural

wages, using HoVmann’s Wgures, and comparing

Wve-year averages, HoVmann (1965), 492–5.

(ii) Appendix Table 4B.3, Table 4.8 in text

P(MIG)kj(dependent

variable)

Numbers born in rural part of region k who are

recorded as living in city j, divided by total

numbers born in rural part of k, calculated from

SDR n.f. 68, pp. 136–53.

NATPOPINCk Natural rate of population increase in the region of

origin, 1880–90, from 1885 and 1895 census

results, per 1,000 head of population, SDR n.f. 52

and VSDR 1897 iv.

AGVALUEk The value of Aktiengesellschaften founded 1871–90from VSDR 1907 iv, p. 373; the Wgures are then

divided by the regional population 1895 (1895

census).

%AGRICk The population occupied in agriculture as a per-

centage of the total occupied population, from the


1895 occupational census, Kaelble and Hohls

(1995), 167–88.

DAGPRODk Percentage increase in agricultural value added per

full-time labour unit, 1880/4 to 1893/7, from

regional accounts prepared for this study (see

Chapter 7).

NATPOPINCj Natural rate of population increase for the recipi-

ent city, 1890–5, from SJDS (1892), table viii,

p. 39, and subsequent issues.

AVEWAGESj Average of Ortsubliche Tagelohne for men and

women, 1892, from SJDS (1913), 827–8.

%CONSTRCTj

%CLOTHj

%TEXTILESj

%UNEMPj

Calculated from the results of the 1895 occupa-

tional census, given in SJDS (1898), 350–1; also in

SDR n.f. 107–8.

Appendix 4B. Econometric Analysis

(i) Analysis of net migration 1871–1910, using census data

The post-1871 census results can be used to look at the rates of transfer

between the eighty-one regions (the pre-1871 Wgures use diVerentregional units and so are not comparable). Analysing each time

period separately can show if the course of migration changed—if

there were some periods which were dominated by the transfer of

surplus labour out of agriculture and others which show a diVerent

pattern. Table 4B.1 presents the results of a series of regressions run

for each time period in succession, with a single explanatory variable:

the percentage of the total occupied population which is occupied in

agriculture (%AGOCCP).

What this table shows is the extent to which the pattern of net

migration by region can be interpreted solely as a transfer of labour out

of the more agricultural regions, either to the industrial regions or to

foreign countries. The t-ratios are all signiWcant at the 95% conWdence

level, which shows that the extent to which a region specializes in

agriculture is always a factor in migration, but it is the R2 statistics

in the Wnal column that are the most eloquent. These show the extent

to which the incidence of this form of migration, the transfer of surplus

labour out of agriculture, can explain all the observed variance in net


migration by region. At the peak in 1885–90, with an R2 of 0.639, this

is evidently the dominant form of migration: 64% of the migration

pattern can be explained by this factor alone. But by 1905–10 it only

accounts for around a quarter of observed variance. By this time migra-

tion was more heavily inXuenced by other considerations.12

The slope coeYcients give an indication of the pace of the transfer.

A high coeYcient indicates that a large proportion of the agricultural

population migrated in that period. This peaks in 1885–90, although it is

also high in 1871–5 and 1895–1900. By 1900–10 it is considerably lower.

This table provides clear statistical support for the view that the

pattern of migration shifted from a ‘labour surplus’ one in the 1870s

and 1880s, to a more ‘normal’ one post-1900, where the movement of

population out of agriculture (which still continues even in advanced

industrial economies) is just one stream amongst other migrant move-

ments, not a dominant inXuence on the labour market as a whole.

(ii) Panel data analysis of migration 1871–1910

The separate analysis of each time period can produce interesting con-

trasts in the slope coeYcients, and in the Wt of the estimated equations,

12 The question of non-linearity in the functional form, which might aVect the R2

calculations, was considered. F-tests generally led to the rejection of additional X2 vari-ables. No signiWcant improvement in the R2s was found to be possible.

Table 4B.1. Regressions of net migration by period on % occupied in

agriculture: dependent variable is net migration per 100,000 (annual

rate) for 81 regions

Period constant coeYcient on

%AGOCCP

t-ratio R2

1871–5 1595 �39.22 7.91 0.442

1875–80 734 �19.87 7.80 0.435

1880–5 1006 �35.06 10.36 0.571

1885–90 1653 �45.44 11.84 0.639

1890–5 677 �23.12 7.23 0.398

1895–1900 1335 �38.15 8.70 0.489

1900–5 769 �23.49 6.25 0.331

1905–10 610 �21.63 5.39 0.269

Note: Regression variables and sources given in Appendix 4A.


but combining the data in a single panel makes it possible to test

hypotheses about the reasons for the changes in the slope coeYcients.

The analysis in this section uses the same sources as the previous

section but analyses the data as a single panel.

There are some restrictions on the analysis which follow from the

nature of the data. Only eight time periods are available for the 1871–

1910 period, and the pre-1871 data is not comparable either in terms of

the regional units used or the lengths of the time periods. This means

that any lag structure has to be a simple one: eVectively the only possi-

bility is the inclusion of a NETMIGkt�1 term (net migration for the kth

region lagged one period) amongst the explanatory variables, implying,

by way of the Koyck transformation, a lag structure of the form

E(Xt) ¼ aXt þ a2Xt�1 þ a3Xt�2 þ a4Xt�3 . . . : for the expected values

of the explanatory variables.

An additional problem is that the sum total of regional net migration

is the external balance, which Xuctuates considerably, as shown in

Table 4.2. Ideally, this would be modelled separately, and in a way

which took into account the eVect of alterations in economic conditions

in the main destination for German emigrants, the United States.13

But, given the limited number of time periods, it is not possible to

estimate a complex model of this type for the regional data set.

So, the data set was analysed as a Wxed eVects panel data set, includ-

ing dummies for all time periods, which means that the external balance

is treated as an exogenous factor. The estimated model only deals with

the internal transfer of population between the diVerent regions.

Results are given in Table 4B.2. The Wrst column gives the basic

model: net migration as a function of two demographic variables (the

current death rate and the birth rate lagged 20 years), the percentage

occupied in agriculture and the amount of land in holdings of over 100

hectares. As shown in the previous section, the percentage occupied in

agriculture has an important eVect on migration, but this table shows

that there is a further eVect from the prevalence of larger farms and large

estates: there is a signiWcant negative coeYcient on the LAND > 100kterm, suggesting that this causes additional out-migration.

This point merited further investigation, and it is one of the themes

examined in the next two chapters. In brief, these results show that,

13 A model of this type, using annual data, was included in chapter 6 of the originalD.Phil. thesis, Grant (2000). It is intended to publish a separate article based on thischapter.


once the issue is examined at a lower level of aggregation, using a Kreis-

level data set (with data for around 400 Prussian Kreise), then it is

discovered that what mattered was the geographical position of these

estates, the fact that they were mainly located in remote eastern areas,

not the fact that they were large as such.

The demographic terms are both signiWcant. Their importance is

somewhat obscured by the lag structure adopted; part of the eVectcomes through the NETMIGkt�1 term. The estimated coeYcients

imply that, after Wve periods (or twenty-Wve years), 41% of an increase

in the birth rate lagged four periods would be exported through in-

creased net migration, and 47% of a decrease in the current death

rate.14

The next two columns in Table 4B.2 introduce variables which are

inter-acted with the %AGOCCPkt variable. These are variables which

might aVect the rate of labour transfer between agriculture and indus-

try, and thus explain the variation in the coeYcients on %AGOCCP

shown in Table 4B.1. A negative coeYcient therefore shows that a rise

in this variable increases the rate of transfer from agriculture to indus-

try; a positive coeYcient shows that it holds it back.

The second column uses the change in the ratio between agricultural

and industrial wages to explain the shifts in the pace of migration. This

model produces an improved Wt. Part of the increase in migration be-

tween the 1870s and the 1890s can therefore be attributed to the eVectof the rise in this ratio. Given the lag structure used, this would still be

aVecting migration for some time after this, even though the ratio

ceased to rise.

The third column uses a diVerent variable: the level of land prices as

an indicator of agricultural prosperity. The estimated coeYcient is posi-

tive, indicating that higher land prices did reduce the rate of migration,

but it is not signiWcant at the 5% level (the hypothesis that there is in

fact no eVect on migration from this variable cannot be rejected with

95% conWdence). So, migration is best explained by the equation used

for column b, which uses the change in the agriculture–industry wage

gap. It was the widening of this gap which was responsible for the high

rate of rural urban migration in the 1880s.

What is also notable about the results in Table 4B.2 is the stability of

the coeYcients on the demographic variables, and on the structural

14 The reason for lagging the birth rate 20 years is that the Berlin census results showthat most migrants arrived in their early twenties.


variables LAND > 100k and %AGOCCPkt. The importance of these

variables in the overall pattern of migration is conWrmed. The labour

surplus was created by a combination of demographic factors and the

relative decline of agriculture.

(iii) Analysis of rural–urban moves from the 1890 census

The 1890 census provides Wgures for moves to the twenty-seven largest

cities from those living outside these cities, divided between twenty-

nine provinces. This excludes city–city moves and rural–rural moves.

This data has been used to create a variable Pkj, which is the proportion

Table 4B.2. Fixed eVect regressions (OLS with time period dummies)

of net migration, annual rates per 1,000 population, Wve-year time

periods 1871–1910

a b c

NETMIGkt�1 þ0.454 þ0.469 þ0.450

(5.86) (5.98) (5.73)

DEATHRTkt þ0.260 þ0.231 þ0.247

(3.43) (3.11) (3.27)

BIRTHRTkt�4 �0.230 �0.214 �0.223

(4.26) (4.13) (4.16)

%AGOCCPkt �0.168 �0.163 �0.170

(6.60) (6.42) (6.65)

LAND > 100k �0.0320 �0.0320 �0.0328

(2.78) (2.84) (2.85)

%AGOCCPkt* �0.0119

DAGINDWGAPt (5.50)

%AGOCCPkt* þ0.00380

LANDPRICEt (1.81)

N 648 648 648

F-test of regression 89.10 89.02 83.18

[12,635] [13,634] [13,634]

R-sqd 0.627 0.646 0.630

adj R-sqd 0.620 0.639 0.623

RSS 112213990 106588139 111304403

Note: t-statistics (in brackets) calculated using White’s heterocedasticity-correctedstandard errors. Appendix 4A gives a full description of variables and sources; alternativerandom eVects estimates of the equations are given in Appendix 2B of Grant (2000).These do not alter the basic results although some of the coeYcients did change.


of the total numbers born in province K, excluding those born in major

cities, who move to city J.

The model is a simple linear log-odds ratio model: log (Pkj=(1� Pkj)

¼ f (Xk1 . . . :Xkn, Xj1 . . . :Xjm), where f(.) is a linear function, and

Xk1 . . . :Xkn, Xj1 . . . :Xjm are characteristics of the region of origin K

and the recipient city J used as explanatory variables. The only non-

linear variable is the distance travelled which is entered in a log form.

The functional form produces a logistic curve, thus preventing pre-

dicted values from falling below zero.

The LOGDISTANCEkj term gives the log of the distance in kilo-

metres between the recipient city and the nearest border of the region

of origin. Where the recipient city was within the boundaries of the

region a notional distance of 2 kilometres was used (this could not be

zero as the term was to be in log form). To avoid distorting the results

by this procedure a dummy variable DUM1CONTkj was introduced

for moves to cities within the regional boundaries.

Table 4B.3 gives the results of four diVerent regressions run using

this data. In the Wrst three columns the only variables included for the

Table 4B.3. Regression results for linear log-odds migration

model, 1890 census

a b c d

Constant �5.982 �5.620 �5.367 �5.082

LOGDISTANCEkj �1.331 �1.334 �1.328 �1.281

(26.3) (26.4) (27.2) (30.4)

DUM1CONTkj �0.655 �0.676 �0.668 �0.499

(2.45) (2.53) (2.55) (2.21)

Characteristics of the region of origin:

NATPOPINCk þ0.01088 þ0.01118

(1.26) (1.30)

%AGRICk þ0.01108 þ0.06684

(2.64) (1.43)

LAND > 100k þ0.0154 þ0.0141 þ0.0156 þ0.0155

(6.90) (6.13) (7.54) (8.73)

AGVALUEk �0.00196 �0.00392 �0.00381

(2.13) (4.63) (5.24)

DAGPRODk 0.0140 0.0140

(5.49) (6.37)


recipient city are the level of wages (AVEWAGESj) and the rate of

natural population increase (NATPOPINCj). A number of explanatory

variables which were found to inXuence migration were introduced for

the region of origin. The results in the Wrst column show that the

prevalence of larger farms and estates (LAND > 100k) has a signiWcant

positive eVect on the likelihood of a move from the region of origin.

This conWrms the Wnding reported in the previous section.

The rate of natural population increase has a signiWcant negative

eVect on the likelihood of migration into the recipient city (a high rate

of population increase reduces migration into the city) but the coeY-

cient on the NATPOPINCk term (population growth in the region of

origin) was not signiWcant. This is an intriguing result: variations in

population growth between cities matter; those in rural areas are less

Characteristics of the recipient city:

NATPOPINCj �0.0120 �0.0120 �0.0120 �0.0101

(�8.95) (�8.98) (�9.13) (�5.13)

AVEWAGESj þ0.0195 þ0.0195 þ0.0195 þ0.0123

(11.55) (11.57) (11.79) (7.896)

%CONSTRCTj þ0.162

(6.156)

%CLOTHj þ0.105

(5.283)

%TEXTILESj �0.040

(9.74)

%UNEMPj �0.140

(7.29)

N 783 783 783 783

S 1.031 1.028 1.010 0.869

RSS 823.3 818.5 790.8 582.5

F-Test 201.6 177.8 214.5 209.4

[8,774] [8,774] [7,775] [11,771]

R-sqd 0.645 0.647 0.659 0.749

adj.R-sqd 0.642 0.644 0.656 0.746

Note: RONATINC and RCNATINC are natural rates of population growth for theregion of origin and the recipient city; %AGRIC is the % of the population occupied inagriculture; %LAND > 100HA and %LAND20/100 refer to the distribution of theusable agricultural area by size of holding; AVEWAGES is the average wages of day-labourers in 1892 for each city; AGPROD82 and DAGPROD82=95 are Wgures for thelevels of agricultural productivity 1880/5 and the increase in 1880/5–1893/7. Thesemake use of regional agricultural accounts prepared by the author. Further details on thesources are in Appendix 4A.


important. This theme will be examined in more detail in the next

chapter.

The percentage of the population occupied in agriculture in the

region of origin (%AGRICk) is a variable whose importance is consid-

erably reduced when compared to the results given in the previous

section. But this is not so surprising. The eVect of using rural–urban

moves as the dependent variable is to eliminate much of the previous

eVect of this variable. More urban regions had lower overall net migra-

tion for two main reasons: Wrst, there were proportionately less farmers

and farm-workers, and secondly, there were more moves to cities within

the region. Moves to cities within the region are now recorded; and by

using moves relative to the rural population as the dependent variable

the importance of the Wrst point is also reduced.15

The second column shows the eVect of including the term

AGVALUEk. This gives the value per head of population of Aktienge-sellschaften (joint stock companies) founded 1871–1910 by region. This

was included in an attempt to examine the extent to which migration

was aVected by the prospects of a region as well as its current position.

The %AGOCCPkt term is a backward-looking measure: it reXects what

has happened in the past, not what is likely to happen in the future.

The formation of Aktiengesellschaften is a forward-looking decision by

investors and should be related to the dynamism of the region, to the

extent to which it has specialized in fast-growing industries for

example.

The introduction of the AGVALUEk variable reduces the signiW-

cance of the %AGRICk term (column b). For the last two columns this

term and the NATPOPINCk term are both dropped. The coeYcient on

AGVALUEk is raised as a result. Some of the eVect of the %AGRICk

variable is captured by AGVALUEk (Aktiengesellschaften were more

likely to be formed in areas which were already industrial). But the

greater signiWcance of AGVALUEk is due to its ‘forward-looking’ or

dynamic quality.

What exactly were these dynamic factors which caused AGVALUEk

to have this inXuence? There were three forces at work. The Wrst was

an increase in the importance of large Wrms, which were more likely to

be Aktiengesellschaften. The second was the impact of foreign trade on

the German economy, which led to greater specialization in those areas

15 The fact that some eVect is still exerted by %AGRICk is explained by the fact thatthere are some regions which were industrial but which did not have large cities.


where Germany had a comparative advantage, and in turn led to faster

growth in regions specializing in these Welds. The third was a tendency

for regional specialization to rise, due to economies of scale internal and

external to the Wrm. Increased vertical integration, leading to the com-

bining of processing and primary production in a single region, was one

aspect of this.

In the third column the increase in agricultural productivity by

region between the years 1880–4 and 1893–7 is introduced (DAGPRODk).

16 This variable is found to have a signiWcant positive eVect on

the likelihood of migration: migration is higher from regions with high

productivity increases. This result can be interpreted as supporting the

view that there was a substantial agricultural labour surplus, and the

release of this surplus both increased productivity and led to out-migra-

tion. But an alternative view is that these are both consequences of a

change in institutions and social relations. The impact of industrializa-

tion led to a transformation of agrarian society which both raised prod-

uctivity and led to increased out-migration.

In the last column (d) the breakdown of the employed population in

the recipient city by occupation is used to produce a series of additional

explanatory variables. The results show that certain industries attracted

migrants, leading to an increase in the likelihood of inward migration

from rural areas. Two which had this eVect were construction and

clothing (%CONSTRCTj and %CLOTHj). Other industries had a

negative eVect. One of these was textiles (%TEXTILESj). The main

period of rapid expansion for this industry had passed by 1890, and the

main textile areas had less need of migrants. The results also show that

there is a negative eVect from the percentage of workers who were

without regular employment according to the census (the variable

is called %UNEMPj although it is not the same as registered

unemployment).

16 These Wgures are derived from regional accounts. The methods used are describedin chapter 5 of Grant (2000), and also in Grant (2002a).


5

Demography and Migration

1. Introduction

The last chapter presented some evidence on the inXuence of demo-

graphic variables on migration. The results were equivocal. Regressions

showed an eVect from demographic variables on net migration for

eighty-one regions 1871–1910, but, when migration between four major

regional groupings was considered, the net eVect of demographic factors

was shown to be small when compared with other explanatory variables.

The analysis of the 1890 census showed that variations in the natural

rate of population increase in the recipient cities had an important eVect

on migration, but that the eVect from population growth in the regions

of origin was less certain.

This chapter will attempt to elucidate these issues, in the context of a

general discussion of the relationship between migration and demog-

raphy in Germany 1870–1913. In this period Germany was going

through the ‘demographic transition’, usually deWned as a movement

from a state in which both fertility and mortality are high to one in

which both are considerably lower. This transition has been much stud-

ied by demographers: one conclusion which has emerged is that there is

no single model of the demographic transition—much will depend on

the circumstances of individual countries or regions. The German

experience conWrms this: there were important diVerences between the

regions and these have implications for migration. SpeciWcally, the

demographic experiences of two regions which were major sources of

potential migrants in the period, the south and the east, were very

diVerent. This had a considerable eVect. It is one reason why there was

heavy out-migration from certain heavily agricultural regions in the

east, while migration from similar southern regions was lower.

So, diVerences between the south and the east will be one theme of

the chapter. A second theme will be a comparison of the demographic

background to two periods of heavy out-migration: that which aVectedsouthern and western Germany in the earlier part of the nineteenth

century and the movement out of eastern Germany in the period after

1870. The intention here is to look for signs of demographic stress,

direct evidence of severe population pressure. To what extent was

migration caused by overpopulation? Are there diVerences between the

two periods?

Comparison between the south and the east leads to an examination

of the role played by agricultural factors. The agricultural systems of

the two regions were very distinct: diVerent sized-farms, diVerent typesof labour contract, diVerent inheritance systems. Did these have demo-

graphic eVects?

As regards the demographic experience of the recipient towns and

cities, this has already been studied in some depth. The high fertility of

the coal-mining regions in the west is a well-known fact.1 This has clear

implications for migration. It is a fairly simple calculation to show that

high fertility reduced the need for migration into the coal-mining areas.

A higher proportion of the required increase in the total labour force

was supplied through the natural increase of the resident population

than was the case for Berlin, for example. This was an important factor

in migration.2

A more general theme is the contribution of demographic movements

to the ‘labour surplus’ phase of the Lewis Model. The period of the

‘demographic transition’, when mortality rates fall ahead of a reduction

in fertility, is potentially one of high population growth, contributing to

the overall labour surplus.

2. The German Demographic Experience in the

N ineteenth Century

An overview of German population movements in the nineteenth cen-

tury is provided in Figure 5.1. This uses calculations made by the

Prussian statistical oYce for all the territories included in Prussia after

the incorporation of Hanover and Schleswig-Holstein in 1866. This

series extends back to 1816. A second series is superimposed, which is

for all the post-1871 German territories, and makes use of a series

1 Wrigley (1961); see also Haines (1977) and (1979).2 Valuable studies of demographic change in Germany in the period include Ballod

(1914), Hohorst (1977), Kollmann (1974), Knodel (1988), Marschalck (1987), Lee(1979a), and Pollard (1980). This chapter also draws on the theoretical analysis of Birdsell(1988) and Habbakuk (1958). The regional classiWcation follows the deWnitions set out inAppendix Table 3A.1.

Demography and Migration 125

produced by the Imperial Statistical OYce. This only goes back to

1862.3 There is a fairly close correspondence between the two series,

which is as expected, given that Prussia accounted for around 61% of

the total German population.

The reliability of the earlier Wgures is open to question. The main

area of concern is the Wgures for stillbirths prior to 1841. These had to

be estimated as they were not then recorded. If these are underesti-

mated then the birth rate and the death rate are aVected, but not the

overall rate of population growth.

The natural rate of population increase peaked in the 1816–22

period, possibly as a reaction to the ending of the disruptions caused by

3 Calculated from SJDR (1915), 22, SJPS (1910), 14.

18160

100

200

300

400

500

Prussian births

Prussian deaths

Germanbirths

Germandeaths

Prussian natural

1826 1836 1846 1856 1866 1876 1886 1896 1906

increase

German naturalincrease

Fig. 5.1. Births, deaths, and the natural rate of population increase, per 10,000

head of population, Prussia and Germany 1816–1913 (post-1871 territorial

units)

Note: Figures are three-year rolling averages.Source: The source used is SJPS (1915), 55; these estimates were endorsed by Kollmann(1980), who also gives figures for the individual provinces.

126 Demography and Migration

the Napoleonic wars, but then fell back sharply in the agricultural crisis

which followed, from 1.9% in 1822 to just 0.2% in 1831. This was

brought about by a combination of a rise in the death rate and a fall in

the birth rate. Thereafter there was an irregular recovery, which peaked

in the mid-1870s, when the natural rate of increase reached 1.5% per

annum. This period saw considerable Xuctuations in both fertility and

mortality, which are apparent even when a three-year moving average is

used. After the 1870s both series are much more stable, and tend to

decline, with the decline in the death rate being more pronounced and

starting much earlier. The birth rate does not show a substantial decline

until after 1900, so that the rate of natural increase remained high,

peaking at 1.56% per annum (reached in both 1898 and 1902).4

The decline of fertility after 1900 was too late to aVect the pre-1913

labour market. The combined eVects of high fertility in previous

periods and declining mortality meant that the rate of increase of the

potential labour force remained high right up to 1913. Comparison of

the employment censuses of 1895 and 1907 shows the male labour force

increasing at a rate of 1.5% per annum and an even faster rate for

female employment.5

The pattern is a classic demographic transition curve. But the peak

was reached at diVerent times in diVerent regions. Figure 5.2 illustrates

this point, classifying each region by the timing of the peak rate of

natural increase between census years (normally at Wve-year intervals).Most regions peaked in either 1895–1900 or 1900–5, as did the

national average. But there was a group which peaked much earlier, in

the 1870s. This included Berlin, Pomerania, and some other central

regions. There is another group in which population growth was still

rising by 1905–10, which had not, therefore, clearly peaked before the

First World War. These are the regions shaded black. This is quite a

scattered group of regions, with all areas represented apart from the

centre. There is no clear pattern; it includes agricultural as well as

industrial regions.

Figure 5.2 helps to explain why there was relatively little eVect from

demographic factors when the pattern of migration between the major

regional groupings was considered in the last chapter. There were

4 German Wgures, SJDR (1912), 22.5 Data from Kaelble and Hohls (1989). There are some changes in the relative positions

of the various regions as a result of diVerent rates of industrial growth between censusdates, but the overall ranking remains remarkably constant. The classiWcation would beessentially the same if the 1882 or 1907 censuses were used in place of the 1895 census.


diVerences within these groupings which had a tendency to cancel each

other out. Population growth falling in Pomerania was balanced by

rising population growth in other eastern regions, for example.

Figures 5.3 and 5.4 demonstrate the eVect of these movements on the

overall rate of population growth by region, comparing the 1870s with

the 1900–10 period. The Wrst map has a ‘mottled’ appearance: most

parts of the country contained regions both of high and of low growth.

Again, this helps to explain why demography had a limited eVect on the

major groupings.

By 1900–10 there is a clearer pattern: the areas of high growth are

predominantly in the west and in the east; the centre and the south

contain the low-growth regions. In most areas the rate of increase was

faster in the 1900–10 period. The main exceptions are in the centre,

conWrming that this was the region which Wrst went into the downward

phase of the demographic transition.

1895−1900

1890−5

1900−5

1905−10

1875−80

1871−5

1880−5

1885−90

Fig. 5.2. The demographic transition in Germany 1871–1910: German

regions classified by the period of highest natural population increase

Source: Data for this figure, and the next two, taken from census publications, SDR(1876), a.f. 20, SDR (1881), a.f. 57, SDR (1886), n.f. 32, SDR (1891), n.f. 68, SDR (1901),n.f. 150–1; VSDR 1897 iv, 1906 iv, and 1911 iv.


From the viewpoint of a study of migration, the most important

map is Figure 5.3. This is because diVerences in population growth in

1900–10, in so far as these reXect diVerences in birth rates, are not

particularly relevant to migration Xows before 1914. A rise in the

number of children will not increase migration immediately; in fact it

could reduce migration if families with children are less likely to mi-

grate. It is only when these children enter the labour market, and take

decisions based on an assessment of their own prospects, that there is

likely to be an increase in out-migration from a region of high popula-

tion growth.

Analysis of the Berlin Wgures from the 1885 census supports these

points. The median age of migrants arrived in the previous year was

22.8; 52.2% of migrants were aged between 17 and 26.6 These Wgures

also show that those aged between 20 and 21 had the highest probability

of migration.

6 Calculated from Berlin, Statistisches Amt (1887), table Vb.

1.6−1.8

1.4−1.6

>1.81.2−1.4

0.8−1.0

<0.8

1.0−1.2

Fig. 5.3. The natural rate of population increase 1871–80 (annual averages)



So, it is the earlier, ‘mottled’ map of 1871–80 which has more sign-

iWcance. The more pronounced regional pattern of 1900–10 was of little

consequence for migration before 1914.

3. A Comparison of Demographic Movements

in the East and the South

The discussion in the previous section implicitly treated demographic

factors as exogenous: as a pattern of behaviour which was unaVected by

economic circumstances and prospects. But there are ways in which

economic conditions can have eVects on population growth, and, once

this is accepted, the main point about the interaction between demo-

graphic factors and migration becomes a rather diVerent one. It is the

extent to which the natural rate of increase responds, or fails to respond,

to diVerences in relative economic opportunities between regions that

will aVect migration.

1.6−1.8

1.4−1.6

>1.81.2−1.4

0.8−1.0

<0.8

1.0−1.2

Fig. 5.4. The natural rate of population increase 1900–10 (annual averages)



Contemporary observers were well aware of mechanisms whereby

economic conditions could aVect both mortality and fertility. In 1879

von Fircks pointed to the coincidence of periods of rising mortality and

high cereal prices.7 German demographers also recognized the import-

ance of delayed marriage—the ‘Western European marriage pattern’—

for the control of fertility. Indeed, they were particularly conscious of

this, as Germany was very much on the frontier of this division: the

area of delayed marriage in central and eastern Europe followed the line

of German inXuence.8 In the German view, delayed marriage was a

consequence of cultural progress, and this was the reason why Ger-

many, and other parts of Western Europe, diVered in this respect from

lands further east: ‘In primitive tribes or in countries with a low level of

cultural development, such as Russia, Serbia or Bulgaria, the average

age at marriage is very low; marriage takes place shortly after puberty

for both sexes.’9

The same study argued that in Germany marriage was linked to

economic conditions. The custom which delayed marriage until the

couple were able to set up as an independent family unit meant that in

favourable economic circumstances this would be attained earlier, while

less good conditions would hold it back.10 Statistical analysis of mar-

riage rates in Prussian provinces, 1816–75, supports the view that eco-

nomic circumstances did inXuence marriage rates. High prices for foods

consumed by the rural labourers and their families, principally rye and

potatoes, were associated with lower marriage rates. However, high

prices for goods which were produced in rural areas but mainly con-

sumed by the better-oV or by city-dwellers, such as wheat and beef,

were linked to higher rates. Marriage was put oV when times were bad

and brought forward when times were good. The full analysis is given

in Appendix 5E.11

With such a relationship, when there was out-migration caused by a

rate of population growth which was excessive relative to available eco-

nomic opportunities, this would set in motion demographic changes

which would eventually reduce population growth in the regions of

7 Von Fircks (1879), 48.8 Including, for example, the future Baltic states, as can be seen from the map at the

end of Coale and Watkins (1986).9 Nadobnik (1908), 69.

10 Ibid. 80.11 Hohorst (1977) also found a relationship between marriage rates and economic con-

ditions for linen workers.


out-migration. A crisis of overpopulation would cause a deterioration of

employment prospects. This would hold back marriage rates and act as

a stabilizing mechanism. The question then becomes: why did this

mechanism work in some regions and not in others?

The importance of this point is illustrated by the results of a classiW-

cation of German regions given in Appendix 5A and summarized in

Table 5.1. The data set is the one used in the previous chapter: net

migration is calculated using the indirect method of census comparison,

and the period of analysis is 1871–1910. The regions were divided into

four equal groups according to the percentage of the population occu-

pied in agriculture (1895 occupational census). These groups were then

subdivided into two according to an estimate of predicted net migration

resulting from population growth. This made use of the statistical an-

alysis reported in Appendix 4B (Table 4B.2).12 As before, the birth rate

is lagged four periods, or twenty years.

The classiWcation shows that there were considerable diVerences be-

tween the demographic records of the diVerent regions. In the high

population growth sub-group of the regions classiWed as highly indus-

trial, the diVerence between the lagged birth rate and the current death

rate was 1.66% per annum; for the low population growth regions it

was 1.29%. The predicted eVect from these demographic diVerences

amounts to a net diVerence of 137 persons per 10,000 resident popula-

tion (108 plus 29). The actual diVerence was 227 (565 minus 238). So,these demographic factors accounted for 42% of the observed diVerence

in net migration rates between the two sub-groups. At the other end, in

the highly agricultural group, the demographic eVect accounts for 48%

of the observed diVerence in net migration between the high and low

population growth groups.13 Demographic factors played a large part in

the determination of migration patterns.

Looking at individual regions, there are some major diVerences

between regions which are not so far apart geographically. In the highly

industrial group there are two comparisons of this type which stand

out: in the west, the Aachen region has one of the lowest rates of

population growth, the Arnsberg region one of the highest; in Saxony

12 The equation used in these calculations was the same as that used forTable 4.5: NETMIGjt ¼ 0:44 NETMIGjt�1 þ 0:26 DEATHRTjt � 0:22 BIRTHRTjt�4.The regions were divided into eight sub-groups of ten; the extra region is in the ‘moder-ately agricultural–low population pressure’ sub-group.

13 For the ‘moderately industrial’ group, however, the predicted eVect is much greaterthan the observed diVerence.


the Zwickau region has a high rate of natural increase, while the Baut-

zen region is much lower.

In the ‘highly agricultural’ group there is a clear geographical div-

ision. Six of the ten ‘high population growth’ regions are in the east,

including the Wve highest; eight of the ten ‘low population growth’

Table 5.1. A classiWcation of 81 German regions according to the

availability of non-agricultural employment and population growth,

1871–1910, unweighted averages (migration and demographic variables

are per 10,000 head of population)

Regions classiWed as

Highly

industrial

Moderately

industrial

Moderately

agricultural

Highly

agricultural

All

regions

A. Regions with low population growth

% occupied in

agriculture

25.2 37.8 45.6 55.3 41.0

Average birth

rate (T-4)

3591 3588 3683 3496 3590

Average death

rate (T)

2305 2433 2555 2447 2435

Predicted

demographic

eVect

þ29 þ82 þ104 þ122 þ84

Actual net

migration

þ238 �206 �344 �469 �195

B. Regions with high population growth

% occupied in

agriculture

16.7 35.6 45.8 58.1 39.1

Average birth

rate (T-4)

4079 3725 3888 4050 3936

Average death

rate (T)

2414 2217 2386 2412 2357

Predicted

demographic

eVect

�108 �72 �63 �94 �84

Actual net

migration

þ565 �246 �567 �921 �292

Source: ClassiWcation of regions uses 1895 occupational census, Kaelble and Hohls (1989).


regions are in the south, including the Wve lowest. The basic point is

clear: in the south population growth was adjusted downwards in

regions which had limited opportunities for non-agricultural employ-

ment; in the east this did not occur. Consequently there was more

migration out of the eastern regions.

The mechanism which achieved this was marriage restriction.

Figures 5.5 and 5.6 make this point. The Wrst shows the overall recordof population growth for the south and the east. It combines Wgures

1818

0

100

200

300

400

500

600

1832 1847 1862 1877 1892 1907

Deaths

Births

South : births

Natural increase

South : deaths

East : births

East : deaths

South : naturalincrease

East : naturalincrease

Fig. 5.5. Births, deaths, and the rate of natural increase, five-year average

annual rates, 1816–1910 (per 1,000 head of population)

Source: Calculated using data from Kollmann (1980) for the pre-1871 figures; the post-1871 figures come from the census publications. For this comparison Mecklenburg-Strelitz and Alsace-Lorraine were omitted from the regional groupings, thus ‘south’ isBavaria, Wurttemberg, and Baden.


given by Kollmann for the pre-1871 period with Wgures produced by

the Imperial Statistical OYce for the period after uniWcation. Kollmann

has estimated stillbirths for periods when these are not available, so

there are reasons to treat the pre-1870 Wgures with some caution. How-

ever, the basic conclusions should not be seriously aVected. These are,

Wrst, that the natural rate of increase was fairly consistently higher in

the east, and secondly that this owed more to fertility than to mortality.

Mortality Xuctuated up to 1880 (rather more in the east than in the

1818

0.0

0.1

0.2

0.3

0.4

0.5

0.6

1832 1847 1862 1877 1892 1907

East: Im index

East: marriages

South: marriages

South: Im index

Fig. 5.6. Crude marriage rates and the Princeton Im index, 1816–1901, five-

year average annual rates

Note: Marriage rate is number of marriages per 100 head of population.Source: Data from Knodel (1974), Appendix table 2.1, and Kollmann (1980).


south), and thereafter fell in both regions. Fertility was generally higher

in the east, apart from the 1871–85 period, when it was the same in

both areas.14

Figure 5.6 compares the marriage Wgures for the two regions, and

includes average values of the Im index used by the Princeton study of

population growth.15 This is a weighted index of the proportion of

women of diVerent ages who are married, the weights reXecting changes

in the natural fertility of women in diVerent age groups. It should

therefore represent an unbiased estimate of the potential eVect of mar-

riage on birth rates. However, the limitations of the available census

data meant that it could not be calculated before 1867. The crude

marriage rate, which is available, is a much less valuable measure. As it

includes remarriages, it is aVected by the death rate: higher death rates

lead to more remarriages of those who have lost partners. The same

problem applies to the average age of those getting married, if this does

not distinguish remarriages (which is the case with the Prussian

Wgures).

Figure 5.6 shows that there is a relationship between the crude mar-

riage rate and the Im index. The low marriage rate in southern Ger-

many before the 1860s clearly aVected the Wrst Im index. Even though

the southern marriage rate was rising in the 1860s this still left the

south well below the east. The southern Im index was 79% of the

eastern Wgure. A higher southern marriage rate in the years 1860–75meant that the gap narrowed, and this is the period in which the south-

ern birth rate rose to the eastern level. But in the late 1870s the south-

ern marriage rate fell again, and both the Im indices and the birth rates

moved apart.

The low birth rate of southern Germany owed something to legal

restrictions on marriage in Bavaria, but this was not the only factor.

While these contributed to the extremely low Bavarian Im index of

0.380 in 1867, the indices for Wurttemberg and Baden, where restric-

tions were lifted earlier, were also low (0.403 and 0.400 respectively). In

addition, while the lifting of the Bavarian restrictions in 1868 certainly

contributed to the high marriage rates in the following years, in the less

favourable economic circumstances of the late 1870s there was a return

to low marriage rates, which contributed to the fall in the southern Im

14 The reasons for high eastern fertility are examined in Section 3.15 All the Princeton indices for Germany are taken from the appendices in Knodel

(1974).


index relative to the east. It seems therefore that the legal restrictions

reinforced a tendency to low marriage rates which had other causes as

well.16

Turning from the incidence of marriage to the level of fertility within

marriage (the Princeton Ig index), there is no sizeable diVerence be-

tween the east and the south on this measure for the period 1867–1910.

The southern average was slightly lower, 0.741 against 0.745, but thiswas balanced by a larger number of births outside marriage.

Statistical analysis of the percentage unmarried did not produce a

clear relationship between this variable and economic circumstances.

The extent to which marriage was sensitive to economic conditions

varied considerably between regions, especially once the non-Prussian

regions were included in the analysis. One reason for this may be that

some areas were restricting fertility within marriage. This was the con-

clusion reached by Knodel in his study of village fertility patterns. If so,

then these areas would have had less reason to put oV marriage in times

of economic hardship.17

Another statistical analysis which did not produce results was an

attempt to Wnd causation running from migration to marriage patterns.

If migration aVected the sex-ratio in regions of out-migration then it

might also aVect the likelihood of marriage for women in these regions.

The lack of any clear relationship between migration and the marriage

prospects of women is demonstrated by Figure 5.7 which gives the

proportion of women aged 49–54 who had never been married, by

region, from the 1880 census.

Figure 5.7 shows that there was no consistent relationship between

migration and the likelihood of marriage. Marienwerder, a region of

high out-migration, had one of the lowest proportions of never-married

women. In general the regions of out-migration in the east had no

tendency to have more never-married women than the recipient regions

of the centre. Some of the reasons for this can be understood from the

discussion in the previous chapter. Women did migrate, even if they

16 Works dealing with the eVect of the marriage restrictions in Southern Germany andAustria include Matz (1980), Ehmer (1991), and Mantl (1997). The eVect of these restric-tions may have been to increase illegitimacy rather than reducing the overall birth rate. Astudy of 5,263 applications in Wurttemberg during the restricted period showed that 6%were refused, but these couples may have applied successfully later, Matz (1980), 267.

17 Knodel (1988). This study of demographic behaviour in 14 villages shows consider-able divergence between the diVerent villages, but the sample is too small, and theregional coverage is too restricted, to draw general conclusions about the reasons for thesediVerences.


made fewer moves than men.18 So women were quite capable of leaving

regions where prospects were poor. Moreover, some of the men who

made temporary moves out of their region of birth in their early twen-

ties returned later to be married in that region. Thus out-migration did

not inevitably lead to a decrease in the likelihood of marriage.19

But in the south and the south-west the position was a diVerent one.

Here there are extremely high proportions of never-married women.

The Bavarian Wgures are almost certainly aVected by the legal restric-

tions in force up to 1868, but high Wgures are also found in Baden,

18 Women also made temporary moves, Wnding employment as domestic servants, Leeand Marschalck (2000), 379, but men were more mobile in general, Hochstadt (1999),93–4.

19 Analysis of Kreis-level results for 1895 did indicate that there was an eVect frommigration on marriage rates inside Prussia by that date: see the note to Table 5D.2 inAppendix 5D.

12−14%

6−8%

<6%

8−10%

10−12%

14−16%

16−18%

18−20%

>20%

Fig. 5.7. Percentage of women aged 49–54 who had never been married,

by region, 1880 census

Source: Calculated from SDR (1881), a.f. 52, pp. 144–87.


Wurttemberg, and Alsace-Lorraine.20 Figures in some western regions

are also above average. These include many of the areas where there

was high out-migration in the Wrst half of the nineteenth century.

This leads to a more general point. Looking at the history of migra-

tion in Germany over the whole nineteenth century it is possible to

identify two distinct movements of large-scale out-migration. The Wrst

was centred on south-western Germany, began in the 1830s, and

peaked in the 1850s. The second was centred on eastern Germany,

began in the 1860s, and peaked in the 1890s.

The Wrst was accompanied by clear signs of demographic distress.

The death rate was high and the marriage rate fell. Partly in conse-

quence, there was a rise in the number of illegitimate births. Kollmann

has referred to this as a period of ‘Ubervolkerungskatastrophe’. 21 In a

process not untypical of pre-industrial Western Europe, an excess

supply of labour relative to economic opportunities was eliminated,

partly by out-migration and partly by reduced marriages. The excess

supply of labour was most probably caused by falling mortality from the

late eighteenth century onwards.22

The second period has none of these characteristics. The death rate

was falling in all areas from the 1870s onwards, including the areas of

out-migration. Marriage rates were high and tending to rise. Illegitim-

ate births fell, and were below the national average in the east.

This is not to argue that all was well in eastern Germany. There is

evidence of poor conditions in an earlier period from a diVerent source,

which is the height of recruits, generally taken as a good indicator of the

nutritional standards of children and young adults.23 Figure 5.8 gives

the proportion of recruits rejected for being less than 5 feet tall, in the

20 Marriage restrictions may also account for the relatively high Wgure for Mecklen-burg, Lubinski (1995); for Bavaria see Lee (1977). The period after the ending ofthe marriage restrictions saw a gradual convergence of marriage rates in the south andsouth-west with those in other parts of the country, but this could also be explained bythe improvement in economic circumstances in these regions after 1870.

21 Kollmann (1976), 15. The reasons for this ‘overpopulation catastrophe’ havebeen debated: see inter alia Lee (1979 and 1980), Marschalck (1987), Kollmann (1974).One important question is the impact of agrarian reforms introduced in the earlynineteenth century. But it is quite possible to see it as a consequence of a period ofrelatively high population growth, set oV by declining mortality in the eighteenth century.

22 SchoWeld et al. (1991), Lee (1980), and Marschalck (1982) discuss this point.23 See, inter alia, Komlos (1995) and Baten (1999). There are other possible reasons for

the low stature of eastern recruits, such as the eVect of diseases, Leunig and Voth (1996),but in the absence of clear evidence of regional diVerences in disease epidemiology, it isreasonable to assume that the main cause of the variation in stature was nutritional.


old Prussian provinces (post-1815 boundaries), taken from a survey in

the 1860s. This would obviously reXect conditions in the 1840s when

these recruits were born.

The range is a large one: fewer than 2% of possible recruits were less

than 5 feet tall in Berlin and Munster; in Marienberg 16.8% were, and

in Posen the Wgure was 22.8%.24 This is a period when the east was

attracting migrants from other parts of Germany. The map shows that

conditions in the eastern Prussian provinces in these years were harsh

and unappealing. It conWrms the view that only acute demographic

pressure in other parts of Germany could have led migrants to move to

these remote areas where living conditions were precarious and food

24 The Wgures are much higher for the Polish areas, which might reXect a disposition toreject Polish recruits, although putting down incorrect heights would have been an oddway to do this when there were other, less easily veriWable, reasons which could have beengiven (such as ‘bodily weakness’).

120−40

100−20

140−60

160−80

20−40

<20

40−60

60−80

>18080−100

Fig. 5.8. Numbers of possible recruits in the 1860s rejected for being less

than 5 feet tall, per 1,000, by Regierungsbezirke

Note: The unit of measurement is the old Prussian Fu�.Source: Meitzen (1868), 322–3.


supplies often inadequate. What is striking is that, in the early part of

the century, migrants moved from relatively rich provinces to the

poorer ones in the east, but after 1860 they moved from the poorer east

to the richer parts in the centre and the west.25

There are, therefore, reasons to see the two migration movements as

representing diVerent types of labour surplus: the Wrst a pre-industrial

labour surplus, driven by demographic conditions, the second, a labour

surplus of the industrial period, not primarily caused by population

growth, but rather by changes in the relative demand and supply of

labour between sectors. Rising industrial demand for labour was sup-

plied by the release of surplus labour from agriculture.

4. Population Growth in the Regions of Origin

The argument has been made that the agrarian reforms of the early

nineteenth century created the conditions for a large increase in popula-

tion, and that this provided a stimulus to industrial growth. This is

sometimes referred to as the ‘Ipsen–Kollmann thesis’ after the two

authors who are most closely associated with this view. The eVect of

the land reforms has, however, been disputed. Some demographers,

including Marschalck and Lee, have argued that the intensiWcation of

work which, they argue, was brought about by the reforms contributed

to a worsening of child care and a consequent rise in infant mortality.26

There is a problem with the chronology of the Ipsen–Kollmann story

which is apparent from Figures 5.1 and 5.5 in the previous section: the

peak of German population growth in the years around 1900 is far

removed from the period of the land reforms. Rising population growth

after 1870 looks like a response to industrialization, not a cause. It can

be argued that the surge in marriages and in births in 1816–22 owed

something to emancipation (though the ending of the Napoleonic wars

is also a possible explanation). This, in turn, meant that there was a

ready supply of labour for aspirant industrialists, as well as providing

25 Kollmann’s Wgures (calculated on a three-year basis using census returns) show thatthere were substantial gains in East and West Prussia, Silesia, and Posen through migra-tion between 1816 and 1846. It is not possible to say exactly where these migrants camefrom, but there was out-migration from Brandenburg in 1816–31, Prussian Saxony in1820–31, and Westfalia in 1831–7 and again in 1840–6. Kollmann (1980).

26 Kollmann (1976), Lee (1979a), Marschalck (1976); see also Harnisch (1977).


the labour required for the expansion of the area under cultivation in

eastern Germany in the Wrst half of the century.

The emphasis placed on emancipation has behind it a ‘top–down’

view of population growth: peasants and rural labourers will marry

early, and have more children, unless laws or other restrictions imposed

from above prevent them from doing so. But this underestimates the

ability of rural society in Western Europe to regulate population growth

in its own interests. Knodel’s 1988 study of fourteen German villages

showed that some were restricting fertility within marriage as early as

1800, while the average age of women at Wrst marriage was raised from

25.5 in 1700–49 to 26.9 in 1825–49.27

This was not exactly a benign process. The burden fell heavily on

women, particularly if reduced marriage opportunities led to an increase

in illegitimate births. The high mortality of illegitimate children (84%

higher than for those born within marriages in Prussia 1875–1900) iseloquent testimony to the social pressures placed on unmarried

mothers.28

If rural society was able to exercise a crude but eVective control over

population increase, then it is possible to interpret the increased birth

rate which followed the agrarian reforms as a response to a perceived

improvement in economic opportunities. But for some these hopes were

disappointed. In the 1830s bad harvests and high prices brought about a

worsening of the condition of the landless agricultural labourer. How-

ever, from 1860 onwards rising non-agricultural demand for labour

brought about a recovery, both in real wages and in births.29

Agricultural institutions have consequences for the fertility decisions

of households. One example is the contractual labour relationship typ-

ical on large estates in the east. The Inste or Deputate received a small-

holding from the landlord (from which they derived much of their

income) in return for an obligation to provide labour at harvest time.

This obligation was not just to provide the labour of a single person,

but of additional Scharwerker, who could be family members or Knechte

or Magde (other labourers living in the Inste household). For a house-

hold living in this system it was valuable to have children who could

27 Knodel (1988), 125.28 Calculated from Wgures in SJPS (1915), 55. The Wgures show that on average 55.5%

of illegitimate children died in their Wrst year, while 19.5% of those born within marriagesdied. This was not just a feature of rural society; the diVerence was just as large for thoseborn in cities.

29 The role of economic factors in population growth is discussed in Birdsell (1988).


provide Scharwerker labour; it was also necessary that, if the family

were to retain the smallholding when the parents were too old or inWrm

to work during the harvest, there should be children willing and able to

perform these duties.30

A second institution which may aVect fertility decisions is partible

inheritance. There are reasons to expect this to aVect marriage decisions

in particular. Studies show that landless labourers married later, so any

institution which spread land ownership more widely should lead to

earlier marriage and consequently higher fertility. But there are factors

which might work against this. If the main concern of parents when

making fertility decisions was the prospect of support from their chil-

dren in old age, then there was a risk that, with partible inheritance,

large family sizes could lead to a position where none of the children had

holdings large enough to provide eVective support to their parents. Ex-

cessive subdivision of holdings would not be in anyone’s best interest.31

To examine the relationship between these institutional factors and

fertility a data set was prepared which was based on the 1895 census

results for the Prussian Kreise. Limiting the analysis to Prussian terri-

tories had the advantage that the legal framework would be the same for

all areas. It also meant that the results of the 1892 and 1901 surveys of

Ortsubliche Tagelohne could be used for rural wages. Finally, Wgures for

30 Weber (1895), 185; see also Harnisch (1984) for a diVerent perspective on tenurialrelations in eastern Germany, but which also stresses the importance of the post-reformsystem as a source of labour for industrialization.

31 See Berkner and Mendels (1980) for a survey of the literature on this point.

Table 5.2. Summary of Wndings from statistical analysis of census

results for 424 rural Prussian Kreise 1895: analysis of total fertility

Factors which increase total fertility (numbers of children under 1 year relative

to the total number of women of child-bearing age)


A large Polish-speaking proportion of the population


High levels of rural wages

Low levels of partible inheritance


A high level of Catholicism

A large Lithuanian-speaking proportion of the population


partible inheritance were available from an 1896–1902 Prussian survey

which recorded what was actually happening to property which

changed hands as a result of inheritance. This is preferable to the use of

a classiWcation of regions according to an assessment of the predominant

system which does not record actual practice, since practice may diVer

from custom. It is the actual splitting of holdings which is likely to have

the strongest eVect on demographic behaviour.32 However, it does

mean that the main areas of partible inheritance in the south and south-

west are excluded from the analysis.

The econometric analysis of this data set is reported in Appendix 5D.

The results are summarized in Tables 5.2, 5.3, and 5.4. The Wrst of

these shows the results of the analysis of total fertility: the numbers of

children aged under 1 year relative to the total number of women aged

15–50. The main Wndings are that total fertility is highest in areas with

a high Polish-speaking population, in areas where wages are relatively

high and in areas where there were large farms and estates. In the rural

east, the presence of large numbers of Poles, and of larger estates,

would tend to raise fertility, but lower wages would have the opposite

eVect. These results support the view that the institutions of east-

Elbian estates, in particular the Scharwerker system, did raise fertility,

and thus lead to out-migration.

Partible inheritance, on the other hand, does not appear to have led

to an increase in fertility; on the contrary, fertility was lower in the


results for 424 rural Prussian Kreise 1895: analysis of marriage rates

Factors which increase the proportion of women aged 15–50 who are married


A low level of Catholicism



The proximity of cities (over 50,000 inhabitants)


High levels of partible inheritance

32 The survey was published in ZdKPSB (1905); studies of inheritance systems includeMiaskowski (1882) and Sering (1899–1910). In the partial inheritance areas of southernGermany the problem of excessive subdivision aVected the attitudes of parents to lateroVspring, and may have contributed to high infant mortality. Lee (1981), 33 and 96.


Prussian areas which practised divided inheritance. If partible inherit-

ance led to earlier marriage, then this did not feed through into higher

birth rates. The analysis of marriage rates (summarized in Table 5.3)shows that partible inheritance did tend to be associated with a higher

marriage rate, though the result is not a particularly secure one, and the

size of the eVect is not large. The analysis also shows that Catholic areas

tended to have low marriage rates, outside the Polish-speaking areas.

This indicates that marriage restriction may not have been the only

reason for low marriage rates in Catholic areas in the south and south-

west: Catholics inside Prussia also had lower marriage rates.

Marriage rates were higher in areas with good economic prospects: in

Kreise with high wages and in areas close to major cities which would have

beneWted from better access to urban markets and improved employment

opportunities. However, the proximity of major cities did not lead to a

rise in overall fertility. If the measure of fertility is altered to reXect the

impact of variations in the marriage rate—dividing the number of chil-

dren under 1 year by the number of women aged 15–50 who are

married—then the statistical analysis shows that fertility on this measure

was lower in areas close to major cities and higher in remoter areas (Table

5.4). This Wgure cannot be regarded purely as a measure of fertility within

marriage since it will include children born outside marriage as well (these

cannot be distinguished from the census results as published). However, it


results for 424 rural Prussian Kreise 1895: analysis of fertility relative to

the number of married women

Factors which increase fertility measured by the number of children under

1 year relative to the number of married women of child-bearing age



A high level of Catholicism



Low levels of partible inheritance

Large distances from cities (over 50,000 inhabitants)




will reXect changes in marital behaviour. The Wnding suggests that

married couples in areas close to cities were restricting fertility and this

was cancelling the eVect of higher marriage rates

Similarly, in areas of partible inheritance, the tendency for marriage

rates to be higher was cancelled by low rates of fertility relative to

married women. The fear of excessive subdivision may have led married

couples in these areas to restrict family size. Outside the Polish area,

Catholic families followed the opposite strategy: high rates of fertility

relative to married women were countered by low marriage rates.

There were therefore three factors which had strong eVects on demo-

graphic behaviour, but which had less eVect on overall fertility, and

hence on overall population growth, largely because of cancelling eVects

from marriage rates on the one hand, and reproductive behaviour

within marriage on the other. These factors were Catholicism, the prox-

imity of large cities, and partible inheritance. This leaves three other

factors which did have substantial eVects on overall fertility: high

wages, the presence of large estates, and the Polish-speaking percentage.

In these cases, the eVect on marriage rates and the eVect on fertility

relative to married women both ran in the same direction.

The impact of this on overall population growth is illustrated in

Table 5.5, which gives predicted values for diVerent areas, distinguish-

ing between the implied rate of fertility in areas dominated by larger

estates and that for areas where smaller holdings predominated.33

33 These are illustrative Wgures, which assume that the behaviour of the non-agricultural population in Kreise dominated by larger estates followed the example set by

Table 5.5. Predicted values for total fertility (numbers of

children aged under 1 year per 1,000 women of child-bearing age)

On large farms

and estates

(over 100 hectares)

On other holdings

(under 100 hectares)

Average values 66.0 57.9

Polish areas 79.6 71.5

Non-Polish areas 64.3 56.2

Daily wages of 1.51 Marks 68.4 60.3

Daily wages of 1.08 Marks 63.3 55.2

Note: The predictions use the coeYcients given in column a of Table 5D.1.


Outside the Polish areas, fertility in the rural east was held back by low

wages: the predicted level for a larger estate paying the average wage

east of the Elbe (1.08 Marks) is only slightly higher than that for a

smaller holding in the western area where wages reached the average

for Kreise west of the Elbe (1.51 Marks). However, in the Polish areas

population growth was much higher, despite low wages.

The previous section raised the question why population growth was

high in the east even in regions which had few opportunities for non-

agricultural employment, creating a labour surplus which had to

migrate to Wnd work. A contrast was drawn between the east and simi-

lar regions in the south, where population growth was restricted.

The analysis presented in this section provides answers to this ques-

tion. It shows that there was a tendency for total fertility to fall in

regions where wages were low, but the operation of this adjustment

mechanism in the east was limited by other structural factors:

1. A higher proportion of women were married in the east; the east

does not appear to have practised marriage restriction as a means

of population control to the same extent as the German-speaking

Catholic areas of southern and western Germany, or the Catholic

areas inside Prussia.

2. The Catholic population of the east was predominantly Polish-

speaking; this minority had high fertility.

3. There was a tendency for fertility to be higher in rural areas

where the farm structure was dominated by large farms and

estates; this was much more common in the east.

4. Partible inheritance was rare in the east; this was a factor which

tended to reduce overall fertility.34

But the demographic experience of the east can be seen in a wider

context. Both partible inheritance and the presence of large farms and

estates are related to the overall distribution of property and income. In

practice, inequality in the distribution of income in rural areas is closely

the estate labourers. This may, however, have been the case. If the presence of largeestates led to a change in demographic behaviour only on these estates, then an interactionterm, combining the percentage of land in larger holdings and the percentage of the totalpopulation occupied in agriculture should have a signiWcant and positive eVect on overallfertility. A variable of this type was tried, but it had no signiWcant eVect.

34 The evidence suggests that partible inheritance areas were quicker to move to someform of fertility restriction within marriage in the late nineteenth century. In the earlierpart of the century, this pattern of behaviour was not widespread in rural areas, so that thedemographic problems of the south and south-west led instead to a policy of fertilitycontrol by marriage limitation (which was encouraged by legislation).


linked to inequality in the distribution of land. And, over time, partible

inheritance should tend to reduce inequality in property distribution.

Modern demographic studies of developing countries have shown a link

between high inequality and population growth. The Wndings reported

in this section point in the same direction.35

5. M igration and Population Growth in Rural Areas

The eVect of population growth on migration can be studied using a

data set for the Prussian Kreise prepared by the Prussian Statistical

Bureau in the interwar years and used in an article by Peter Quante,

published in 1933.36 This made estimates of the rate of migration out of

the population occupied in agriculture by applying to the agricultural

population rates of fertility and mortality recorded for the population

residing in rural areas. These rates of natural increase were then used to

derive estimates of net migration, using the indirect method (subtract-

ing the estimated rate of natural increase from the actual increase as

recorded by the occupational censuses).37

There are some problems with this approach. The population occu-

pied in agriculture was a large component of the total population resi-

dent in rural areas but there were others, occupied outside agriculture,

who also lived in rural areas. If rates of mortality or fertility for this

rural, non-agricultural, population were to diVer markedly from those

for the agricultural population, there would be a bias in the estimates.38

The second problem is that, although the data set is described as one

which gives migration out of agriculture (landwirtschaftliche Abwander-

ung) it is not in fact one which necessarily records migration. Someone

who switched from agricultural employment to a non-agricultural occu-

pation without leaving the Kreis would be recorded as a move. Given

the nature of agricultural employment in the late nineteenth century,

35 The empirical evidence, and the theoretical reasons for expecting such a link, aresurveyed in Birdsell (1988).

36 Quante (1955); the data was also used in Quante (1958).37 These estimates are derived from the occupational censuses, which were conducted

in June. These Wgures would therefore include some movements which were purelyseasonal (such as migration to sugar beet areas to do hoeing and other cultivations)although the main period of seasonal migration came later in the year, at harvest time.

38 In 1907 55.2% of the German population was occupied in agriculture, while 41.5%lived in rural areas (taking an average of the 1905 and 1910 censuses). So around 15% ofthe rural population had an occupation outside agriculture.


which typically involved living on the farm where the worker was

employed, a switch to non-agricultural work would usually involve a

move of some description, but this could be no more than a shift to a

nearby town or village. The data set is best described as one which

gives the net agricultural ‘quitting rate’: the rate at which persons occu-

pied in agriculture switched to other occupations. The results of this

analysis are summarized in Table 5.6; the full econometric results are

given in Table 5D.3 in Appendix 5D.

The most striking result from this analysis is the close relationship

between natural population increase and migration. The estimated rela-

tionship indicates that at least 100% of any population increase in rural

areas was exported to other sectors. There was no capacity to absorb

additional labout within agriculture. Any population growth fed directly

into migration (at least within the Kreis).

The quit rate was heavily inXuenced by the position of the Kreis; itwas much higher in areas close to the border with Russian Poland and

fell progressively for Kreise further and further to the west. This would

have led to a high rate of migration out of the east but for the tendency

for the quit rate to be higher on smaller holdings than on larger estates.

Both the presence of large estates (over 100 hectares), and, to a lesser

extent, the presence of medium-sized holdings (20–100 hectares) are

associated with lower quit rates than those on smaller holdings of below

20 hectares.

Table 5.6. Summary of results of statistical analysis of moves out of

agriculture for 390 rural Prussian Kreise, 1882–95


Factors which increase net moves out of agriculture:

A high rate of natural population increase

Being closer to the border with Russian Poland

Factors which decrease net moves out of agriculture or lead to net gains:

Presence of large farms and estates (over 100 hectares)

Presence of medium-sized farms (20–100 hectares)




Factors which increase net moves out of agriculture:

High levels of non-agricultural employment in the Kreis


This Wnding contrasts strongly with the view expressed by many

contemporary writers that large estates led to a higher rate of out-

migration. As the previous section has shown, one explanation of this is

that large estates were associated with higher birth rates and it was this,

not any tendency to shed labour as such, that produced an association

between large estates and out-migration. This issue will be considered

in more detail in the next chapter.

Quit rates were also lower in the Polish and Lithuanian areas. In

these areas population pressure was high, particularly so in Polish-

speaking areas, and this was the main factor driving out-migration.

Table 5.7 gives the predicted quit rates for the 1882–95 period for

Kreise at diVerent distances from the border with Russian Poland, con-

trasting the predicted rates for smallholdings and larger estates. In areas

close to the border with Russian Poland the predicted rates are consid-

erably higher than in the centre of Germany (approximately 500 kmfrom the border). The predicted quit rates are also much lower for the

Polish areas, but the calculations are based on the assumption that the

rates of natural population increase were similar, and they were in fact

much higher in these areas. If the Polish-speaking population had had a

similar rate of population increase to the Germans living in the rural

east, then Polish migrants would have been a relatively unimportant

factor in the overall pattern of German migration.

Table 5.7. Predicted quit rates (per 1,000 of the total agricultural

population) for rural areas with similar rates of natural population

increase, based on analysis of Wgures for Prussian Kreise 1882–95

On large farms and estates

(over 100 hectares)

On small farms

(under 20 hectares)

Average values 166.5 313.4

Distances from the

border with Russian

Poland:

50 km 225.8 372.7

500 km 148.0 294.9

Polish areas 69.2 214.9

Non-Polish areas 182.6 328.3

Note: The calculations use the coeYcients from column c of Table 5D.3, and averagevalues for all other variables.


On average the quit rate for Kreise east of the Elbe was 21% higher

than for Kreise west of the Elbe (284.5 per 1,000 versus 235.5 per

1,000). About half of this can be attributed to higher rates of population

increase feeding directly into migration. The remainder was due to

other factors. Chief amongst these was the proximity of the border with

Russian Poland. Cheap seasonal migrant labour from Russian

Poland could be used to displace permanently employed workers. The

availability of this labour source lowered wage rates in the areas close to

the border, providing an additional reason for farm labourers in these

regions to look for employment elsewhere in Germany.

6. Population Growth in the Recip ient C it ies

The regressions in Chapter 4 showed that the rate of natural population

increase of recipient cities was an important inXuence on migration

rates into those cities. Table 5.8 compares lists of the 20 cities with the

highest rates of natural population increase and those with the lowest

rates (selected from a list of 96 for which Wgures were available).

The signiWcance of these diVerences is evident when compared to the

average rate of growth of German cities in this period, which was 2.6%

per annum (total increase of population from all sources). The cities in

the Wrst list could have supplied much of this increase from the natural

increase of the resident population; those in the second list would have

needed to take in migrants.

A few points emerge from this comparison. The cities in the Wrst list

are predominantly industrial cities, with a heavy concentration in the

Ruhr. Those in the second list are well scattered around the country,

and tend to be longer-established cities, often the capitals of their

regions, or cities which for other reasons might be expected to have a

strong middle-class presence.39

There are two views of the demographic relationship between the

cities and their surrounding regions. The Wrst stresses the leadership

role of the cities, and of the middle class within these cities. Middle-

class attitudes and behaviour would Wrst aVect the working class living

within the cities, and then transfer to the surrounding rural areas. The

39 Laux (1989) shows that natural population growth rates were lower in cities whichwere predominantly commercial or administrative, and these cities tended to have higherrates of in-migration in consequence.


second view is the one expressed by Kollmann: migrants from rural

areas bring their behaviour patterns with them to the cities.

The results reported in Tables 5.2, 5.3, and 5.4 showed that there

was a relationship between demographic behaviour and the proximity of

major cities. In areas nearer to cities the marriage rate was higher, but

fertility relative to the number of married women was lower compared

to more remote regions. This could be interpreted as supporting the

view that changes in demographic behaviour were initiated in the cities

and then moved outwards to contiguous areas. However, the analysis

also indicated that fertility in rural areas responded to factors which

were speciWc to those areas: the presence of large estates and the

Table 5.8. German cities with highest and lowest rates of natural

population increase 1908–12

20 cities with highest rates 20 cities with lowest rates

Annual rate of

increase (%)

Annual rate of

increase (%)

Hamborn 3.16 Potsdam 0.28

Borbeck 3.09 Hildesheim 0.34

Herne 2.93 Regensburg 0.47

Gelsenkirchen 2.93 Ulm 0.51

Recklinghausen 2.63 Gorlitz 0.51

Oberhausen 2.61 Mulhausen 0.62

Bochum 2.51 Frankfurt a. Oder 0.66

Konigshutte 2.31 Metz 0.70

Essen 2.21 Braunschweig 0.72

Gleiwitz 2.18 Liegnitz 0.73

Dortmund 2.12 Dessau 0.75

Dusseldorf 2.10 Rostock 0.76

Rixdorf 2.04 Koblenz 0.76

Mulheim a.d. Ruhr 1.99 Bromberg 0.77

Mannheim 1.92 Wiesbaden 0.79

Plauen 1.88 Berlin 0.82

Ludwigshafen 1.86 Solingen 0.83

Duisburg 1.86 Augsburg 0.83

Saarbrucken 1.75 Wurzburg 0.84

Osnabruck 1.74 Brandenburg 0.84

Source: Calculated from SJDS (1911), table xii, pp. 56–7, and subsequent issues.


institutions associated with these in eastern agriculture, and the preva-

lent inheritance system. Demographic behaviour outside cities was not

just a lagged response to the changes taking place in urban areas.

If migrants from rural areas did transfer demographic behaviour

from their regions of origin then it should be possible to Wnd this eVect

by statistical analysis. The birth rate in rural areas was higher than in

cities, and the diVerence was increasing. The crude birth rate in Prus-

sian cities was 97.6% of the rural level in 1876–80 but only 82.5% in

1906–10.40 Despite this it was not possible to Wnd a statistical relation-

ship between migration and urban fertility. Two data sets were tried.

The Wrst made use of the 1895 census results for the major German

cities (showing the proportion of the population born in rural areas);

the second used Wgures for movements into German cities in the period

1905–10.41

One reason for this may be that the demographic behaviour of mi-

grant families diVered in ways which tended to cancel each other out.

Migrants tended to marry later, which should have reduced overall

fertility.42 But they may have been more reluctant to adopt ‘modern’

patterns of reproductive behaviour, which would have pushed fertility

up. Studies based on household Wgures show relatively small diVerences

in the size of completed families.43

Demographic behaviour was inXuenced by conditions within the cities.

Table 5D.4 in Appendix 5D gives the results of an econometric analysis of

the birth and death rates in 79 German cities in 1910–12. These are

summarized in Table 5.9. The main inXuences came from the occupa-

tional structure of the cities. In cities which had high levels of industrial

employment—which also meant relatively low levels of employment in the

professions and other middle-class activities, and in domestic service—the

40 SJPS (1915), 55–7.41 Using the 1905–10 data set, no signiWcant coeYcients were found either for net

migration or total moves per head of population. The dependent variable was the crudebirth rate. For 1895, a relationship was found between the average marital fertility of theregions from which migrants were drawn (using the Princeton Ig index) and maritalfertility in the recipient cities. The problem here is that most migrants were drawn fromthe provinces within which the cities were situated, and so this result could also beinterpreted as demonstrating the inXuence that demographic behaviour in cities had onthe surrounding countryside. An important drawback with this approach is that it doesnot consider the actual behaviour of migrant families, only the aggregate Wgures for citieswith high levels of migration. A family-level study may well show diVerent results.

42 Lee and Marschalck (2000), 383–4.43 Lee and Marschalck (2000), 384 report that completed family size was 4.3 children

for in-migrants in the city of Bremen 1820–75, but 4.6 for native-born households.


birth rate tended to be higher. It was higher still for the cities dominated

by ‘heavy industry’, mining and metal production.

There was an association between low birth rates and a high vote for

the Social Democratic party in the 1912 election. These were both

indicators of a change in working-class attitudes. Arguably, both were

signs of a greater self-conWdence: an attempt on the part of working-

class households to control or at least inXuence their own destiny. They

were no longer prepared to accept that their lives should continue to be

dominated by forces outside their control. This turned out to be a more

important and consistent eVect than Catholicism, which had little eVect

on demographic behaviour once industrial structure had been included

in the statistical analysis.

Death rates were also inXuenced by industrial structure and by wage

levels: low death rates were found in cities with relatively high levels of

industrial occupation compared to other forms of employment, and in

Table 5.9. Summary of results of statistical analysis of birth rates

and death rates in 79 German cities, 1910–12

Secure statistical results

I. Factors which are associated with high birth rates:

A high level of industrial employment relative to other occupations

A high level of employment in mining and metal production relative to

other industrial occupations

A low Socialist vote in the 1912 election

II. Factors which are associated with low death rates:

High wages


A low level of employment in mining and metal production relative to

other industrial occupations (less secure result)

III. Factors which are associated with high rates of natural increase:


A high level of employment in mining and metal production relative to

other industrial occupations

High wages

A low Socialist vote in the 1912 election

Factors which have no consistent eVect on the rate of natural increase:

Catholicism

A high level of migration


cities with high wages. On the other hand, the heavy industry sector

was associated with higher death rates when compared to other indus-

tries (although the result is not a particularly secure one in statistical

terms). This was most probably due to the nature of the work, and the

illnesses and accidents which resulted from this.

The rates of natural increase which resulted from these birth rates and

death rates show eVects from the same factors: population growth was

highest in industrial cities, with high wages, and relatively low socialist

votes. The positive eVect of high birth rates in the ‘heavy industry’ cities

outweighed the tendency for these to have higher death rates, so the

overall eVect from this factor on population growth was a positive one.

Again the analysis shows no consistent eVect from Catholicism, and

migration, as measured by moves in and out relative to population, was

not found to have any eVect on demographic behaviour.

Table 5.10 illustrates these points for ten selected cities, those with

the highest and lowest rates of population increase. The list of cities

with low rates of increase includes several with small Catholic percent-

ages, but alongside them is Muhlhausen, which was 75.8% Catholic.

Table 5.10. Predicted rates of population increase (per 100,000

resident population) for selected cities, 1910–12

Catholic

%

Wages

1912

Industry

%

Mining/

metal

%

SDP

vote

1912

Rate of

increase

A. 5 cities with lowest rates of increase

Potsdam 8.7 3.25 32.8 0.15 47.4 23.3

Gorlitz 14.0 2.50 49.7 0.13 48.8 38.7

Mulhausen 75.8 2.40 63.3 0.01 56.1 40.3

Wiesbaden 31.8 3.20 37.4 0.13 35.8 50.7

Brunswick 6.5 3.20 50.2 0.13 55.6 53.7

average 27.3 2.91 46.7 0.11 48.7 41.3

B. 5 cities with highest rates of increase

Beuthen 55.1 3.40 66.4 42.8 35.7 189.7

Duisburg 52.8 3.25 65.8 33.0 30.9 192.3

Bochum 50.4 3.30 68.2 45.8 36.8 229.0

Gelsenkirchen 49.7 3.40 74.3 56.5 36.8 252.0

Recklinghausen 75.2 3.30 73.5 64.2 27.3 262.5

average 56.6 3.33 69.6 48.5 33.5 225.1


Other Catholic cities with low rates of population increase were

Wurzburg, Augsburg, and Koblenz. It is this inconsistency that makes

the relationship between Catholicism and population growth such a

weak one.

On the other hand the low growth cities are distinguished by gener-

ally low levels of industrial employment, relatively low wages and

high SDP votes. But above all, they are not cities which have large

mining or metal production sectors. By contrast the cities with high

rates of population growth are all ones where the heavy industry sector

was important. They tended to have lower socialist votes, but wages

were high.

The classiWcation of regions given in Appendix 5A shows that, over

the whole 1871–1910 period, comparison of the death rate and the

lagged birth rate identiWes the Ruhr regions, Arnsberg and Dusseldorf,

as two of the regions which were both highly industrial and which had

high population growth. But a comparison with other industrial regions

shows that the Ruhr was not so unusual when values for the whole

1871–1910 period are considered. Population pressure was as high in

parts of Saxony, Leipzig, and Zwickau, for example. What made the

natural rate of increase in the Ruhr so distinctive is that it remained

high while population growth in other urban areas was tending to fall.

The more isolated working-class culture of the Ruhr was a factor in this

delayed demographic transition.

7. Population Growth and the Lewis Model

So far the analysis in this chapter has concentrated on relative diVer-

ences in population growth between regions. For migration this is the

most important factor. Is the pattern of demographic change in line

with the distribution of economic opportunities? If it is not, then there

will be an adjustment role for migration.

But, in the context of the Lewis Model, the aggregate level of popu-

lation increase also matters. Lewis argued that demographic response

would be a component of the ‘unlimited supplies of labour’ assumed by

his model. He placed most emphasis on the decline in the death rate:

Because the eVect of development on the death rate is so swift and certain,

while its eVect on the birth rate is unsure and retarded, we can say for certain

that the immediate eVect of economic development is to cause the population to

grow; after some decades it begins to grow (we hope) less rapidly. Hence in any


society where the death rate is around 40 per thousand, the eVect of economic

development will be to generate an increase in the supply of labour.44

The reasons Lewis gave for the ‘swift and certain’ decline in the death

rate were, Wrst, that improved communications would reduce the inci-

dence of local famines, and secondly, that there would be improvements

in public health and medicine.

The relationship between development and the death rate does not

appear quite so certain now as it did to Lewis: rather more is known

about high urban mortality in industrializing cities. Life expectancy in

German cities remained low into the twentieth century. Table 5.11

gives life expectancy at birth for the Prussian population, as calculated

by the Prussian Statistical OYce using mortality tables for 1901–5. This

44 Lewis (1954), 144.

Table 5.11. Life expectancy at birth, Prussia 1901–5

Men Women

Towns

and cities

Rural

areas

Towns

and cities

Rural

areas

East Prussia 37.6 42.0 44.8 45.2

West Prussia 37.5 43.4 42.9 45.9

Berlin 41.8 - 47.3 -

Brandenburg 41.7 43.2 47.0 47.5

Pomerania 37.9 46.5 43.6 48.6

Posen 38.7 45.5 43.9 48.5

Silesia 36.5 40.0 41.7 43.5

Pr. Saxony 41.1 44.9 45.2 47.7

Schleswig-Holstein 44.9 52.9 49.7 54.8

Hanover 44.7 50.3 49.4 51.4

Westphalia 41.7 46.6 46.6 48.6

Hesse-Nassau 45.2 49.6 49.5 50.9

Rhineland 42.9 45.8 47.6 47.7

Note: The reasons for high mortality in the east are discussed by Spree (1988), 58. Seealso Hennock (2000) for a discussion of attitudes to urban sanitation. These Wgures mayunderstate the problems of eastern cities. Appendix 5B gives life-expectancy Wgures forindividual cities. The results show that male life expectancy in the major eastern citieswas 29–35 years. There had been an improvement in urban life expectancy since the1870s, but this was still below rural life expectancy.Source: Zdp/SB (1908), i. 26–7.


table shows that in all regions, and for both sexes, life expectancy in

urban areas was below that in the countryside, and the gap could be a

substantial one: 7 years for men in Posen, 81⁄2 years in Pomerania.

National averages tend to obscure this diVerence, as the urban popula-

tion was concentrated in the west and centre, regions where life expect-

ancy was high, while the east had proportionately more of the rural

population. The ‘centre of gravity’ of the rural population was further

to the east compared to that of the urban population.45

One evident conclusion is that, in the absence of balancing variations

in fertility, cities would have to draw in migrants from rural areas to

make up for their higher mortality rates. Lower rural mortality should

have made it easier to Wnd migrants from the countryside nearby.

Migration over longer distances had a mixed eVect on mortality. On

the one hand, it transferred population from rural areas to cities, which

would tend to increase mortality. On the other, it moved population

from the east to the west, which should have led to a reduction. But, as

most moves were over shorter distances, and did result in net gains by

cities, it can be seen that the overall eVect would have tended to raise

mortality.

If urbanization tended to raise mortality, then the decline in the

German death rate from 1870 onwards becomes even more impressive.

However, by European standards German life expectancy in the period

just before the First World War was not high. At 49.0 years in 1910–11(both sexes) it was below France (50.4), England and Wales (53.5), and

the Netherlands (56.1). In Sweden, Norway, and Denmark life expect-

ancy was over 57 years. Comparative studies show that the decline of

mortality in Germany was relatively late; in other Western European

countries it had started earlier.46

45 Life-expectancy tables for the early 1900s show a substantial diVerence betweenurban and rural areas for the male population at all ages, while there was little diVerencefor women past the Wrst year of life (urban infant mortality was high for both sexes). Formen the gap was 3.8 years at birth and 3.6 years at age 10, Vogele (1998), 50–1. Thiscalculation, for the population as a whole, underestimates the gap between urban areas andcontiguous rural areas.

46 SchoWeld et al. (1991), 47. There had been considerable progress in reducing mortal-ity in German cities since the 1870s: see Vogele (1998) and Vogele and Woelk (2000) forstudies of this. The fact that German life expectancy was still not high by Europeanstandards is an indicator of the severity of the problems faced in the 1870s and 1880s.The high rates of migration in this period had contributed to a major crisis of mortality,which was only gradually overcome. By 1905, however, migrant life expectancy in themore advanced cites was below that of the native-born population, Lee and Marschalck(2000), 381.


On the other hand, Lewis’s ‘uncertain and retarded’ decline in fertil-

ity set in fairly quickly. The Princeton study showed that fertility

within marriage started to fall from the late 1870s onwards. This was

initially oVset by a rise in the proportion married. If there had not been

this decline in marital fertility, then population growth in the 1880s and

1890s would have been much higher.

The Princeton study showed that the decline in marital fertility started

earlier in urban areas, but village studies, and the evidence of the regres-

sions presented in this chapter, make the point that this was not a com-

pletely new phenomenon. Some rural areas had already reduced fertility

within marriage. If rural areas tended to place more emphasis on mar-

riage restrictions, it may well be that the reason for this was that the

degree of social control required to maintain a high average age at Wrst

marriage was easier to achieve in village communities.

It seems therefore that the decline in marital fertility which began in

the 1870s had its roots in attitudes and objectives which were also

found in rural society, and which had already found expression in the

Western European pattern of delayed marriage. EVectively, population

growth was linked to economic objectives: the household unit sought to

regulate population increase so as to achieve a desired standard of living

for itself and its children. Late marriage was particularly linked to the

institution of the nuclear family, and to the custom that children should

set up a separate household on marriage. Parental approval of marriage

would depend on the couple having suYcient means to do this, and to

achieve a respectable standard of living in the new household.47

This implies that there was a gradual upwards adjustment of the

objective. Even in pre-industrial Europe there was some increase in

living standards, although this was intermittent and uncertain. If the

household made no adjustment of the desired standard of living, then

in times of prosperity there would have been a surge of population

growth suYcient to wipe out any gains achieved. By and large this did

not happen.

If this argument is correct then the German rural population which

moved into industrial cities carried with it attitudes which predisposed

it to control fertility within marriage. The incentive for a married

couple to move out of a family home consisting of a one- or two-room

47 This meant that eVective population control could be exercised simply by refusingpermission to set up a new household, as happened in Mecklenburg for example, Lubinski(1995).


apartment in a Berlin Mietskaserne (or tenement block) was at least as

great as the incentive to move in a rural village. The strategy may have

changed. Increased employment opportunities for married women with-

out children would have made it more attractive to marry early but put

oV having children. But the target was the same.48

Moreover, the ability to adjust objectives would have also aVected

behaviour. Cities oVered new opportunities: white-collar jobs, appren-

ticeships, accessible secondary schools. Even if migrants themselves

were not able to take advantage of these possibilities, their children

might, if there were not too many of them. In cities with a strong

middle-class presence these opportunities may well have been greater,

or at least more apparent.49

The demographic aspect of the Lewis Model was therefore curtailed.

Germany had to deal with quite a high rate of population increase in

the 1880s and 1890s, but this would have been higher still if cities in

central Germany had not started to reduce fertility from the 1870s

onwards. This in turn made these cities take in migrants from other

areas where the shift in demographic behaviour was less advanced.

The intention in this chapter has been to concentrate on those

aspects of demographic behaviour which aVect migration. This meant

that particular attention was given to mechanisms which might adjust

population growth to available economic opportunities within a region.

The main conclusion is that this did not happen in the rural east, and

that this was an important reason why the region was a major source of

migrants in the 1871–1913 period. The evidence of the regressions at

the Kreis level shows that demographic factors had major eVects on

migration. But this does not mean that the east was in a crisis of

overpopulation in 1870–1913 similar to that of the early nineteenth

century. There was less evidence of demographic distress. It was not

just population pressure which led migrants to leave the east.

48 Neumann (1978) provides evidence of fertility control in urban working-class house-holds.

49 The idea that marriage in Western Europe was linked to economic circumstances isfairly widely accepted. This may have been a long-established pattern of behaviour, asargued by Habbakuk (1958). But this has the implication that European households wereaccustomed to adjust demographic behaviour in general to economic circumstances, andmay therefore have been unusually predisposed to reduce fertility in response to improvedopportunities to invest in human capital, leading to the downward phase of the demo-graphic transition. The extent of the responsiveness of demographic variables to economicconditions may be a European peculiarity.


The next two chapters will look more closely at conditions in the

rural east. The position of the rural labourer will be considered. Evi-

dence on wage movements, and on changing social relationship will be

presented. The role of changing technology will also be examined. But

it is important to remember that the east was not the only agricultural

region. Changes were taking place in agriculture in other regions. It was

the combination of the fact that the east was predominantly agricultural,

with a rate of population increase which was high relative to opportun-

ities in local agriculture or industry, that made migrants leave the east.


Appendix 5A. A Class if ication of German Regions

A. Highly industrial regions

1. Regions with high population growth: 2. Regions with low population growth:

Arnsberg (43.0/23.3) Starkenburg (37.4/22.4) Wiesbaden (33.7/20.5) Aachen (36.1/23.2)

Zwickau (48.6/28.9) Brunswick (35.6/20.7) Koln (38.7/24.9) Bremen (35.5/22.9)

Dusseldorf (41.0/22.8) Dresden (40.3/24.7) Saxony-Coburg-Gotha (35.4/22.2)Hamburg (35.1/22.8)

Leipzig (42.3/25.1) Berlin (38.7/23.7) Rheinhessen (35.3/22.4) Lubeck (32.0/20.3)

Reub -a.l. (42.4/25.5) Potsdam (38.1/23.9) Reub -n.l. (40.3/26.2) Bautzen (36.6/24.6)

B. Moderately industrial regions


Neckarkreis (41.6/24.5) Sch.-Lippe (33.1/18.7) Magdeburg (38.8/24.7) Hildesheim (34.0/22.3)

Erfurt (39.2/22.9) Saxony-Meiningen (35.9/21.4)Mannheim (38.0/24.8) Breslau (40.6/28.4)

Anhalt (38.3/21.8) Schleswig-Holstein (33.6/20.0)Schwarzburg-Rudolfs. (35.1/21.9) Upper Alsace (34.6/25.1)

Merseburg (40.7/24.6) Saxony-Weimar (35.0/22.0) Munster (32.0/23.3)

Schwarzburg-Sond. (35.7/21.1)Hanover (35.8/21.7) Saxony-Altenburg (40.9/27.9) Lorraine (29.4/22.0)

Karlsruhe (38.0/24.5)

C. Moderately agricultural regions


Danzig (44.2/27.0) Trier (37.5/22.5) Kassel (36.0/22.4) Upper Alsace (34.7/24.0)

Stettin (40.1/24.3) Frankfurt a. Oder (38.3/24.1) Oldenburg (33.0/20.7) Mittelfranken (38.4/27.6)

Oppeln (43.7/27.3) Schwarzwaldkr. (41.7/26.7) Koblenz (35.3/22.7) Liegnitz (37.4/28.2)

Pfalz (38.1/22.8) Minden (35.8/21.8) Donaukreis (41.5/27.9) Oberpfalz (40.6/30.9)

Lippe (36.0/21.1) Mecklenburg-Schwerin

(32.9/20.4)

Stralsund (36.1/24.0)

Mecklenburg-Strelitz (32.1/21.7)

Oberbayern (39.6/30.3)

D. Highly agricultural regions


Konigsberg (40.7/22.0) Aurich (32.8/18.9) Waldeck (35.2/21.5) Konstanz (34.7/24.2)

Marienwerder (46.7/26.7) Gumbinnen (42.4/27.1) Sigmaringen (37.8/26.1) Unterfranken (34.2/24.1)

Potsdam (44.2/25.1) Stade (34.7/21.0) Oberfranken (34.6/23.5) Freiburg (31.5/22.5)

Bromberg (45.8/27.1) Jagstkeis (40.3/25.8) Oberhessen (31.2/20.7) Schwaben (39.6/29.8)

Koslin (42.6/26.5) Osnabruck (34.3/20.8) Luneburg (31.0/20.8) Niederbayern (39.5/31.0)

Note: Regions classiWed according to percentage occupied in agriculture 1871–1910. Figures in brackets are average birth rate per 1,000lagged 20 years followed by the average death rate.

Appendix 5B. L ife Expectancy at B irth in Major C it ies ,

in Years , Calculated from 1893–4 Mortality Tables

Men

East Prussia 36.2 Konigsberg 29.5 Rest of Province 36.6

West Prussia 38.6 Danzig 33.6 Rest of Province 38.9

Brandenburg 37.9 Berlin 35.8

Pomerania 40.5 Stettin 29.2 Rest of Province 41.0

Posen 39.5

Silesia 34.9 Breslau 30.0 Rest of Province 35.2

Prussian Saxony 39.6 Magdeburg 36.2 Rest of Province 39.8

Halle 36.0

Schleswig-Holstein 43.9 Altona 37.0 Rest of Province 44.4

Hanover 43.2 Hanover 41.0 Rest of Province 43.3

Westphalia 40.9

Hesse-Nassau 42.6 Frankfurt am Main 38.5 Rest of Province 42.8

Rhineland 39.6 Koln 33.6 Rest of Province 40.1

Dusseldorf 37.5

Elberfeld 39.7

Barmen 42.0

Crefeld 36.6

Aachen 35.0

All Prussia 39.0

Women

East Prussia 40.1 Konigsberg 35.8 Rest of Province 40.3

West Prussia 42.2 Danzig 39.5 Rest of Province 42.4

Brandenburg 41.9 Berlin 41.1

Pomerania 43.5 Stettin 35.2 Rest of Province 43.8

Posen 42.9

Silesia 38.5 Breslau 35.2 Rest of Province 38.6

Prussian Saxony 42.8 Magdeburg 39.6 Rest of Province 42.9

Halle 41.9

Schleswig-Holstein 46.4 Altona 42.1 Rest of Province 46.7

Hanover 44.8 Hanover 43.5 Rest of Province 44.8

Westphalia 42.9

Hesse-Nassau 44.7 Frankfurt am Main 44.4 Rest of Province 44.7

Rhineland 42.5 Koln 38.0 Rest of Province 42.6

Dusseldorf 43.0

Elberfeld 46.6

Barmen 45.8

Crefeld 41.4

Aachen 41.1

All Prussia 42.1

Source: Kucyzinski (1897).


Appendix 5C. Descr iption of Regress ion Variables and

Sources

(i) Tables 5.2 and 5.5

TOTFERTILITY Numbers of children aged under 1 year per

1,000 women aged 15–50, 1895 census, PS

(1897–8) vol. 148, pp. 60–176.MARRFERTILITY As above, children per 1,000 married women

aged 15–50.

%MARRIED As above, percentage of women aged 15–50

who were married.

LAND>100ha The percentage of the total agricultural area

in holdings of over 100 hectares, from 1895

Agricultural Census, SDR n.f. 212.

%CATHOLIC The percentage of the population who were

Catholic, 1890 census, PS, vol. 121, pp. 152–

98.

%POLISH As above, the percentage of the population

who spoke Polish.

%LITHUANIAN As above, the percentage of the population

who spoke Lithuanian.

RURALWAGES Average wages (Ortsubliche Tagelohne) for

rural areas; the Wgures are averages of male

and female daily rates, 1892 and 1901,

ZdKPSB (1904), 320–8.

%PARTIBLE The percentage of land transferred by inher-

itance in 1896–1902 which was divided,

ZdKPSB (1905), 262–7.

CITYDISTANCE(b) Distance in kilometres from the Kreis mid-

point to the nearest city with at least 50,000

inhabitants in 1900 (measured from contem-

porary maps).

(ii) Tables 5.4 and 5.6

Net quit rate

(dependent variable)

The estimated number of net moves out of

agriculture, by Kreis, 1882–95 and 1895–

1907, per 1,000 of the average population

occupied in agriculture, from Quante (1933),38–58.


NATINCREASE As above, estimated numbers of total births

less total deaths, per 1,000 of the average

population occupied in agriculture.

RURALWAGES92 As for RURALWAGES, using only 1892

Wgures.

RUSSDISTANCE Shortest distance from the Kreis mid-point

to the border with Russian Poland.

CITYDISTANCE(c) Distance to the nearest city with at least

20,000 inhabitants in 1900.

(iii) Appendix Tables 5D.4 and 5D.5

Birth rates, death rates,

rates of natural increase

(dependent variables)

All Wgures are rates per 10,000 head of

population, taken from SJDS (1911), 67–8;

(1912), 70–1; and (1913), 76–7.CATHOLIC% The percentage of the population who were

Catholic, 1910 census, SJDS (1911), 678–9.

WAGES1912 Average wages (Ortsubliche Tagelohne) for

male workers, aged 16 and over, from SJDS

(1913), 826.

INDUSTRY% The share of the total occupied population

(both sexes) occupied in industry, 1907

Occupational Census, SDR (1909), n.f.

volumes 207 and 209.

MINING/METALS% As above, the share of the industrial work-

force occupied in the mining and metal pro-

duction sector.

SDPVOTE1912 Social Democrat share of the vote from Die

Reichstagswahlen von 1912, SDR, vol. 250

(1913).

MOBILITY Data from SJDS 1911–13: total moves for

each city in 1910–12 divided by total popu-

lation.

(iv) Appendix Table 5E.1

MARRIAGEt


The number of marriages in each province

per 1,000 head of population in each period

(annual data), from Kollmann (1980).


WARDUMMY Dummy for war years (1866, 1870, and

1871).

RYEPRt Average rye prices for each province, from

ZdKPSB (1907), 84–92.

POTATOPRt,

WHEATPRt, BEEFPRt,

PORKPRt and

BUTTERPRt

As above, for potatoes, wheat, beef, pork,

and butter.

WAGESt Wages paid to male workers in Prussian state

forests, daily rate in Marks, from Eggert

(1883), decade averages.

Appendix 5D. Econometric Results

(i) The analysis of fertility in the rural Prussian Kreise

The aim of this analysis was to examine factors which aVected fertility

decisions in rural areas, so as to consider the eVect of agricultural

conditions on migration when these factors operated indirectly—

through their eVect on the demographic balance in the rural area. Mi-

gration would tend to occur when the supply of labour exceeded the

ability of the local economy to supply employment. Information on

fertility was derived from the 1895 census, by comparing the numbers

of children to the numbers of women of child-bearing age.

Two measures of fertility were used as dependent variables: TOT-

FERTILITY is the number of children aged under 1 year per 1,000women aged 15–50; MARRFERTILITY is the number of children

aged under 1 year per 1,000 married women aged 15–50. This second

variable has the disadvantage that it includes all children, not just those

born within marriage (these are not distinguished by the published

census results). A third dependent variable is the percentage of women

aged 15–50 who were married.50

The explanatory variables include two which measure the ethnic

composition of the population, speciWcally the presence of Polish- or

50 In societies with high infant mortality these fertility variables will be biased due todeaths in the Wrst year. On average around 8% of children would have died before theywere recorded by the census. However, being rather brutal about this, from a migrationviewpoint, what matters is not the number of births but the number of children whosurvived. So this bias is not a serious problem.


Lithuanian-speaking minorities. The inclusion of these variables is im-

portant if the eVect of Catholicism is to be accurately estimated. The

coeYcients on %LITHUANIAN are not signiWcant, which is not so

surprising as there were only six Kreise where there was a substantial

minority (or majority—Lithuanian-speakers were 58% of the popula-

tion of Kreis Heydekrug in East Prussia). The variable was left in the

equations to avoid creating a bias in the other estimates.

The behaviour of the Polish-speaking and German-speaking Cath-

olics is shown to have been very diVerent. Overall fertility was 3%

higher, compared to the Prussian average, for German Catholics, but

Table 5D.1. Regression results for the analysis of fertility in rural

Prussian Kreise, 1895, estimation by OLS, t-ratios in brackets

Dependent variable is:

a

TOTFERTILITY

b

MARRFERTILITY

c

%MARRIED

Constant 89.5 183.3 23.0

LAND>100ha þ0.214 þ0.342 �0.0007

(4.90) (4.56) (0.13)

%CATHOLIC þ0.0406 þ0.494 �0.0399

(2.01) (14.25) (15.5)

%POLISH þ0.324 þ0.234 þ0.0350

(9.44) (3.96) (8.00)

%LITHUANIAN þ0.216 þ0.508 �0.0244

(1.32) (1.81) (1.17)

RURALWAGES þ21.0 þ24.2 þ2.10

(6.92) (4.64) (5.42)

N 477 477 477

S 13.90 23.89 1.77

R2 0.319 0.442 0.385

adj R2 0.311 0.436 0.378


RSS 91148 269470 1485

Note: DeWnitions and sources are given in Appendix 5C. The analysis is a crude one bythe standards of modern demographic studies: there is no attempt to measure fertility bycohort or to make allowance for diVerences in the age structure of the female populationin the diVerent Kreise. The main reason for this was that the data needed to do this werenot available from the sources used. The analysis serves the needs of this study, which isto examine the eVect of demographic factors on migration.


27% higher for Polish Catholics. The reasons for this are of interest.

Although fertility within marriage was high for German Catholics, they

were much less likely to be married (16% less likely than the rest of the

Prussian population).Thus the practice of late marriage found in south-

ern Germany was also found in the Catholic areas of Prussia (something

which can be seen from Figure 5.7 as well).

The presence of large farms and estates is shown to raise fertility,

though it does not aVect the likelihood of marriage. The coeYcient

implies that the rural population in areas where farms were over 100

hectares had fertility rates 16% higher than the rest of the population.

But the eVect of this on the east was cancelled by the Wnding that low

wages depressed fertility.

In Table 5.D2 the eVect of adding additional variables is considered.

The addition of %PARTIBLE (the percentage of land transferred by

inheritance in 1896–1902 which was divided) and CITYDISTANCE(a)

(the distance to major cities) did not have a large eVect on the other

estimated coeYcients, although the overall eVect of Catholicism was

reduced.51 The eVect of partible inheritance was to reduce overall fertil-

ity, which may surprise some. The likelihood of marriage was increased,

however, which leads to the conclusion that, barring major changes in

the percentage of children born outside marriage, there must have been

a major diVerence in fertility within marriage in the regions where

partible inheritance was common. The implied reduction in marital

fertility is about 15%. What this suggests is that, in these regions,

families were making eVorts to reduce fertility in order to avoid further

subdivision of the family property.52

Remoteness, as measured by the distance from major cities, had the

eVect of raising fertility within marriage, and reducing the percentage

married. Rural areas closest to cities had a more ‘modern’ demographic

pattern: earlier marriage, and lower fertility within marriage. The over-

all eVect on total fertility was not signiWcant.The same data set was also used for the analysis of mortality rates.

The results were of no particular interest. Neither LAND>100ha nor

51 One reason could be that there were 55 Kreise observations which had to be droppedbecause %PARTIBLE was not available, and these were predominantly in the Catholicareas of the Rhineland.

52 According to a contemporary observer, Wydgozinski in Sering (1899–1910), i. 116,there was indeed a tendency in the partible inheritance areas to restrict fertility in order toavoid the excessive splitting of holdings: a ‘two-child’ system was practised (Zweikinder-system).


%PARTIBLE had any signiWcant eVect on mortality. The only clear

result was a tendency for mortality to rise where wages were low.

(ii) Analysis of migration by Kreis, 1882–1895

The connection between migration and population growth can be exam-

ined by making use of an analysis conducted by the Prussian Statistical

Table 5D.2. Regression results for the analysis of fertility in rural

Prussian Kreise, 1895, estimation by OLS, t-ratios in brackets

a b c

TOTFERTILITY MARRFERTILITY %MARRIED

Constant 88.4 166.7 24.8

LAND>100ha þ0.195 þ0.346 �0.0040

(4.34) (4.52) (0.72)

%CATHOLIC þ0.0393 þ0.0435 �0.0350

(1.61) (10.45) (11.50)

%POLISH þ0.336 þ0.267 þ0.0337

(9.17) (4.33) (7.43)

%LITHUANIAN þ0.216 þ0.411 �0.0147

(1.31) (1.47) (0.72)

RURALWAGES þ23.0 þ33.0 þ1.33

(6.58) (5.58) (3.05)

%PARTIBLE �0.149 �0.330 þ0.0112

(2.94) (3.81) (1.77)

CITYDISTANCE(b) �0.00013 þ0.00601 �0.000658

(0.11) (3.14) (4.74)

N 424 424 424

S 13.9 23.7 1.72

R2 0.345 0.436 0.333

adj R2 0.334 0.426 0.321

F-test 31.4 45.9 29.6

RSS 80443 233943 1238

Note: This data set was also used to examine the eVect of out-migration on marriagerates. The inclusion of a term giving rates of out-migration between 1895 and 1905improved the Wt, and the variable had a t-statistic of 11.00, indicating a highly signiWcantresult. As the migration Wgures refer to a period after the census used for the dependentvariable, the result cannot be regarded as an entirely satisfactory one, and there mustbe a question about the direction of causation. For this reason, the result has not beenreported, although it certainly suggests that marriage rates were depressed in regions ofout-migration.


Bureau and used in an article in 1933.53 This made indirect estimates of

the rate of migration out of agriculture by applying rates of fertility and

mortality recorded for the rural population. These estimated rates of

natural increase for the population occupied in agriculture were then

used to derive estimates of net migration, using the indirect method

(comparing census returns with the predicted natural increase).

The dependent variable is best described as the net agricultural ‘quit-

ting rate’: the rate at which persons occupied in agriculture switched to

other occupations, since it is not clear whether or not the people who

left agriculture actually migrated. It may be surmised that most did

make some move: although some may have found non-agricultural em-

ployment in their village, the majority probably had to leave for a local

town at the very least. But many may not have left the Kreis, particu-

larly if this was in an area where non-agricultural employment was

rising rapidly.

The data set makes it possible to examine the eVect of variations in

population growth on the net quitting rate, along with the inXuence of

other variables, such as agricultural structure. Bivariate correlation an-

alysis shows a positive relation between the presence of large farms and

estates (where LAND>100HA is high) and the net quitting rate: out-

ward migration was apparently higher on the larger estates. However, in

multiple regression analysis, this eVect is reversed once the rate of

natural population increase is introduced into the estimated equations.

In other words, the apparent relationship between large estates and out-

migration is entirely due to demographic factors.

The Wrst column of Table 5.D3 gives the results of Ordinary Least

Squares estimation of the basic equation. The estimated coeYcient on

LAND>100HA is negative, as is the coeYcient on LAND20–100HA,

though smaller. Large estates were the ones with the lowest quit rates;

quit rates were highest on holdings of less than 20 hectares.

The quit rate was reduced by the presence of Poles or Lithuanians.

The fact that Poles, in particular, have Wgured quite heavily in accounts

of migration, has much to do with high birth rates: these non-German

groups had no great inclination to migrate, and were rather less likely to

leave agricultural employment than Germans in similar circumstances.

It was population pressure which drove up migration from the Polish-

speaking areas.54

53 Quante (1955); the data were also used in Quante (1958).54 If NATINCREASE is omitted from the estimated equation, the coeYcient on

%POLISH becomes positive and signiWcant.


The negative coeYcient on %AGOCCP1882, the percentage occu-

pied in agriculture, shows that the availability of opportunities to Wnd

non-agricultural employment within the Kreis increased the quit rate.

This, however, would not lead to increased net out-migration. In the

Table 5D.3. Regression results for moves out of agriculture by

Kreis, 1882–95

Estimation method

OLS OLS IV IV

a b c d

Constant 184.4 182.2 155.9 190.4

LAND>100HA �1.245 �1.230 �1.469 �1.550

(6.42) (6.30) (4.10) (3.84)

LAND20-100HA �0.585 �0.565 �0.643 �0.624

(3.76) (3.58) (3.65) (3.10)

%AGOCCP1882 �0.507 �0.540 �0.361 �0.431

(3.19) (3.27) (1.42) (2.15)

RUSSDISTANCE �0.161 �0.159 �0.173 �0.154

(10.30) (10.00) (7.95) (5.21)

%POLISH �0.826 �0.843 �1.109 �1.183

(5.33) (5.38) (2.70) (2.25)

%LITHUANIAN �1.687 �1.698 �1.749 �1.711

(2.99) (3.00) (3.01) (2.94)

NATINCREASE þ1.051 þ1.043 þ1.218 þ1.271

(21.39) (20.70) (5.32) (3.66)

CITYDISTANCE(a) þ0.110

(0.76)

%PARTIBLE �0.17

(0.02)

RURALWAGES92 �0.39

(0.84)

N 390 390 390 339

S 46.45 46.47 47.15 46.41

R2 0.672 0.673 0.474 0.483

adj R2 0.666 0.666 0.464 0.469

F-test 111.84 97.82 49.16 34.12

RSS 824158 822914 849148 708756

Note: Dependent variable is the estimated net quitting rate per 1,000 for the totalagricultural population.


more industrial areas the quit rate was high, but many of those leaving

found non-agricultural work within the locality.

But the most important result to emerge from this analysis is the

importance of the rate of natural population increase. The eVect is a

large one. The estimated coeYcient in column a is just over 1.0, indi-

cating that, in Kreise with above-average rates of population increase,

over 100% of this surplus left agriculture. There was no capacity to

absorb surplus labour within agriculture.

In regions where non-agricultural employment was increasing, sur-

plus labour could Wnd work without migrating. In other regions where

there were limited non-agricultural opportunities the surplus had to

migrate. Demographic pressure fed directly into migration in these

regions.

There are problems of possible endogeneity bias in this equation.

High migration rates aVect the population balance and this will have an

inXuence on the rates of natural increase. In areas with high inward

migration, the population will tend to be younger, and this will reduce

the crude death rate, pushing up the rate of natural increase. One way

to deal with this is to use instrumental variables, if a suitable instrument

can be found. The distance to large cities turned out to be a possible

instrument. Its inclusion had little eVect on the estimated equation

(column b) but it was well correlated with the rate of natural increase.55

Using this as an instrument produced some changes in the estimated

values (comparing column a with column c), and reduced the overall Wt

of the equation, but pushed up the coeYcient on NATINCREASE to

over 1. The new coeYcient suggests that 120% of the natural increase

in the population left agriculture. This implies that the overpopulated

Kreise were releasing a surplus built up in previous periods of high

population growth as well as the surplus created by high current rates

of population increase.

The Wnal column introduces other variables. There is no eVect frompartible inheritance on the quit rate. High rural wage levels have a small

negative eVect on the quit rate, but the coeYcient is not signiWcant.

(iii) Analysis of demographic behaviour in urban areas

Econometric analysis of demographic variables for German cities can be

used to show the eVect of industrial structure and other factors on rates

55 A correlation coeYcient of 0.274 (p-value of 0.000).


of natural increase. Cities with low rates of natural increase would need

higher rates of inward migration to provide the increased labour force

needed for industrial expansion.

The basic model uses three variables: the Catholic percentage of the

total population, the level of wages, and the percentage of the occupied

population who are occupied in industry. In column a of Table 5D.4

the dependent variable is the birth rate for the years 1910–12. Theresults show that the birth rate was highest in Catholic industrial cities.

The Wnding that the industrial percentage has an eVect on the birth rate

is consistent with the view that the working class in these cities was

relatively isolated and therefore less likely to emulate middle-class

reproductive behaviour, although other explanations cannot be ruled out.

In column b the eVect of industrial structure is examined by includ-

ing MINING/METALS% (the percentage of the industrial workforce

occupied in mining and metal production) amongst the explanatory

variables.56 The inclusion of this variable has an eVect on the coeY-

cients estimated for the other variables. The eVect of Catholicism is

greatly reduced, as is the eVect of industrial employment in general.

The mining and metal production sector was characterized by large-

scale enterprises and a strong ‘isolated’ working-class culture.

The eVect of Catholicism is further reduced, eVectively eliminated,

by the inclusion of SDPVOTE1912, the Social Democrat vote in the

1912 election as a percentage of all votes. This should not be inter-

preted as a direct causation; rather, both changed demographic behav-

iour, and a willingness to vote for the SPD, should be taken as

indicators of changes in working-class mentality. The more conWdent,

and independent, working class which was emerging in many German

cities was prepared to express this in votes for the more radical pro-

gramme of political and social change espoused by the SPD. Similar

changes in social attitudes led to an alteration in reproductive behav-

iour, and a reduction in family size, which led to a fall in the birth rate.

The Wnal column of Table 5D.4 gives the results of a similar analysis

for the death rate. Here, the importance of wage levels is much greater,

having had an insigniWcant eVect on birth rates. Mortality was consider-

ably lower in better-oV cities. Mortality was also lower in more indus-

56 Other variables of this type were also considered (for other industrial sectors and forthe service sectors). This was the one which the analysis showed to have had a consistentand signiWcant eVect on demographic behaviour.


trial cities, as long as the industry involved was not mining or metal

production: these two variables have opposite signs.

Combining the birth rates and the death rates produces the overall

rate of natural increase, which is the most signiWcant variable for the

explanation of migration patterns. This is analysed in the regressions

reported in Table 5D.5. These show that the eVect of Catholicism is

not secure or consistent. High wage levels have a positive eVect on rates

of population increase, and this coeYcient is signiWcant at the 5% level

once MINING/METALS% is included in the explanatory variables

(columns b to d). The industrial structure variables (INDUSTRY%

and MINING/METALS%) have a consistent and signiWcant eVect on

population growth. The eVect of heavy industry is somewhat reduced

by the tendency of this sector to produce high death rates as well as

high birth rates, but the overall eVect is a positive one.

Table 5D.4. Regression results for the analysis of birth and death

rates in 79 major cities, 1910–12


Birth rate Birth rate Birth rate Death rate

a b c d

Constant 180.3 232.0 241.2 259.4

CATHOLIC% þ0.641 þ0.293 þ0.038 þ0.162

(3.67) (2.03) (0.22) (1.88)

WAGES1912 �6.8 �8.6 þ0.8 �25.1

(0.57) (0.93) (0.08) (4.51)

INDUSTRY% þ1.609 þ0.738 þ0.932 �0.680

(4.14) (2.16) (2.76) (3.33)

MINING/METALS% þ2.220 þ2.057 þ0.352

(6.88) (6.48) (1.82)

SDPVOTE1912 �0.916

(2.57)

N 79 77 77 77

S 42.78 33.11 31.89 19.79

R2 0.297 0.584 0.620 0.361

adjusted R2 0.269 0.561 0.593 0.325

F-test of regression 10.56 25.29 23.13 10.16

RSS 137246 78929 72210 28194

Note: Dependent variables are annual rates per 10,000 inhabitants.


The socialist vote continues to be associated with a reduced rate of

population increase (column c). The Wnal column includes MOBILITY

(the number of total moves in and out of each city relative to the total

resident population) amongst the explanatory variables. This has no

signiWcant eVect on demographic behaviour. This conWrms the Wnding,

reported in the chapter, that it is not possible to discover any relation-

ship between migration and demographic behaviour in urban areas,

using the sort of aggregated analysis reported in this chapter. A micro-

level study, using individual household data, might produce a diVerent

result.

Statistical analysis of results for an earlier period, the years 1899–1901, broadly conWrmed the Wndings of this section. The more limited

data availability meant that the analysis had to be conWned to 27 major

cities, with a resulting loss of degrees of freedom. But the econometric

results also showed a strong eVect on the rate of natural increase from

Table 5D.5. Regression results for the analysis of the population

growth in 79 major cities, 1910–12

a b c d

Constant �69.2 �27.4 �18.6 �11.5

CATHOLIC% þ0.432 þ0.131 �0.112 �0.176

(3.01) (1.13) (0.83) (1.05)

WAGES1912 þ16.9 þ16.4 þ25.5 þ28.9

(1.74) (2.21) (3.34) (3.42)

INDUSTRY% þ2.186 þ1.418 þ1.603 þ1.620

(6.84) (5.18) (6.04) (5.34)

MINING/METALS% þ1.868 þ1.713 þ1.726

(7.22) (6.87) (5.67)

SDPVOTE1912 �0.873 �0.971

(3.12) (2.97)

MOBILITY �0.3568

(0.9)

N 79 77 77 69

S 35.19 26.54 25.07 26.15

R2 0.460 0.697 0.733 0.700

adj R2 0.439 0.680 0.714 0.671

F-test 21.34 41.36 39.03 24.09

RSS 92860 50721 44620 42410

Note: Dependent variable is the annual rate of natural increase per 10,000 inhabitants.


the percentage occupied in industry, and from heavy industry, and also

conWrmed that population growth was highest where wages were high.

Catholicism was found to be relatively unimportant.

Appendix 5E. Marriage and Economic C ircumstances

The argument that late marriage was a distinctive Western European

institution, which linked marriage rates to economic conditions, and

Table 5E.1. Panel data analysis of marriage rates in Prussian

provinces 1816–75

Random eVects model Fixed eVects model

a b c d

Constant 3.83 3.98 � �MARRIAGEt�1 þ0.515 þ0.526 þ0.503 þ0.52143

(15.48) (16.09) (9.44) (10.50)

WARDUMMY �1.176 �1.151 �1.209 �1.161

(8.09) (7.93) (7.07) (6.78)

RYEPRt �0.0149 �0.0155 �0.0140 �0.0151

(5.90) (6.19) (5.16) (5.93)

POTATOPRt �0.0012 �0.0011 �0.0014 �0.0011

(2.30) (2.01) (2.76) (2.29)

WHEATPRt þ0.0120 þ0.0123 þ0.0116 þ0.0121

(6.55) (6.72) (5.91) (6.30)

BEEFPRt þ0.0266 þ0.0301 þ0.0261 þ0.0306

(3.52) (4.11) (3.14) (3.73)

PORKPRt �0.0104 �0.0141 �0.0098 �0.0146

(1.69) (2.44) (1.45) (2.05)

BUTTERPRt þ0.0014 þ0.0011 þ0.0014 þ0.0011

(0.52) (0.39) (0.48) (0.38)

WAGESt �0.0001 �0.0050 þ0.0027 �0.0048

(0.02) (1.37) (0.47) (1.42)

TIME �0.0072 �0.0101

(1.67) (1.79)

N 711 711 711 711

R2 0.668 0.667 0.687 0.685

RSS 0.256 0.246 0.226 0.228

Note: t-statistics in brackets, calculated using White’s heterocedasticity-corrected standarderrors.


thus formed a ‘positive check’ on population growth, to use Malthus’s

terminology, is associated with the work of Hajnal, although it is cer-

tainly of much older origin.57 The idea can be found in the work of

German demographers in the late nineteenth century, and indeed

delayed marriage is Malthus’s moral check on population growth.

There has, however, been criticism of the view that non-Western soci-

eties inevitably tended to push population growth to the point where

wages were reduced to the subsistence level. Some form of population

control seems to have operated in Tokugawa Japan and in Imperial

China, although the mechanism did not involve delayed marriage.58

If marriage was related to economic circumstances in Western

Europe, then it should be possible to Wnd some connection between

economic variables and marriage rates. In the period before the trans-

formation to an industrial society these variables will be primarily re-

lated to agricultural conditions. Good harvests should produce gains in

real wages for agricultural workers, as a result both of lower living costs

and of higher demand for rural labour. High urban demand for agricul-

tural produce should also raise rural living standards, and lead to higher

marriage rates.

To test this hypothesis, a data set was prepared which included

annual marriage rates for the original Prussian provinces (1816 bound-

aries), and average prices for each province from the oYcial price

surveys. The data set ran from 1816 to 1875. This made it possible to

include wage data from a survey of day-labourer rates in the Prussian

state forests for each region, which should be a good guide to rural

wage rates. The wage Wgures refer to decades. It would have been

desirable to have had annual wage data, but a source of this type is not

available for the period. The analysis runs for a period when Prussia

was still a predominantly rural society.

The data set was analysed as a panel data set; the results are given in

Table 5E.1. The Wrst two columns give the results when a random

eVects model is used; the second two when the analysis uses a Wxed

eVects model (with dummies for each province). The two sets of results

are very close, indicating that the random eVects results can be regarded

as secure.

57 Hajnal (1965).58 Infanticide was one means employed: Lee and Wang (1999), Hanley and Yamamura

(1977).


The eVect from wage levels is weak and not secure in statistical

terms. The introduction of a time trend (columns a and c) eliminates

the eVect of this variable. Butter prices were also found to have little

eVect on marriage rates. The other price variables were associated with

signiWcant changes in the marriage rate.

There are three price series which have negative coeYcients: rye,

potatoes, and pork (although the result for pork is not a robust one).

These were the basic items of consumption for the poorer sections of

the agricultural population, and the less well-oV urban dwellers as well.

High prices for these commodities hit the living standards of the agri-

cultural workers and led to a postponement of marriage plans. The

eVect of rye prices is a particularly strong one.

There are two price series which are positively linked to marriage

rates: those for beef and wheat. These are items which did not form a

major part of the diet of the agricultural labourers and their families,

but which were mainly consumed by the better-oV in the countryside,

or sent to urban markets. High prices for these commodities were an

indicator of high urban demand, and a general rise in prosperity. As the

agricultural sector exported these products to the cities, while consum-

ing relatively little itself, high prices raised rural incomes. So, the posi-

tive link with marriage rates shows that rising rural prosperity was a

factor which brought forward marriages. It can also be argued that high

prices for these ‘luxury goods’ were indicators of rising urban wages,

which may have led to increased marriage rates in urban areas.

In general these results support the view that marriage in Prussia in

the pre-industrial phase and the early stages of industrialization was

inXuenced by economic circumstances, particularly in the rural sector.

When conditions were favourable, leading to gains by the rural sector,

then the marriage rate rose. More couples were able to aVord the cost of

setting up a household on their own. But when times were hard, then

this was put oV.This analysis leaves open the question whether regions which had

poor economic conditions had permanently low marriage rates. The

fact that no satisfactory result is obtained from the wage data argues for

the view that there was no permanent eVect; marriages were brought

forward or put oV according to circumstances, but the rate was not kept

down over the long term in regions where wages were low. The price

data might obscure this eVect if poorer regions had relatively high

prices for rye and potatoes but relatively low prices for beef and wheat,

though there is no evidence that this was in fact the case. Against this,


the econometric results for the Kreise given in Section 5D(i) do show a

positive eVect from rural wages on marriage rates in 1895, so it may be

that the analysis reported in this section is too aggregated to Wnd this

eVect. It might be found at the Kreis level but not at the provincial

level.


6

Migration, Farm Size, and the

Condition of the Agricultural Labourer

1. Introduction

One theme which ran very strongly through contemporary writings on

migration, particularly migration out of eastern agriculture, was that

there were institutional problems east of the Elbe which caused the

high rates of migration. These were particularly related to the preva-

lence of large estates, and the contractual labour system practised on

these estates. The property-owning peasant farmers typical of other

regions were, it was thought, less likely to leave. The landless east-

Elbian agricultural workers were drawn to the cities, leaving empty

dwellings and uncultivated Welds behind them. Or, even more seriously,

they were replaced by Polish workers, which threatened the population

balance in the disputed eastern provinces which Prussia had gained as a

result of the partitions of Poland in the eighteenth century.

Many of these studies expressed the view that there was something

wrong with agriculture east of the Elbe, and this feeling has tended to be

reproduced in modern historical studies: the east was backward, bound by

tradition, hostile to modernity, and so on.1 This chapter and the next one

question some of these stereotypes. The theme of this chapter is the insti-

tutional structure of east-Elbian agriculture, and the part this played in

migration out of the rural east. The next chapter considers the techno-

logical transformation of eastern agriculture in the late nineteenth century.

2. A Modern Perspective on the Institutions

of Eastern Agriculture

Adam Smith and the classical economists of the eighteenth and

early nineteenth centuries were hostile to share-cropping and other

contractual systems which deviated from the employer–wage labourer

1 A work published in 1898 described large eastern estates as the principal obstacles tothe establishment of a ‘just society’ and a ‘modern exchange economy’, Oppenheimer(1898), 9. See Frauendorfer (1957), i. 407–8 for a summary of Oppenheimer’s work.

or landlord–tenant norms. The reason for this was the eVect that such

institutions had on the incentives faced by the share-cropper or tenant

farmer: a share-cropper would only receive a proportion (typically 50%

or less) of the return to additional eVort, or to investments in new

productive knowledge. By contrast, a tenant farmer who paid a Wxed

money rent would receive the entire reward, which should lead to the

adoption of more progressive techniques, greater eVort on the part of

the farmer, and higher output.

The classical economists also favoured the use of cash payments: the

certainty provided by wages paid in a unit of known and stable value

would be preferable to the uncertainty of payments in kind, or payment

in the form of access to land or other beneWts. It would take extreme

conditions, where the value of a currency was uncertain or changing

rapidly (such as a period of hyper-inXation), to persuade rational and

well-informed economic agents to give up the beneWts of money and

return to barter and other forms of non-monetary exchange.

From this perspective, the traditional institutions of eastern agricul-

ture were not conducive to the rational pursuit of economic gain. An

example of payments and beneWts provided to a Deputat household on

an eastern estate in the 1800s included:

free accommodation and heating;1⁄2 Morgen fruit or vegetable garden;

1 Morgen potato land (fertilized and cultivated);

housing and fodder for one cow, plus a calf every other year;

the possibility of having hens or geese, or 1–2 sheep;

permission to keep a pig;

68 ScheVel of rye and other cereals for each adult male worker and

30 ScheVel for each female worker.2

As the nineteenth century progressed, there was a gradual shift to

cash payments, but, even in the years just before the First World War,

payments in kind remained an important source of income for eastern

agricultural workers. Table 6.1 gives income sources for households in

Wve eastern regions (East Prussia, Pomerania, Brandenburg, Silesia, and

Mecklenburg-Schwerin) compared to the average for all regions.3 This

shows that eastern workers received 60% of total household income in

2 Finck von Finckenstein (1960), 198.3 Averaging all the eastern regions obscures the fact that there were considerable diVer-

ences between the eastern provinces—cash payments were much more common in Silesia,for example. See Wygodzinski (1917), 26 for a more detailed analysis of payment systems.

182 Migration, Farm Size, Condition of the Labourer

the form of harvest shares and other payments in kind, considerably

more than the national average.

EVectively, the eastern worker and his family had two roles: the

household practised smallholder production on the land supplied by the

landlord; but the household was also a small contractor, who cultivated

and harvested the estate Welds, and looked after estate livestock in

return for a share in production, as well as the land allocated to the

household (and some cash payments).

The critical view of share-cropping which can be found in the work

of the classical economists was widely accepted by economists and other

social scientists until the 1950s and 1960s, and formed a part of a

generally dismissive attitude to the institutions and values of traditional

rural society. But from the 1970s onwards a series of studies have

appeared which have taken a more positive stance. Although the main

focus has been on the advantages and disadvantages of share-cropping,

the analysis has broader implications for the study of agricultural insti-

tutions. The authors responsible for this new approach have been

working mainly in the Welds of development economics and principal-

agent theory; but their insights can be applied to the historical study of

east-Elbian agriculture.

The essence of this approach lies in the concepts of ‘monitoring

costs’ and ‘missing markets’, particularly for credit and insurance. The

monitoring of a worker’s eVort and diligence is costly for employers,

and this is particularly the case where the workplace is a diVuse one

Table 6.1. Sources of income of workers in eastern agriculture,

1910–14 (in Marks), compared with the German average

East of the Elbe All Germany

Value of house plus Wrewood 127 103

Value of household production using

land supplied by landlord

138 94

Value of harvest shares 210 124

Other payments in kind 29 34

Other beneWts 18 13

Cash payments 338 501

Total household income 860 870

Cash payments as % of total 40.5 57.6

Source: Averages calculated from budgets given in Mittler (1929), 210–15.

Migration, Farm Size, Condition of the Labourer 183

(like an eastern estate) rather than concentrated (as in a modern fac-

tory). But the obvious answer to this would be to make the workers into

independent tenant farmers paying a Wxed rent, who would then have

every incentive to monitor their own activities as they would gain the

full rewards from additional eVort or increased diligence.

What is the reason why this option is not taken? One possibility lies

in the economies of scale available to large holdings, which may have

access to advanced machinery and management skills which are not

available to smaller farmers. In this case the ‘missing market’ is the one

which might allow other farmers to use such factors. Another possible

reason is that credit markets are closed to smaller farmers, or that credit

is prohibitively expensive. A third reason may be that smaller farmers

wish to have some insurance against crop failures and other disasters,

and, in the absence of formal insurance markets, they may be obliged to

enter into contractual arrangements with landlords which provide some

element of support in bad years. Finally, the costs of getting goods to

market may be so high that only larger estates can run the risk of

producing marketable crops.

Recent theory predicts that, in circumstances where there are missing

markets and high monitoring costs, then crop-sharing contracts will be

eYcient solutions to these problems. They combine an incentive elem-

ent to promote self-monitoring and higher eVort, and also provide solu-

tions to the problems which hinder smaller farmers’ access to product

markets, credit, insurance, and modern technology.4

The situation in east-Elbian agriculture in the early part of the nine-

teenth century was one which combined many features of the various

‘missing markets’ hypotheses: credit was not easy to obtain, crop yields

and prices were highly variable but insurance against these risks was not

available, new technologies such as sugar beet cultivation were being

introduced which favoured larger estates, urban markets were distant

and diYcult to predict, and transport was costly.5

4 Articles which set out the modern approach to agricultural contracts include Stiglitz(1974) and Newbery (1977). Bardhan (1989) provides an overview of the various ap-proaches. Two articles which were particularly useful in the preparation of this sectionwere Newbery (1989) and LaVont and Matoussi (1995). Newbery deals with the eVect thatmarketing costs have on the optimal choice of contract, and argues that a mixture of arental contract and a wage labour contract might provide an eYcient solution. The insti-tutional arrangements found in east-Elbian agriculture provided such a combination.LaVont and Matoussi analyse the eVect that missing credit markets have on the optimalchoice of contract.

5 The analysis of the importance of marketing costs follows Dasgupta (1993), 232: ‘TheWxed rental system is particularly ineYcient . . . where information about market conditions


The contractual system of eastern agriculture was well suited to these

circumstances. It provided a guaranteed amount of work for the farm-

worker and his family at harvest time, which protected them against

Xuctuations in labour demand. Although payments by harvest share

meant that they were exposed to these Xuctuations, the crops grown on

the household smallholding provided an alternative source of food.

Access to buildings and fodder for livestock reduced the capital cost of

keeping a cow and other animals.

The main disadvantage from the viewpoint of the farmworker was

that it only provided full employment at harvest time (when the system

often required the labourer to provide additional Scharwerker). It could

be described as a form of institutionalized underemployment: in oV-peak

periods the family occupied itself as best as possible on the household

smallholding or in other tasks. This was equally one of the principal

advantages from the landlord’s point of view: there was no requirement

to provide tasks to keep the estate workers busy during the winter.

The second advantage for the land-owner was that it economized on

cash outgoings. Although eastern estate owners had advantages over

small farmers in that they had better access to credit, and that they

were also better placed to face the costs and the risks of bringing crops

to market, they were also credit-constrained even if at a rather higher

level, and marketing costs were still high. There could be years when

market prices would not cover the costs of bringing in the crops from

remote estates. The normal practice in such years was to feed grain not

required for human consumption to livestock. This would not produce

any income until the animals were sold. EVectively, by making pay-

ments in kind, part of the risk associated with price Xuctuations was

transferred to the estate workers, the value of whose ‘wages’ would

Xuctuate with price changes. But as most of their income was spent on

food anyway (or would have been if they had been paid money wages),

this was not such a serious disadvantage.

Industrialization promotes institutional change in rural society partly

because it provides solutions to many of these ‘missing market’

dilemmas. The spread of banks and other credit institutions eventually

makes it possible for small farmers to borrow, and also to save (as a form

of self-insurance). Improved employment opportunities can also reduce

the need for insurance: if work on construction sites, for example, can be

is of importance in generating farm proWts, and where the tenant does not possess thisinformation to any great extent.’


found in times when agricultural incomes are low then it would be less

necessary to accept disadvantageous labour contracts which provide an

element of insurance. As urban centres grow, and transport links are

improved, market opportunities improve, integrating markets which

were previously separated, and thus reducing price variability.

Institutional arrangements such as the contractual relationships of

eastern agriculture form a part of a wider system of values and arrange-

ments, behavioural norms and beliefs, which characterize pre-industrial

rural societies. Revenue-sharing norms can support high levels of under-

employment, causing families to share resources with other members of

an extended kinship network, and with other villagers. In traditional

societies revenue-sharing and other forms of reciprocal behaviour can be

substitutes for missing markets in credit and insurance. Industrialization

will tend to alter these attitudes, mainly because wider employment

opportunities will reduce the need for insurance mechanisms. The move

to increased mobility will damage support mechanisms which rely on a

close-knit network of mutual obligations between diVerent kinship

groups. Partha Dasgupta has written about this:

Costs of mobility involve those associated with transportation, but they involve

many other things besides. Markets being thin, the outside world presents

considerable uncertainties to the individual. Life in the village may be harsh,

but there is an element of security provided by the family and the community.

This is sustained by norms of behaviour by which people abide. You can’t

simply pack up and leave for better prospects in a neighbouring village. For one

thing, you may not be welcome; for another, your departure would be seen as

an unsocial act, and this could inXict costs on your kin through a reduction of

the support system your village provides for them.6

The implications of this are, Wrst, that the cost of migration includes

not only the cost to the individual of giving up the support available in

the home village, but also the consequences for those family members

that remain of this ‘unsocial’ act. But, secondly, the system of reci-

procity itself will be at risk once there is signiWcant out-migration. If

families with members who migrate are perceived as having withdrawn

from a system of mutual obligation, then the system will cease to be one

which involves the whole community.7 While it might survive a few

6 Dasgupta (1993), 234.7 Dasgupta does not draw this conclusion, but it is a logical consequence of the system

he describes. The system can, however, survive considerable seasonal or short-termmigration.


defections, there will be a critical level below which it will no longer

function. Once this happens, the cost of migration will fall dramatically,

since migrants will no longer have to set the loss of this insurance

beneWt against the gains from higher urban wages.

In this analysis, as in Max Weber’s, one consequence of industrializa-

tion is the erosion of traditional relationships in rural society. Initially,

this may be a slow process, but the system could collapse quite quickly

once migration is established. These traditional relationships confer a

beneWt on those living in rural areas which is additional to the income

received from farming (or other activities). The consequence of the

removal of this beneWt will be the release of surplus labour onto urban

labour markets.

Dasgupta also provides an analysis of the eVect of eYciency wages on

rural unemployment. The model is a simple one: workers are either

permanently employed or employed only at peak periods or perman-

ently unemployed. Those without work are obliged to feed themselves

using common-property resources. Labour eVort is related to nutri-

tional status, therefore land-owners pay eYciency wage rates which

exceed the market-clearing rate, so there is involuntary unemployment.

If there is a reduction in the rural population through out-migration,

then the Wrst consequence is an improvement in the average amount of

nutrition available from the common Welds and other common

resources. This in turn provides an improved nutritional carryover for

those who go on to work during the peak period. In consequence, land-

owners are able to cut peak period eYciency wages, resulting in

increased employment and a general rise in production. As Dasgupta

puts it: ‘agricultural output would be expected to increase were popula-

tion to decline in size. This is surplus labour with a vengeance.’8

More generally, the existence of eYciency wage rates either in indus-

try or agriculture will tend to create unemployment if wages are thereby

raised above market-clearing levels. In modern economies most of this

will be open and recorded, due to the availability of beneWts. In the

nineteenth century, when the availability of beneWts was restricted or

non-existent, underemployment was hidden; people scraped by on the

occasional casual job, joined the informal urban sector, went into

domestic service, or returned to family farms. The size of this pool of

underemployed labour is not easy to estimate, and much will depend on

8 Dasgupta (1993), 509 and 518–23.


the deWnition of underemployment. Nevertheless, the reabsorption of

discouraged or temporarily unemployed urban workers was one way in

which agriculture helped to stabilize urban labour markets. In this case,

the source of the surplus may lie as much in the operation of the urban

labour market as in the state of agriculture.

3. Contemporary works on M igration

out of Eastern Agriculture

There are a number of distinctive features of Germany’s position which

aVected contemporary attitudes to migration.9 These included:

(a) Germany’s geographical position: Germany was not an island, but

linked to the labour markets of central and eastern Europe. Migra-

tion could Xow across Germany’s eastern borders relatively easily.

(b) Political considerations: the size of the rural vote was an import-

ant factor in the political structure of imperial Germany, and

those who sought to maintain it had positions of inXuence or

authority within the system.

(c) The strength of German socialism: from an early stage Germany

had a well-organized social democratic party, which polled over

50% of the vote in three German cities in the Reichstag elections

of 1877, and nearly 40% in Berlin in that year.

(d) The British example: Germans had the opportunity to study

conditions in British industrial cities. Many found this an

alarming prospect for Germany: something to be avoided, or

improved upon, if possible.

(e) Ethnic composition: parts of the German east contained substantial

numbers of Poles and Lithuanians, and the departure of German-

speaking migrants from such areas had implications for the govern-

ment’s policy of strengthening the German identity of these regions.

Attitudes to migration were certainly inXuenced by the social prob-

lems associated with urbanization. Adolf Weber, writing in 1908, wasable to use oYcial statistics to demonstrate that crime was higher in

cities, as were divorce, suicide, adultery, the incidence of venereal dis-

ease, and the number of children on welfare, while church attendance

was lower. He added a number of other charges: city art was ostenta-

9 Haberle (1938) surveys German studies before 1938.


tious, there was less patriotism, and city life hindered the moral and

spiritual progress of the nation. Despite this, Weber did not condemn

all aspects of city life: he thought that productivity was higher and

educational opportunities were better.10 Most serious commentators

recognized that urbanization had its beneWts, and that it was an inevit-

able consequence of economic growth.

So, although the debate over the question ‘Agrar- oder Industriestaat’attracted attention in the 1890s, in practice no serious eVort was made

to hold back urbanization or the growth of industry. The most that it

was hoped to achieve was that Germany should remain both ‘Agrar-

und Industriestaat’, to use the title of a work by one participant in the

debate, the historical economist Adolf Wagner.11 It was also considered

to be important to preserve the Mittelstand, and peasant proprietors

constituted a large part of this class. This led to a concern with the

interests of small and medium-sized farmers: as Gustav Schmoller put

it, ‘the issue of the preservation of the position of the peasantry involves

our entire future as a society’.12

This piece of rhetoric requires some elucidation. Schmoller feared

that industrialization would lead to the concentration of property, and

the polarization of society between a propertyless working class and a

rich capitalist minority, just as Marx had predicted. Membership of the

Mittelstand was deWned not just by income level but also by the owner-

ship of at least a small amount of property. Gustav Knapp, speaking

after Schmoller in the same debate, praised labour relations in agricul-

ture in western Germany: the Heuerlingen of this region had some

animals of their own, long-term contracts with their employers (them-

selves not aristocratic land-owners but larger peasant farmers), and

worked ‘shoulder to shoulder with the employer’.13 The consequence

was that social democracy had made no progress in this area.14

Migration transferred agricultural workers into a very diVerent soci-

ety. In large-scale industry there was little personal contact between

employers and employees; employment was insecure and contracts were

10 Weber (1908). Adolf Weber (1876–1963), economist, posts included chairs at theuniversities of Cologne, Breslau, Frankfurt, and Munich.

11 Wagner (1902); the debate is described in Lebovics (1967) and Barkin (1970). AdolfWagner (1835–1917), economist, professor at Berlin university, leading member of thehistorical school, and one of the founders of the Verein fur Sozialpolitik. He representedthe Christian Social Party in the Prussian State Assembly 1882–5.

12 Verein fur Sozialpolitik (1893), 2.13 Ibid. 12.14 Ibid. 29.


short-term; workers had little property of their own. This was fertile

ground for socialism.

The concern of the historical economists with the polarization of

society and the decline of the Mittelstand was intensiWed by a series of

studies of Prussian tax returns which appeared to show that inequality

was widening. The Wrst of these, by Engel, appeared in 1868. Some of

the subsequent studies were produced by members of the Verein furSozialpolitik; Schmoller and Wagner were amongst those who worked

on these estimates.15

These studies led to a debate about the reasons why inequality was

widening. Adolf Wagner thought that the cause was technological pro-

gress favourable to large enterprises. But the essence of the historical

school’s analysis of this, and other, economic problems was to stress the

importance of the historical context: society did not move along prede-

termined lines. Decisions were made which aVected the outcome. They

held the view that the state could bring about the reconciliation of

capital and labour, both by direct action and by encouraging favourable

institutional developments.16

Some made the same connection between rural conditions, migration,

and urban labour markets which lies behind the Lewis Model. Max

Sering, in a report presented to the Verein fur Sozialpolitik in 1893,

blamed excessive migration for depopulating rural areas in the east, as

well as holding down the wages of industrial workers:

while the cities and industrial districts are accumulating an ever larger reserve

army, which rarely Wnds full employment, even in times of maximum economic

activity, and which is forcing down the living standards of other workers, the

wide Welds of the eastern arable districts are short of labour . . . 17

Interesting points in this passage include the use of the term ‘reserve

army’, familiar from Marxist terminology.

15 Engel (1868 and 1875); Schmoller (1895) and Wagner (1904); there had been atradition of concern about the distribution of income in German political economy, goingback at least as far as von Thunen (1826), whose views are given in Pogge von Strand-mann (1999). Ernst Engel (1821–96), statistician, director of the Prussian Statistical OYce1860–82, represented the National Liberal Party in the Prussian State Assembly, 1867–70,Neue Deutsche Biographie, iv. 500–1.

16 Wagner (1871) and Schmoller (1900); also HelVerich (1915); the views of the histor-ical economists on income distribution are discussed in Dumke (1991) and Grimmer(1999); other writers considered ways to reconcile the interests of capital and labour,amongst them Walter Rathenau, Pogge von Strandmann (1988a). Karl HelVerich was adirector of the Deutsche Bank and secretary of the Treasury during the First World War.

17 Sering (1893), 7.


For a group of writers on agricultural matters, the extent of migra-

tion out of agriculture was far greater that could be justiWed by the

diVerence between urban and rural wages. This group, which included

Knapp, Max Weber, and Sering, blamed developments within agricul-

ture.18 Sering noted that in 1885–90 the eastern provinces had lost

640,000 people through migration, 75% of the natural increase in the

population. This was an indication of a profound social malaise: ‘it

suggests that all social classes had a feeling of profound dissatisfaction,

that there was a sickness of the social organism itself’.19 The view that

migration was a symptom of some major problem within rural society

was an inXuential one, and led to schemes of ‘internal colonization’: the

settlement of landless rural workers, or disaVected urban workers, in

colonies of land-owning peasant farmers.20

Migration from rural areas into German cities was a matter of con-

cern to a range of opinion in Germany. A number of authors took the

view that it was caused by a deterioration in agricultural conditions, in

particular by changes in the position of landless agricultural workers.

One strand of this analysis concerned the eVect of the agrarian

reforms of the early nineteenth century. In 1873 Theodor von der

Goltz pointed out that migration was highest from areas of Großgrund-

besitz—the large east-Elbian estates—and unknown in areas dominated

by smaller peasant holdings. In 1887 Gustav Knapp argued that the

reforms had led to a loss of traditional rights and increased the number

of landless labourers:21 In Knapp’s view it was the poverty and lack of

freedom of the eastern agricultural worker which was responsible for

out-migration.

In the western provinces, where there is often serious poverty amongst small

farmers as a result of the splitting of holdings, there are certainly poor agricul-

tural workers, but nobody who feels to the same extent the lack of freedom in

every aspect of their lives. . . . The poorest forestry worker in the Black Forest,

the most down-trodden hay-maker in Kanton Uri, have a higher station in

society than the Insten in our wide eastern provinces.

18 Knapp (1887), Weber (1893 and 1894): also von der Goltz (1875). G. F. Knapp(1842–1926), economist and statistician, professor of statistics at Leipzig and Strasburg,NDB xii. 152. T. A. von der Goltz (1836–1905), farmer and agricultural historian, heldchairs of agriculture at Konigsberg and Jena, NDB vi. 635–6. See Frauendorfer (1957),i. 392–4, 394–6 and 436–7 for the background to this debate.

19 Sering (1893), 7.20 These schemes also served nationalist aims, Grimmer (1999), 293–4.21 von der Goltz (1875), Knapp (1887 and 1891).


Knapp blamed the operation of the Stein-Hardenburg laws for the

original loss of land and status by the agricultural workers east of

the Elbe. But he also thought that their position was deteriorating as

the land-owners intensiWed their demands for labour to work the estate

Welds, thus reducing the time the labourers had for their own small-

holdings:

The aristocratic landlords want to have all the labourers’ time for their own

enterprises . . .Where this development has been completed . . . the worker on an

estate will be turned into a state of utter dependency . . . he is no longer a true

Inst: he has no house, he has no arable acres of his own, he no longer has a cow.

These early studies had an eVect on opinion. The view that large estates

were predominantly to blame for migration out of agriculture became

widely accepted. A study published in 1930 commented: ‘It is a com-

monly held view that the large landholdings considerably increase mi-

gration out of agriculture.’22

The debate produced three major surveys: one of the conditions of

the peasantry in 1883, another of the conditions of agricultural labour-

ers in 1890–1, and a survey of Prussian inheritance systems in 1899–

1910. These are valuable surveys of conditions in German agriculture in

the period.23 There are, however, some diYculties in using the infor-

mation they provide. They are, in eVect, collections of articles, which

reXect the concerns and interests of individual authors. Data is not

generally provided on a comparable basis.24

The 1890–1 survey was in three parts. The third of these, covering

Germany east of the Elbe, was written by Max Weber. It was his

second major publication. His contribution is useful for a number of

reasons. It was more systematic than the others. It includes useful data

on wage rates and payments in kind, and attempts to link these Wgures

to those produced by earlier surveys. And Weber attempted to provide

an explanation of the process of social change revealed by the survey.

22 Golding (1930), 214. See Oppenheimer (1898) for a particularly forceful presentationof this view.

23 These were published in the Schriften des Verein fur Sozialpolitik (1883), vols. 22–4and (1892–3), vols. 53–5, and in Sering (1899–1910).

24 A comment which reXects the frustrations of the modern statistician, faced withwhat could have been immensely valuable sources of data. There are some very goodarticles in these surveys: one example is the article by Paasche on Mecklenburg-Schwerin,Paasche (1883).


4. Max weber and the Condition of the

Agricultural Labourer in the East

Of all Weber’s major works this is probably the least known. It remains

untranslated. Yet it is a thorough survey of conditions in eastern agri-

culture, a valuable source of material, and one of the earliest works to

consider the question of a ‘labour surplus’ in traditional agriculture:

how is it created? under what conditions is it released?

The questionnaire results were used as a basis. Weber reported some

of the more reactionary views of the respondents without comment, as

for example the opinion that it was not desirable to educate the rural

population.25 But he also used the survey to express his own views on

the social relationships found in eastern agriculture. He developed this

analysis in an article in the 1894 Archiv fur Soziale Gesetzgebung und

Statistik, which also responded to some of the criticism made of the

original study at the 1892 Colloquium of the Verein fur Sozialpolitik. In

1908 he returned to the theme in a lecture in the United States.26

Weber followed Knapp in that he thought that the larger estates were

primarily responsible for out-migration. He saw the creation of the

labour surplus as a result of an alteration in social relationships on these

estates. Paternalist attitudes were replaced by a commercial, capitalist

approach. This in turn led to the introduction of more intensive arable

production, associated with sugar beet cultivation, and increased use of

mechanization. Contractual German workers were replaced by seasonal

Polish workers.

Weber’s attitude was inXuenced by political concerns, a desire to

maintain and, if possible, increase the German population in the dis-

puted lands of the Prussian east. He was a member of the nationalist

organization, the Ostmarkverein (also known as Hakatisten from the

initials of the organization’s founders), which argued for the compul-

sory purchase of Polish estates and the settlement of German farmers

on lands previously occupied by Poles.27 Weber supported the organiza-

tion’s aims. A newspaper report of a speech he gave in March 1893

went: ‘above all else, Herr Weber argued for the pursuit of a policy

25 Weber (1893), 197–8.26 Weber (1894 and 1948). See Honigsheim (1949) for an assessment of Weber’s work

in this area.27 See Hagen (1980) and also Galos et al. (1966) for accounts of the conXict between

Germans and Poles in the east. Schmoller and Adolf Weber were also members of theOstmarkverein, Hagen (1980), 269.


which served German interests, he was against the equal treatment of

Poles and Germans’.28 His criticism of the Junker estates should be

seen in this context. In the same month, in another speech, he said: ‘the

presence of large estates is the factor which is, currently, most respon-

sible for the increase in the Polish population in the East’.29

However, in the 1890s the concern over foreign agricultural workers

was premature. In 1907 there were 195,000 foreign-born agricultural

workers in Prussia, constituting 1.8% of the agricultural labour force:

1.0% in East Prussia, 1.6% in West Prussia, 1.3% in Posen, and 2.7%

in Pomerania. The largest concentration was in Prussian Saxony, where

they amounted to 4.8% of the population occupied in agriculture.30

This suggests that, up to this point, the main cause of migration from

outside Germany had been the high seasonal demand for labour in the

sugar beet Welds of central Germany.

It has been estimated that the number of foreign workers rose to over

a million by 1913 (including 356,000 employed in industry). This is

diYcult to check given the lack of an occupational census after 1907.

The Preubische Feldarbeiterzentrale issued a total of 412,000 permits in

1907–14, but, as Russian-born Poles, in particular, were obliged to

return home every year, it is possible that these were only used for a

few years on average.31

In general, permanently settled foreigners do not seem to have

replaced Germans to any considerable extent before 1907. There was

some displacement by temporary seasonal migrants, although this could

be interpreted as the replacement of German labourers who left volun-

tarily to seek better wages in the cities. This question is related to the

claim that the position of the agricultural labourers in the east was

deteriorating, socially or economically. In this respect there is a diVer-

ence between Knapp on the one hand, who saw in the persistence of

payments in kind and contractual labour services evidence of the

dependent status of the eastern workers, and explained migration as an

attempt to escape this, and Weber on the other, who took a rather more

favourable view of the traditional labour relationships.

Weber’s view was that the Inste and Deputate of eastern agriculture

had had the status of Kleinwirte or small contractors, bound individually

to the landlords in a system which was certainly one of subordination,

but which also emphasized common interests (through harvest shares

28 Weber (1993), 825. 29 Ibid. 179 30 Ibid. 489.31 Figures are from Wehler (1995), 546.


and access to communal grazing). The keeping of livestock by the

labourers was a crucial sign not only of material well-being, but also of

status and of ‘cultural-level’. The point here was that, in Weber’s view,

a reduced level of cow-keeping (such as Weber found in Silesia) was

generally associated with a higher level of female participation, in work

both oV and on the holding, and this was a backward step culturally:

‘the ability to keep, preferably, a milking cow, or at least some cattle, is

the most important precondition for a well-adjusted family life’.32

Weber held that this system of patriarchal labour relations had come

under threat from two directions. The Wrst was the shift to a ‘capitalist’

agriculture: more intensive use of the land and greater livestock produc-

tion. This had led to the reduction of the smallholdings given to

workers, and the division of the common grazing Welds: ‘the loss of

access to grazing on the village commons at the time of enclosure was

the most grievous blow that could have been struck at the property

owning labourer’.33 He described the reduction of the Masurian labour-

ers to the status of Milchdeputate (looking after the landlord’s cows

instead of their own) as ‘something which embittered social relations

and which destroyed the sense of having interests in common’.34

The second problem was the unwillingness of the children of agricul-

tural labourers to remain part of the system. The Instleute households

were required to provide two or three workers at harvest time, so the

participation of children was vital, especially as it became more diYcult

to Wnd substitutes (‘cripples, old men, women of loose morals, have

been oVered as such’).35 Weber did not believe that there was a connec-

tion between the migration of the children of agricultural labourers to

the cities and the level of wages. Indeed, he thought that migrants

experienced a decline in wages. In his view it was a somewhat perverse

desire for an illusion of freedom which led the children of eastern

agricultural labourers to prefer proletarianization and lower wages in

cities to the paternal system and higher ‘cultural level’ on oVer in

eastern agriculture: ‘according to the survey returns, the motive was a

desire for independence even if it could only be achieved at the cost of a

reduction in living standards’.36

It should be borne in mind that Weber was making these remarks

while reporting on the results of a survey of the condition of agricul-

tural labourers, conducted by means of a questionnaire sent out to

32 Weber (1893), 184. 33 Ibid. 34 Ibid.35 Ibid. 185. 36 Ibid. 193.


landlords. Weber had no other evidence to support the statement that

migrants out of agriculture did not improve their economic position

apart from the opinions expressed by these respondents.

The eVect of out-migration was damaging to the institutional struc-

ture of eastern agriculture as a result of its impact on the Scharwerker

system. If the Inste were unable to provide additional workers, then

they were not fulWlling their side of the bargain which led to the land-

owner oVering a smallholding in return:

The farmer’s own children move away, at the latest, after completing military

service, though many leave earlier, so that they can devote themselves to life in

the cities as soon as possible, without considering whether this will bring about

the collapse of the family business. As a result, the Inste are no longer able to

fulWl a contractual obligation to provide three workers to the estate, as used to

be commonly required in East Prussia, many are unable to even provide two, if

their wives don’t work with them.37

The allegation that, as a result of the shift to capitalist values, labour-

ers had less opportunity to keep animals themselves is not supported by

the evidence, which shows continued cow-keeping on smallholdings,

although it was probably true in certain districts.38 However, the fact

that cow-keeping was lower in industrial districts like Silesia is not

necessarily an indicator that the position of the labourers was worse in

these areas. It could just as well show that better opportunities to supple-

ment agricultural earnings through industrial work made the keeping of

cows less attractive. Whether or not the increase in the keeping of live-

stock on smallholdings was suYcient to compensate for the undoubted

reduction in grazing rights on uncultivated lands, as these were enclosed

and brought into arable production, is a more open question.

The argument that changed values had brought about a deterioration

in the living standards of the agricultural workers is also not supported

by the evidence. On the contrary, there is good evidence (some pro-

vided by Weber himself) to show that agricultural wages were rising.

Weber supplied a mass of evidence on the conditions of employment of

37 Weber (1893), 185. Another factor which might have aVected the Scharwerkersystem was the enforcement of laws against child labour during school hours. Weber didnot mention this as a factor, but, according to other authors involved in the survey, thishad a signiWcant eVect on the use of child labour in some regions, Schriften des Vereins furSozialpolitik, 52, p. 332.

38 Between 1881 and 1895, the value of livestock kept by holdings of less than 2hectares rose by 16.2% and it rose by 17.4% on holdings of between 2 and 5 hectares; therise on holdings of over 100 hectares was only 9.5%, Rauchberg (1900), 590.


diVerent classes of agricultural labourers, and comparisons with the

position in 1849 and 1873 (the latter date uses von der Goltz’s Wgures).

Where a high proportion of the payments are in kind or through har-

vest shares or the allocation of smallholdings the comparisons are hard

to make. There seems to have been an increase in cash payments to

groups traditionally paid in kind. But the value of these other beneWts

may have altered, and there is some evidence that the labour contribu-

tions demanded by landlords had risen.

However, comparisons are possible for those workers paid in cash

ohne Kost (without provision of board and lodging). There are a number

of localities where Wgures are available for 1849 and 1892 and these have

been brought together to form Table 6.2. What this table shows is a

very considerable improvement in the wages paid to agricultural

workers in the east over the period. The last line gives indices which

show wages rising by a factor of 2¼–21⁄2 1849–92.This is important for a number of reasons. First, the suggestion that

eastern agricultural workers experienced declining wage rates is encoun-

tered in some publications.39 This does not appear to be the case.

Secondly, it puts the problems of eastern grain producers in a new

perspective. Between 1870/4 and 1890/4 arable producer prices fell by

10.3%, while this table shows wages rising by 34–35%. Producers were

being squeezed from both directions.

5. M igrat ion , Farm S i ze and Agricultural Techniques

The regressions in the previous chapter showed that the apparent rela-

tionship between farm size and migration was reversed once allowance

was made for demographic factors. Large farms contributed to out-

migration indirectly, through the positive relationship between farm

size and the natural rate of population increase. There was no tendency

for the quit rate to be higher on larger holdings once the eVect of high

population increase was removed. Indeed, the quit rate was shown to be

lower on larger holdings.

The data set used in this analysis made use of a 1933 study of net

migration in the Prussian Kreise from 1882 to 1895. In this, an indirect

estimate of the quit rate for the agricultural population was produced

39 Dickler (1975), for example.


Table 6.2. Daily earnings in Marks of ‘free’ agricultural day-labourers (ohne

Kost) in full-time employment where possible (earnings of those who are only

temporarily employed are in bold type)

Summer wages Winter wages

1849 1873 1892 1849 1873 1892

East Prussia: Gumbinnen

Ragnit 0.5/0.6 0.85 1.0/1.8 3/0.5 0.62 0.8/1.2

Gumbinnen 0.5/0.6 0.85 1.2/1.4 0.3/0.4 0.62 0.8/1.0

Insterburg 0.6/0.7 1.55 1.5/2.0 0.4 0.66 1.0/1.2

Sensburg 0.5/0.6 1.1 1.0/1.6 0.4 0.57 0.8/1.2

East Prussia: Konigsberg

Mohrungen 0.5/0.6 1.23 1.4/2.0 0.4/0.5 0.73 0.9/1.2

Rossel; Allenstein;

Ortelsburg

0.5/0.6 1.25/1.3 1.0/1.5 – 0.7/0.81 0.7/1.0

West Prussia: Marienwerder

Stuhm 0.8/1.0 2.03 1.25/2.5 – 1.41 –

Rosenberg 0.8 – 1.25/1.75 0.4 – 1.0/1.2

West Prussia: Danzig

Stargard 1.0/1.5 1.62 2.75/4.0 – 0.92 1.0/1.5

Putzig 0.75 – 1.0/1.75 0.5 – 0.75/1.0

Pommern: Koslin

Schlawe;

Rummelsberg; Stolp

0.8 0.6 – – 1.5/3.0 0.8/1.25

Pommern: Stettin

Anklam 1.0 1.68 1.5/2.0 0.62 1.0 1.25/1.75

Greisenberg 0.75 1.53 1.5/2.0 0.6 1.31 1.2

Pommern: Stralsund

Greiswald 1.0/1.2 2.47 3.0 – 1.18 –

Silesia: Oppeln

Neustadt 0.5 1.38 2.0 0.4 0.9 1.5

Silesia: Liegnitz

Liegnitz 0.5 0.96 1.5 0.4 0.73 1.0

Brandenburg: Frankfurt

Lebus 0.6 1.5 1.5/2.0 – 0.87 1.0/1.25

Brandenburg: Potsdam

Barnim 1.0 1.56 2.1 0.75 1.06 1.2

Unweighted

averagesa0.74 1.37 1.83 0.52 0.87 1.17

as an index

(1849¼100)

100 185 247 100 167 225

aGaps in the data have been Wlled by extrapolation (such as using the average relationship betweensummer and winter wages for the year in question); ranges were entered as the mid-point values.


using a combination of results from the occupational censuses of these

years, and records of births and deaths by Kreis.

An alternative approach is to examine the movement of the agricul-

tural population directly using the results of the occupational censuses.

This eVectively excludes the inXuence of demographic factors by

looking only at the trend of agricultural employment. Implicitly, it

assumes that any demographic surplus will be exported, either to non-

agricultural sectors within the region, or to other regions. The analysis

presented in the previous chapter showed this to be a relatively well-

founded assumption.

The data set prepared for this analysis covered a rather longer period,

1882 to 1907, and included 797 Kreis-level units, covering the whole of

Germany not just Prussia. The results reinforced the analysis in the

previous chapter. Although there is a simple negative relationship be-

tween the movement of the agricultural population and the presence of

large holdings, this is a spurious result. More sophisticated analysis

showed that the reason why the permanent agricultural population de-

clined in areas where there were larger farms and estates (or rose less

rapidly) is that these estates were situated in areas where the agricul-

tural population was declining for diVerent reasons: these were remote

areas, with poor access to urban markets, and they were situated in

regions close to the border with Russian Poland, which meant that

seasonal Polish workers could be used.40

The results of the statistical analysis are summarized in Table 6.3.

The most important inXuences on the movement of the agricultural

labour force were the level of non-agricultural employment within the

Kreis in 1882—there tended to be a decline in the more agricultural

regions—and geographical location. In purely agricultural areas the pre-

dicted level of agricultural employment was static, meaning that any

demographic surplus would be exported, while in the areas which had

high levels of non-agricultural employment, agricultural employment

rose by 35–40% (37% for a region which was 80% non-agricultural in

1882). In these regions, quite substantial rates of population increase

could be accommodated within the agricultural sector.

Remoter regions tended to maintain a higher level of agricultural

employment. The predicted level of employment in regions 100 kilo-

metres from major cities (over 100,000 inhabitants) is 7% higher in

1907 compared to a region with a similar starting point in 1882, but

40 The full details of this analysis were originally given in chapter 4 of Grant (2000)and have been published separately in Grant (2002c).


which was 20 kilometres from a major city. This suggests that regions

closer to major cities were quicker to shed surplus labour.

There was also a major east–west division: the agricultural population

declined fastest, or rose less rapidly, in the rural east. Table 6.4 illus-

trates this for two hypothetical regions: Region A with 60% of land in

holdings of over 100 hectares, Region B with only 10%. Both have 65%

of the population initially occupied in agriculture, and 75 kilometres

between their mid-point and the nearest large city.

This table shows the importance of location. In areas close to the

border there is little predicted increase in the agricultural population.

The average German rate of natural population increase of 1.3%

(1871–1913) implies an increase of 38% over 25 years. The rates in

Table 6.3. Summary of results of statistical analysis of the change

in the agricultural population 1882–1907 in 793 German Kreise


Factors which are associated with a less rapid rise in the population occupied in

agriculture or with a decrease in the agricultural population:

A high rate of agricultural employment in 1882

A situation close to the border with Russian Poland

Factors which are associated with a less rapid decrease in the population occu-

pied in agriculture or with an increase in the agricultural population:

The presence of large farms and estates

Being a long distance from large cities (over 100,000 inhabitants)

Having a high level of sugar beet production in 1907

Having high use of mechanized threshing in 1907

Table 6.4. Predicted change in the agricultural population over

25 years for two hypothetical regions

Distance from the border with Russian Poland in km

50 100 200 500

Predicted change in the agricultural population over 25 years

A 60% in

large farms

þ3.5 þ6.5 þ12.4 þ30.3

B 10% in

large farms

þ3.3 þ4.8 þ7.7 þ16.6


some eastern provinces were substantially higher (1.6% per annum in

West Prussia, 1.7% in Posen). In consequence, the anticipated rate of

out-migration would be high from all the agricultural regions close to

the border. But at a distance of 500 kilometres (approximately the

German average) the position changes. Here the larger estates increased

employment opportunities more than small or medium-sized farms did,

and so Region A is predicted to increase the agricultural population by

more than Region B. The implied rate of out-migration from Region

A is substantially lower than from Region B.

The same analysis can also be used to examine the eVect of more

advanced agricultural techniques, such as the introduction of sugar beet

cultivation and also the use of labour-saving machinery such as steam-

threshing. The results given in Table 6.3 show that the intensiWcation

of arable production associated with the cultivation of sugar beet led to

an increase in employment opportunities: when controlling for other

factors the eVect of sugar beet cultivation is a positive one. The more

intensive agricultural crops such as sugar beet made increased demands

on the farm labour force, for weeding and harvesting, and also were

associated with increased livestock numbers (who could be fed on the

pulp left after the sugar had been extracted), which also raised labour

requirements.

Similarly, the use of steam-threshing was not associated with a

decline in employment opportunities, when this factor was considered

separately from other factors inXuencing agricultural employment.

Rather, the use of steam-threshing is associated with rising agricultural

employment. The larger estates did often use steam-threshing, but, as

already noted, these holdings were tending to increase employment for

other reasons. If steam-threshing did replace labour previously used to

thresh by hand (and this was the main advantage of the technology)

then it appears that these workers were redirected to other tasks. The

positive connection between the use of something which was quite

clearly a labour-saving technology, and the change in agricultural em-

ployment, suggests that this technology tended to be used mainly in

areas which were already short of labour.

6. Influences on Rural Wage Levels

It was a commonly held view that large holdings caused out-migration

partly because they paid low wages. A study published in 1907


commented: ‘the accusation has often been made against large farms

and estates that by their nature, as well as by paying poor wages, they

have created migration out of agriculture, or at least caused it to in-

crease’.41

If so, it should be possible to Wnd this eVect by examining variations

in rural wage levels. These can be studied making use of the surveys

conducted to provide information for the Reichsversicherungsamt. Theaccident insurance legislation required that the Ortsublicher Tagelohn

should be recorded for each locality. The Prussian surveys are particu-

larly useful in that wage rates in rural areas are distinguished from

those in towns. The data was used to construct estimates of average

rural wages for each Kreis, combining the 1892 and 1901 surveys. The

full statistical results are given in Appendix 6B, Table 6B.1; Table 6.5

provides a summary of the main Wndings.

The main inXuences on wage levels came from geographical location

and from the costs of living in the locality: wages were higher in areas

where food prices were high. The impact of the border with Russian

Poland was also an important factor. Wage rates were substantially

lower in Russian Poland: male day-labourer rates were 0.52 Marks in

Russian Poland in 1890, compared to 1.13 Marks in a German border

region such as Gumbinnen (in 1892), and 1.66 Marks in a more pros-

Table 6.5. Summary of results of statistical analysis of rural

wages levels in 392 Prussian Kreise 1892/1901


Factors which are associated with a higher level of rural wages:

High food prices

A situation further from the border with Russian Poland

Factors which are associated with lower wage levels:


A high level of household employment in 1882

Having a high level of sugar beet production in 1897


Factors which are associated with a higher level of rural wages:

The presence of large farms and estates

The proximity of large towns and cities (over 50,000 inhabitants)

41 Broesike (1907), 22.


perous central region such as Magdeburg.42 The equivalent Wgures for

female day-labourers were 0.37, 0.71, and 1.01 Marks respectively. In

the areas closer to the border, the availability of cheap migrant labour

had a depressing eVect on German wage rates.

Table 6.6 gives the predicted levels of wages for rural areas at diVer-

ent distances from the border with Russian Poland.43 The strong eVect

of the border on rural wage levels is shown. The importance of non-

agricultural employment is also demonstrated. Wages were low in the

east mainly because of the proximity of the border and the relative

shortage of non-agricultural employment.

Once allowance is made for these factors, it can be shown that the

presence of large farms and estates did not lead to lower wages. On the

contrary there is a slight tendency for wages to be higher in areas where

there were signiWcant numbers of holdings of 100 hectares or more.

There is no justiWcation for the view that larger estates held down

wages.44 Wages were low in the rural east for other reasons.

Other factors which were associated with low wages include a high

level of household employment in 1882. Cottage industry tended to be

found in areas of high population pressure, and could therefore be an

Table 6.6. Predicted average daily wages in Pfennigs, in rural

areas, 1892–1901

Distance from the border with

Russian Poland (km)

50 100 200 500

Predicted level for a region with 20%

of the population occupied in agriculture

121.6 124.6 130.8 149.2

Predicted level for a region with 80%

of the population occupied in agriculture

98.2 101.2 107.4 125.8

42 Figures for Russian Poland are from Knoke (1911), 32, who obtained them from theWarsaw Statistical OYce and converted these into Marks; German data from Neuhaus(1904), 320–28.

43 The calculations use the coeYcients given in column a of Appendix Table 6B.1(other variables take average values).

44 The statistical result showing that large holdings tended to push up wages is not astrong one. However, the alternative hypothesis, that large estates paid signiWcantly lowerwages, can be rejected with a high degree of conWdence.


indicator of surplus labour conditions within the Kreis.45 The Wnding

that wage levels within agriculture were lower in regions with high

levels of household employment supports this view.

The statistical analysis also found an association between low wages

and sugar beet cultivation: wages were lower in Kreise with high areas

under sugar beet relative to other arable crops in 1897. This is a sur-

prising result when considered alongside the Wnding reported in the

previous section, that sugar beet cultivation led to an increase in agri-

cultural employment. It appears to support Weber’s contention that

there was a change in the behaviour of land-owners in market-

orientated agriculture, such as the areas practising intensive arable

farming associated with sugar beet cultivation. Sugar beet cultivation

raised the demand for labour but led to a fall in the price of labour, in

apparent contradiction of the basic laws of supply and demand.

The reason seems to lie in the relationship revealed by the import-

ance of the proximity of the border with Russian Poland: the availability

of migrant labour from Russian Poland was a major factor in wage

determination in German agriculture. Much of the eVect of migrant

labour is captured by the distance variable, as the most important factor

was the proximity of the frontier. However, in the sugar beet areas the

use of foreign migrant labour was established early, and this led to the

development of networks of agents who supplied labour on a regular

basis. It was therefore easier for farmers in these regions to obtain

migrants for sugar beet cultivation and other tasks. This had a depress-

ing eVect on wages.

7. D irect Evidence on Underemployment in Rural Areas

This section looks for evidence that underemployment in German agri-

culture was a factor in the relative decline of the rural population. Was

there a shift away from areas where there was underemployment? If so,

then this would support the concept of a pre-industrial rural labour

surplus, which is then released during industrialization.

The analysis reported in Section 5 made use of a data set covering all

the German rural Kreise. This section restricts the analysis to the Prus-

45 See for example the description of conditions in the Taunus uplands, in BauerlicherZustand in Deutschland (1883), i. 156–7, where the main cottage industry had been nail-making.


sian Kreise, which makes it possible to use other data sources, in par-

ticular the more detailed publication of the occupational census results

found in Prussian statistical sources. Table 6.7 provides a summary of

the Wndings; the full econometric results are given in Appendix 6B,

Table 6B.2.

The main results repeat those shown earlier when the full data set

was used: the agricultural population declined faster or rose least in

areas close to the border with Russian Poland, and in areas where there

was a high level of agricultural employment in 1882. There was a

tendency for the decline to be held back in remote areas (measured by

the distance from major cities).

A number of authors thought that the excessive division of land in

areas of partible inheritance had led to surplus population and under-

employment.46 If so, then this might have produced a negative relation-

ship between partible inheritance and the movement of the labour

force, as migration reduced the labour surplus in these regions. How-

ever, the analysis shows that the relationship is a positive one not a

negative one: partible inheritance is associated with a slightly larger

46 Sering (1899–1910), Conrad (1891).

Table 6.7. Summary of results of statistical analysis of the change

in the agricultural population 1882–1907 in 397 Prussian Kreise


Factors which are associated with a less rapid rise in the population occupied in

agriculture or with a decrease in the agricultural population:


A situation close to the border with Russian Poland

A high level of household employment in 1882

A high level of self-employment in 1882



Being a long distance from large cities (over 100,000 inhabitants)




A high level of partible inheritance


increase in the agricultural labour force in the Kreise where this was

practised, although the result is not a secure one in statistical terms.

The previous chapter has provided a possible reason for this. In areas

of partible inheritance demographic behaviour had adjusted so as to

avoid excessive subdivision. There was, therefore, no reason for there

to be any more underemployment, or surplus labour, in these regions

than there was in other rural areas.47

The analysis also considered some factors which might be indicative

of a high level of rural underemployment. These are the percentage of

the non-agricultural labour force which is self-employed and the per-

centage occupied in cottage industry. The justiWcation for including

these variables is that underemployment within agriculture in a particu-

lar Kreis may be related to underemployment outside. If there is surplus

agricultural labour, some of this will take non-agricultural work, even if

there is not full employment in this sector either. Expected returns in

both sectors will tend to equalize. And total annual returns will be a

multiple of expected daily earnings and the anticipated number of fully

employed days. So, unless there is a compensating diVerence in

expected earnings, underemployment in one sector will spill over into

the other.48

The results show that high levels of household employment and self-

employment are associated with slower increases in agricultural employ-

ment. In Kreise with this ‘labour surplus’ pattern of employment the rate

of natural population increase was considerably greater than the pre-

dicted increase in agricultural employment, leading to out-migration.

Table 6.8 illustrates this point, by providing predicted rates of increase

in agricultural employment for diVerent Kreise, ranked by the estimated

labour surplus (which is a weighted combination of the percentage in

self-employment and the percentage in household employment).

The estimates show that, in a Kreis in the top decile for ‘estimated

labour surplus’, with high levels of household employment and self-

employment, the predicted increase is well below the average rate of

natural increase (38% over the whole period). The implied rate of out-

migration from rural areas for such a Kreis is 25% of the population

in 1882. By contrast, a Kreis in the lowest decile would have had a

47 A number of regressions, using diVerent data sets and diVerent periods, were runlooking for a relationship between migration and partible inheritance. All these producedresults which were insigniWcant.

48 This implies that there is some Xoor to wages in both sectors, so eYciency wages arepaid.


predicted increase in agricultural employment nearly equivalent to the

rate of population increase, leaving a migration Xow equivalent to just

7% of the population in 1882. This measure provides strong evidence

for the view that labour surplus conditions in rural areas contributed

heavily to migration in the period.

8. The Lewis Model and East-Elbian Agriculture

The focus in this chapter has been on the eVect that large farms and

estates had on migration out of rural areas. A central theme has been to

examine the view of Max Weber that there was a change in the values

and attitudes of land-owners which led to deterioration in the condition

of the agricultural worker and the release of surplus labour.

The evidence does not support the view that there was a general fall

in real wages after 1870, nor was there any overall tendency for larger

farms to shed labour. On the other hand, it does appear that the shift to

more intensive arable production systems, characterized by the intro-

duction of sugar beet, did lead to a change in labour relations, involving

the increased use of migrant labour, and this had a depressing eVect on

Table 6.8. Analysis of the change in the agricultural population,

1882–1907: predicted rates of change for 397 Prussian Kreise ranked by

estimated surplus labour

Ranking of Kreis by

estimated labour

surplus

Household

employment

1882 (%)

Self-employment

1882 (%)

Predicted change

in agricultural

population

1882–1907 (%)

Top decile average 18.74 39.36 þ12.95

Top quartile average 9.10 38.05 þ16.84

Average all Kreise 3.68 28.89 þ23.34

Lower quartile average 1.60 19.51 þ28.85

Lower decile average 0.87 15.59 þ31.09

Note: The ranking is obtained by using the coeVicients given in column c of Table 6B.2for %HOUSEIND and %SELFEMPLOYED, and applying them to the Kreise data toproduce an estimate of the labour surplus. Note also that the increase in the agriculturallabour force shown between 1882 and 1895 was at least in part due to the under-recording of female labour in the earlier census.


wages in the areas most aVected. This eVect was strongest in the east,

due to the proximity of the border with Russian Poland.

One constant theme running through the results presented in this

chapter is the inXuence of Russian Poland, both on wage rates and on the

numbers occupied in agriculture. This inXuence seems to have increased.

In Weber’s view this was due to a change in the behaviour of land-owners.

There are, however, other factors which might account for this.

In the Wrst place, Russia was starting to industrialize. From the 1880s

onwards growth was rapid.49 This would have set oV a similar process

of increased mobility and migration to the one this study has described

for Germany. This could have made it easier for German farmers, and

their agents, to obtain migrant Polish workers. Another consequence of

this rise in mobility was a dramatic increase in the emigration of Rus-

sian Poles to the United States.

Secondly, there may have been a change in the attitude of the

workers themselves. The bargain represented by the traditional con-

tractual relationship of eastern agriculture was one-sided. Although

Weber interpreted the contractual relationships of eastern agriculture as

reXecting a pre-capitalist desire on the part of the land-owners to

employ as many as possible, the evidence he provides is not consistent

with this. The essence of the Inste system was that it did not oVer year-

long full-time employment (as some landlord who sought only to maxi-

mize employment might have done). Rather it ensured that the peak

seasonal labour requirements of the land-owner were met, even if this

meant considerable underemployment for the rest of the year. The

provision of smallholdings and grazing rights meant that this ‘surplus

labour’ could be absorbed in the Inste household economy during the

slack periods. The system he described was one motivated by self-inter-

est not paternalism.

Industrialization made it less necessary for agricultural workers to

accept this. There is evidence that some eastern workers, and in par-

ticular their children, gave up their contractual status voluntarily, in

response to a general rise in the demand for labour and increased wages

outside eastern agriculture.50 So land-owners were obliged to look else-

where for labour to meet peak seasonal requirements. Russian Poland

was the nearest alternative.

49 Gregory (1997).50 There is a report on this in the Jahrbuch fur die Amtlich Statistik des Preußischen

Staats II (1867), 284.


Taking the long view, the movement of German farmers into the

eastern provinces, and into other regions of eastern Europe, was a

product of high population pressure in other parts of Germany, and a

lack of suitable land for settlement nearby. In the eighteenth century

the Prussian state was able to attract settlers from Wurttemberg and

Switzerland.51 But around 1860, this eastwards shift of the German

population, which had been going on for centuries, came to an end.

Instead there is a movement of the German population to the centre

and west, and this is followed by an inXow of migrants from lands

further east. The reason for this was the rise of industry in the west

and the centre.

Weber, and other nationalists, saw this as a problem. But for those

who left eastern agriculture it was an opportunity. Conditions in indus-

trializing cities were harsh, but for many they represented an improve-

ment on what they had left behind. It was no longer necessary to accept

institutional underemployment.

The broad conclusion of this chapter is favourable to the Lewis

Model and to the concept of the rural labour surplus. The evidence

points to the existence of a reserve of underemployed rural labour

available for use in industry. But it also makes the point that geograph-

ical factors are important. Germany was not an island. It was connected

to conditions in the labour markets of eastern and central Europe.

Surplus labour in these markets was available for use in Germany. The

operation of the Lewis Model in one country can be aVected by condi-

tions in neighbouring economies.

51 Keyser (1959), Hauser (1965).



and Sources

%DAGPOP Percentage change in the agricultural population

(includes dependants of occupied population)

1882–1907, from occupational censuses, SDR

n.f. 2 (for 1882) and SDR n.f. 209 (for 1907),

adjusted to 1882 Kreis boundaries where these

had changed.

FOODPRICES A weighted index of food prices for 1899–1901,

for 165 Prussian towns, weights are 1900 shares

in private consumption from HoVmann (1965),applied to the nearby Kreise, prices from Neu-

haus (1904).

RURALWAGES Average wages in Pfennigs (Ortsubliche Tage-

lohne) for rural areas; the Wgures are averages of

male and female daily rates, 1892 and 1901,

ZdKPSB (1904), 320–8.

%LAND>100HA The percentage of the total agricultural area in

holdings of over 100 hectares, from 1895 Agri-

cultural Census, SDR n.f. 112, pp. 489–500.

%AGOCCP1882 The population occupied in agriculture as a

percentage of the total occupied population,

from the 1882 occupational census, SDR n.f. 2.

RUSSDISTANCE Shortest distance in kilometres from the Kreis

mid-point to the Russian border (measured

from contemporary maps).

CITYDISTANCE(a) Distance in kilometres from the Kreis mid-


inhabitants in 1900.

CITYDISTANCE(b) As above, for cities with at least 100,000 inhab-

itants.

%SUGAR1897 Sugar beet area as % of the total arable area, by

Kreise, for 1897, from PS vol. 154 (1898), 8–163.

%PARTIBLE The percentage of land transferred by inherit-

ance in 1896–1902 which was divided, ZdKPSB(1905), 262–7.

%HOUSEIND The percentage of the non-agricultural occu-

pied population occupied in Hausindustrie, 1882

occupational census, from PS vol. 83 (1884),

part 2, pp. 3–466.

%SELFEMPLOYED As above, for those recorded as self-employed.


Appendix 6B. Results of Econometric Analysis

(i) Statistical analysis of wage levels in rural Prussian Kreise 1892/1901

The dependent variable is taken from the Wgures provided by the Reichs-

versicherungsamt. These were used to calculate average rural wages for

each Kreis combining the 1892 and 1901 surveys.52 In the surveys

diVerent wage rates were given for male and female workers, and for

those aged below 16. The Wgures used are weighted averages of these,

the weights being the share of these categories in the total agricultural

52 Figures are from Neuhaus (1904).

Table 6B.1. InXuences on rural wages in the Prussian Kreise

a b c d

Constant 69.0 84.6 77.8 87.8

FOODPRICES þ0.581 þ0.486 þ0.516 þ0.491

(4.40) (3.72) (3.86) (3.78)

RUSSDISTANCE þ0.0615 þ0.0649 þ0.0681 þ0.0626

(15.08) (16.05) (14.35) (14.96)

%AGOCCP1882 �0.390 �0.458 �0.459 �0.442

(8.38) (9.45) (9.31) (9.04)

%HOUSEIND �0.308 �0.288 �0.271 �0.282

(3.94) (3.77) (3.48) (3.70)

%SUGAR1897 �4.310 �4.142 �4.953

(4.39) (4.13) (4.82)

%LAND>100HA þ0.060

(1.15)

CITYDISTANCE(a) þ0.011

(0.66)

CITYDISTANCE(b) �0.0480

(2.03)

N 392 392 392 392

S 13.76 13.42 13.43 13.37

R2 0.738 0.752 0.753 0.754

adj. R2 0.735 0.748 0.748 0.750

F-test of regression 272.9 233.6 167.0 196.9

RSS 73440 69564 69253 68830

Note: Dependent variable is average daily wages in Pfennigs, 1892 and 1901 surveyscombined (RURALWAGES), estimated using OLS, t-ratios in brackets.


workforce for the whole of Germany (from the 1907 occupational

census). They are thus standardized estimates, removing the eVect of

variations in the age or sex distribution of the labour force by Kreis.

These were then used as the dependent variable in the regressions

shown in Table 6B.1.

The Wrst column gives the basic results. One important inXuence on

wages was the cost of living in the locality. Ideally, it would be possible

to construct an index of real wages, making allowance for housing costs,

and the prices of all consumer goods, but the data sources are not

available for such a calculation. However, there was a survey of food

prices in 165 Prussian market towns, published in the Zeitschrift des

Koniglich Preußischen Statistischen Bureaus. The data is not given at the

level of the individual Kreise, and only Wve items are covered (rye,

wheat, potatoes, beef, and pork).

This survey was used to construct a weighted index of food prices

(FOODPRICE) for nearby Kreise. The Wrst line of Table 6B.1 shows

that this does indeed have an important eVect on wages. Wages are

higher in areas with high food prices.

Other inXuences include the level of employment in non-agricultural

sectors in the Kreis: the coeYcient on %AGOCCP1882, the percentage

occupied in agriculture according to the 1882 occupational census, is

strongly negative. Wages were lower in regions where there was little

alternative employment. However, the coeYcient on employment in

‘cottage industry’ (%HOUSEIND—non-agricultural household

employment) is a negative one. As the presence of other forms of

non-agricultural employment has a positive eVect, this implies that the

presence of cottage industry may well be an indicator of underemploy-

ment. Cottage industry was found in areas which had a long-term

problem of overpopulation and underemployment.

However, the most important eVect on the structure of wages in

German rural areas was the distance in kilometres from the frontier

with Russian Poland (RUSSDISTANCE). This is an important eVect:

wages rise by 6–7 Pfennigs per 100 km. A regression equation with just

RUSSDISTANCE as the sole explanatory variable has an R2 of 0.673,

indicating that this factor alone explains over two-thirds of the observed

variance in money wages.

Adding sugar beet (column b) improves the overall Wt. The coeY-

cient is negative, indicating that wages in sugar beet regions were lower

than would be expected. This coeYcient is not greatly aVected by the

inclusion of other variables in columns c and d.


The third column introduces %LAND>100HA, the percentage of

land in large holdings. If large estates did pay low wages then this

variable should have a negative coeYcient. In fact the estimated coeY-

cient is positive although not signiWcant at the 5% level. There is no

evidence to support the view that larger farms and estates paid low

wages; the evidence suggests that they may have raised wages.

The CITYDISTANCE variables introduced in columns c and d are

used to test the robustness of the estimated coeYcients on the other

variables. The strongest eVect on wages comes from CITYDISTAN-

CE(b) (the distance to cities with over 100,000 inhabitants). The co-

eYcient is negative—wages are lower further away from the major

cities—which is as expected. The other results are not seriously aVected.

Table 6B.2. Underemployment and the change in the agricultural

labour force

a b c

Constant 21.6 30.3 35.2

%AGOCCP1882 �0.500 �0.458 �0.555

(7.63) (5.90) (6.86)

RUSSDISTANCE þ0.0468 þ0.0485 þ0.0477

(10.52) (11.09) (11.86)

CITYDISTANCE(a) þ0.123 þ0.123 þ0.181

(5.14) (5.24) (7.39)

%PARTIBLE þ0.107 þ0.099

(1.30) (1.23)

%HOUSEIND �0.330 �0.334

(3.25) (3.00)

%SELFEMPLOYED �0.359 �0.512

(2.58) (3.50)

N 347 347 397

S 17.47 17.05 19.47

R2 0.409 0.442 0.432

adj. R2 0.402 0.432 0.424


RSS 104361 98505 148182

Note: Dependent variable is change in total agricultural population 1882–1907 as apercentage of the 1882 Wgure (%DAGPOP), estimated using OLS, t-ratios inbrackets.


(ii) Underemployment and migration

The analysis in this section looks at the connection between under-

employment and migration, making use of data sources which are only

available for the Prussian Kreise. The dependent variable in Table 6B.2

is the change in the total agricultural population in the Prussian Kreise

between the two occupational censuses of 1882 and 1907. The Wrst

three variables are those found to have the largest eVect on the move-

ment of the agricultural population. The decline was steepest in areas

which were closest to the border with Russian Poland: by contrast,

farms in central and western Germany tended to raise employment.

The decline was also steeper in areas close to major cities, as is shown

by the positive coeYcient on CITYDISTANCE(a).

There is a strong negative eVect from the percentage occupied in

agriculture in 1882 (%AGOCCP1882). This can be interpreted as

showing that there was underemployment in many predominantly agri-

cultural areas at that time, so that the decline in the agricultural labour

force was higher in Kreise which had little non-agricultural employ-

ment. The data set was used to study the eVect of partible inheritance,

using %PARTIBLE (the percentage of land transferred by inheritance

1896–1902 which was partitioned). Columns a and b show that the

coeYcient on this variable is positive and not signiWcant. There was no

tendency for partible inheritance to produce out-migration; there was a

slight eVect in the opposite direction.

Column b adds two other variables. These are employment in cottage

industry and self-employment (%HOUSEIND and %SELFEM-

PLOYED). Both are expressed as percentages of total non-agricultural

employment, so they are possible measures of underemployment out-

side agriculture.

The results in columns b and c show that coeYcients on both these

variables are negative and signiWcant. So, summarizing these results,

there is good evidence that there was underemployment in rural areas

in 1882, and that, in 1882–1907, this was reduced through migration, as

the rural labour surplus was transferred to urban labour markets.


7

Agricultural Productivity,

Labour Surplus, and Migration

1. Introduction

In the previous chapters a number of diVerent explanations of the

agricultural ‘labour surplus’ have been considered. Some have been

rejected, some have found at least partial support: the agrarian reforms

did create a class of landless labourers, there was considerable demo-

graphic pressure in parts of rural Germany, there was underemploy-

ment in the agrarian sector and its ‘cottage industry’ oVshoot, there

were institutional and social changes in agriculture which contributed

to out-migration.

In Chapter 1 there was a discussion of Theodore Schultz’s critique of

the concept of the labour surplus. The essence of Schultz’s view is that,

although productivity may be low in traditional agriculture, this is not

due to ineYciency or underemployment, but rather to a lack of

resources, in particular to a lack of ‘non-traditional resources’: educa-

tion, science, new productive knowledge. Schultz also stressed the im-

portance of access to produce markets and to capital.

Industrialization brings to agriculture access to improved inputs. The

creation of a railway network and other transport improvements make it

possible to transport bulky inputs like fertilizers and animal feedstuVsover long distances. This also makes it easier to get agricultural produce

to market, and, by integrating markets, helps to reduce price volatility.

Education is another productive factor which is likely to be aVected

by industrialization. Industrializing societies, particularly ones like

Germany which develop a specialization in advanced technology, Wnd it

desirable to invest in educational improvements so as to raise the skill

level of the workforce. If the system is a universal one, then the rural

sector will also have access to a higher level of general education. In

specialist Welds such as scientiWc education, advances in basic science

often have implications for agriculture. Nineteenth-century Germany

was a classic case of this: advances in the understanding of organic

chemistry fed into an improved understanding of plant nutrient needs

and how to supply these.

Industry also supplies improved inputs directly. Steam power could

be applied to agricultural tasks, such as ploughing and threshing. In-

dustrial by-products such as ‘basic slag’ turned out to be valuable fertil-

izers. Improvements in mechanical engineering made it possible to

design and build machines such as reaper-binders or seed drills, which

were suYciently robust to stand up to being pulled over Welds while in

operation.

These ‘non-traditional inputs’ can, therefore, lead to productivity

increases and improved yields as farmers react to the stimulation pro-

vided by rising urban demand and better transport links, by investing

in new technology. This does not lead directly to out-migration, but it

does have indirect eVects which might produce this. In the Wrst place, it

will lead to social change within agriculture: as this becomes more

market-orientated, traditional social attitudes and institutions will tend

to be eroded. Secondly, it creates the possibility that the nutritional

needs of the country could be supplied by a much smaller agricultural

sector (measured by the relative size of the agricultural population).

Whether or not there is a relative decline in the agricultural sector as a

result depends on the evolution of the foreign trade position. It is quite

possible for countries to industrialize, to develop modern industrial

sectors, and at the same time remain major exporters of primary prod-

ucts, including agricultural goods. Examples include Brazil and Austra-

lia in the twentieth century and the United States in the nineteenth. In

these examples, the industrial sector concentrated on supplying the

home market. Under these circumstances, the relative importance of the

primary-producing sectors can remain high despite industrialization.

But consideration of these examples shows quite clearly that they

were all ‘resource rich’ countries, with comparatively small populations

relative to the availability of land and other resources. Late nineteenth-

century Germany was not ‘resource rich’ in these terms, especially as

the expansion of European settlement and falling transport costs were

bringing new areas into the world trading system which had far more

favourable land–labour ratios. Germany’s comparative advantage in

world trade lay in her scientiWc base, in the education system, and in

the skills and knowledge possessed by urban workers. Germany was

‘skill rich’ not ‘resource rich’.

The logic of Germany’s position was, therefore, that the German

economy should export skill-intensive industrial goods and import food

216 Agricultural Productivity, Labour Surplus, Migration

and raw materials. This meant that demand for home-produced agricul-

tural goods was limited to a declining share of the home market. Under

these circumstances, improvements in agricultural productivity meant

that the available demand could be supplied by a smaller labour force.

As Germany became more integrated in the world trading system, re-

sources, including labour, would shift to the exporting industrial sector,

and leave the sector exposed to rising imports, which was agriculture.

The context for the discussion of these issues is the more general

question of the relationship between agriculture and industry in the

circumstances of the late nineteenth century, when cheap US grain

exports reduced prices on world agricultural markets. European coun-

tries reacted in diVerent ways, inXuenced by internal political and social

pressures, and by economic conditions. Many tried to mitigate the

eVect on their agricultural sectors by increasing tariVs on agricultural

goods. Germany was not exceptional in this.

An important point about Germany is that it was already industrial-

izing, and that it had a comparative advantage in industrial goods. This

meant that it was able to pay for increased agricultural imports by

exporting industrial goods. This was not an option available to all

European countries. Perhaps only Britain and Belgium were in a similar

position.1

A second point about the German position is that the arrival of cheap

US grain exports in the 1870s came at a time when Germany was still

predominantly rural, and when German cities were already under pres-

sure from rapid growth. By contrast, the agricultural population in

Britain had become relatively small by that time, and British cities had

been growing steadily for a century or more. The consequences to

British society of a free-trade policy were absorbed much more easily.

A simple calculation illustrates this point. Suppose that a country

with the rural–urban balance of Britain in 1871 (42% of the population

in towns with at least 20,000 inhabitants) maintained a free-trade policy

in face of rising agricultural imports, and that this caused 10% of the

rural population to move to the cities over the next ten years. The

additional migrants would amount to 14% of the initial urban popula-

tion: a burden for the urban areas but not an insurmountable one. The

same calculation for a country with a similar balance to Germany in

1 For a general discussion of these issues, see O’Rourke (1997). There is an importantdistinction to be made between countries which had the capacity to raise industrial exportsand other less industrial countries, whose best option was probably to reorientate agricul-tural exports to the livestock sector, as Denmark did.

Agricultural Productivity, Labour Surplus, Migration 217

1871 (12.5% in towns with at least 20,000 inhabitants) would produce a

movement equivalent to 70% of the initial urban population. This

would have raised the growth rate of German cities from 3.0–3.5% per

annum to over 8%. Such a rate would have created intolerable strains.

If there is a relationship between the social costs of migration and the

rate of urban growth, there may be good reasons for a government to

seek to slow down the decline of the agricultural sector. Agricultural

protection is probably undesirable on economic grounds, but it may be

a reasonable response for a society which fears the consequences of a

collapse of the agricultural sector.2

2. Agriculture in the German Economy

The relative importance of the German agricultural sector fell in 1870–1913, despite protection. Table 7.1 presents some calculations, from

Wgures provided by HoVmann, which show how the changing structure

of the German economy aVected the position of agriculture.

HoVmann made a reasonable adjustment to the agricultural labour

force Wgures to allow for under-recording of family labour in the earlier

censuses.3 The resulting estimates show the agricultural labour force

rising at a slow rate, 25% in 42 years (1871 to 1913), while the non-

agricultural labour force more than doubled (a rise of 125%). The

estimated rise in agricultural productivity 1870/2 to 1911/13 is 1.0%

per annum, which is relatively high compared to other European coun-

tries (such as Britain), but still below the rise in the non-agricultural

sector (1.4% per annum). Consequently, the fall in the agricultural

share of total net value added is rather greater than the fall in the share

of the total labour force.4

It is possible to decompose the fall in the agricultural share of net

domestic product into two parts: that which is due to changes in the

structure of domestic demand and that which is due to changes in

foreign trade. As the table shows, there was a small net deWcit on

2 There are other reasons to support agriculture, including the strategic one. A countrydependent on food imports is in a vulnerable position, as was demonstrated by theGerman experience in the First World War, a point discussed in OVer (1989).

3 On the under-recording of female labour in the 1882 census, see the discussion inTipton (1976) and also comments in the introduction to the 1895 survey, SDR n.f. 112.

4 The increase in labour productivity in British agriculture averaged 0.5% per annum1873–1913, Matthews et al. (1982), 228–9.


agricultural trade in 1870–2, equivalent to 0.7% of NDP (1913 prices).

By 1911–13 this had risen to 4.5% of NDP. The change in the external

balance contributed 20% of the total fall in the agricultural share of net

value added.5 But this shows that there would have been a substantial

fall in the agricultural share even without the contribution of the exter-

nal sector. The shift in the structure of domestic demand had rather

more eVect than the change in the foreign trade position.

Comparison of the net import Wgures with the Wgures for total agri-

cultural production (not just value added) shows that agricultural self-

suYciency in all sectors, deWned as total domestic production as a per-

centage of total consumption, fell from 98.7% in 1870–2 to 88.1% in

1911–13. This exaggerates Germany’s ability to feed itself without

using imports, since it takes no account of imported fertilizers and

other non-agricultural inputs. Moreover, the importance of fertilizers,

in particular, was far greater than that implied by the Wgure of their

cost relative to total production.6

Figure 7.1 concentrates attention on trade in cereals. In the period

before 1870 Germany exported cereals in most years. Germany sup-

plied 23.5% of British cereal imports in 1856–60, but this share was

declining, in 1871–5 it was only 8.2% (the share of the United States

5 This Wgure is based on an estimate of the size of the agricultural share in NDPassuming that the deWcit on agricultural trade had remained at the 1870–2 Wgure. It takesaccount of the point that imports were themselves Wnal products; their importance tends,therefore, to be exaggerated by a comparison with domestic value added.

6 A calculation of self-suYciency including the eVect of imported fertilizers is noteasily done. Some fertilizers also had non-agricultural uses (nitrates were used in themanufacture of explosives). Germany exported potash and imported phosphate and nitro-gen. Simply adding up tons imported and exported will give a misleading impression.

Table 7.1. Agriculture in the German economy 1870–1913 (all

calculations use 1913 prices)

Agricultural share in: Exports as % of NDP Imports as % of NDP

NDP Labour

force

Non-

agricultural

Agricultural Non-

agricultural

Agricultural

1870–2 38.7 49.2 10.1 2.7 13.0 3.4

1911–13 23.0 34.9 18.0 1.8 16.4 6.3

Source: Data from Hoffmann (1965); Hoffmann builds up his accounts from estimates of netvalue added by sector, the final figure is therefore net domestic product (NDP).


rose from 18.8% to 40.9% over the same period).7 But before conclud-

ing that it was US competition that was the cause of this loss of share,

it should be noted that German producer prices actually rose in

1856/60 to 1871/5: by 13% for wheat, and 15% for rye. The eVects of

increased US competition in third markets was outweighed by buoyant

domestic demand.

A careful examination of the timing of the fall in cereal producer

prices in the 1870s (given in Figure 7.2) shows that this came after the

shift to a net import position. The 1870s were a period of relatively

high producer prices, certainly better than the early 1860s. There is no

evident downward trend in cereal prices until after the peak in 1873–4.

7 Lambi (1963), also Hobson (1997).

1910190018901880187018601850

30

20

10

0

−10

Fig. 7.1. Net imports of cereals as a percentage of total domestic consumption

(including use as animal feed), Germany (post-1871 boundaries) 1850–1913

Note: It is difficult to find any clear evidence in this diagram of effects from changes intariff policy, Bismarck’s duties of 1879–80 produced, at the most, a brief interruption inthe upwards trend in imports. The diagram also prompts the question: was it conceivable,in the context of pre-1914 power politics, for any country to have been as dependent onoverseas trade for vital supplies, and yet not build a navy to protect this trade? See Offer(1989) for a discussion of the consequences. The issue is also considered in Chapter 10 ofthis study.Source: Calculated from Hoffmann (1966), 292–3, 530–5, and 537–43.


It was the low prices of the 1878–9 period that changed the attitude of

the German agricultural lobby, which had previously supported free

trade.

At the same time domestic demand for food was buoyant, increasing

the viability of a strategy which relied more on protected domestic

markets, and less on exports. Table 7.2 gives some Wgures for per capita

consumption. Between 1850–4 and 1870–4 there were substantial rises

in most categories, apart from pulses and other cereals. The increases in

the consumption of sugar and pork are particularly notable. As the

German population rose by 16% over this period, the overall increase

in demand was high.

Rising demand for livestock produce increased the amount of cereals

required as livestock feed: principally oats, but also some barley and

rye; potatoes were also fed to livestock. This reduced the area that

could be allocated to exports. One consequence of this period of relative

agricultural prosperity was a rise in average rents, from 14.10 Marks

189018851880187518701865186018551850

300

250

200

150

100

50

0

Wheat

Rye

Barley

Oats

Fig. 7.2. Cereal producer prices 1850–95 (in Marks per ton)

Note: The ranking of the crops by production in 1870–2 is, first, rye (40% of the total byweight), then oats (28%), wheat and spelt (17%), and finally barley (15%).Source: From Hoffmann (1966), 552–4.


per hectare in 1850 to 38.16 in 1880.8 This is a clear indication that

problems in third markets had not been the principal cause of the

reduction in exports.

After 1880 the connection between rising imports and falling produ-

cer prices becomes clearer. There is a rise in net imports from 7% of

total consumption in the period around 1880 to 15% in the mid-1890s,and there is an evident downward trend in producer prices. TariVs did

have some eVect in holding back the fall in prices, as Table 7.3 shows.

The fall in producer prices was 53 Marks per ton (1871/80 to 1891/

1900), but the Hamburg price fell by 93 Marks per ton over the same

period. Comparison of this price with the prices in Nebraska and New

York shows the eVect of falling transport costs: a total reduction of

55%, 70% of this coming from a reduction of costs within the United

States.

After 1895 net imports of cereals remained at around 15% of total

consumption, but net imports of all agricultural produce continued to

rise. In 1911–13 net imports were 15.6% of total food consumption; net

imports of foodstuVs other than cereals and sugar having risen from

8 Henning (1977), 223.

Table 7.2. Per capita food consumption: five-year averages

Wheat

and rye

Potatoes Pulses and

other cereals

Sugar Beef &

veal

Pork Milk Eggs

Annual consumption per capita in kg: In numbers:

1850–4 90 120 26 2 10 7 272 47

1870–4 114 169 18 5 12 13 296 54

1890–4 127 242 12 9 15 18 313 72

1909–13 128 200 11 18 17 25 381 102

Note: There are reasons to question the 1850–4 figures. The implied increase in percapita calorie consumption of these foods between the first two periods is high (20.8% in1850/4 to 1870/4 against an increase of 7.8% in 1890/4 to 1909/13). Some of this maybe explained by the increased use of cereals and potatoes for domestic alcohol production.But, in the earlier years, Hoffmann may also have underestimated personal consumptionof pulses and other cereals (as opposed to their use as livestock feed). High potatoconsumption in 1890–4 may also be spurious, if a high proportion was fed to livestock inthese years. But in general, the implied calorie consumption figures for the period after1870 are not unreasonable.Source: Calculated from Hoffmann (1966), 622–33.


2.7% of total consumption in 1870–2 to 12.0% in 1894–6 and 15.5% in

1911–13. The rise after 1894–6 was not due to a increase in meat

imports, which fell slightly in the period to 1911–13 (including trade

in live animals), but to rises in other categories such as fruit (up

168%) and eggs and dairy products (up 180%) over the same period.

By 1911–13 Germany was importing nearly 11% of total consumption

of eggs and dairy produce.9

The impressive performance of German agriculture in raising pro-

duction has, therefore, to be set against a background of buoyant

demand. Domestic production was not suYcient to supply all the food

required by the expanding urban population. But it was important that

productivity rose by as much as it did. A fairly simple calculation shows

that, if agricultural productivity had risen by 0.8% per annum instead

of 1.0%, domestic food production in 1911–13 would have been 26%

9 Figures from HoVmann. The question of the relative rates of protection given todiVerent agricultural sectors has been a matter of controversy since Gerschenkron’s 1943study, Gerschenkron (1943), Hardach (1967), and Webb (1982).

Table 7.3. Wheat prices, tariffs and transport costs

(a) Average wheat prices, by decade

(Marks per Ton)

1871–80 1881–90 1891–1900

Nebraska 106 99 81

New York 195 155 123

Hamburg (f.o.b.) 227 164 135

Average German producer prices 224 187 171

(b) German duties on imported wheat (minimum duty payable)

1879–85 1885–7 1887–91 1891–1902 1902–6 1906–10

Tariff in Marks/

Ton

10 30 50 35 55–75

(sliding

scale)

55

Source: Prices in Nebraska, New York, and Hamburg taken from Henning (1977), 219,German producer prices calculated from Hoffmann (1966), 552–4, tariffs from Tracey(1982). These are the minimum rates, so the lower rates for 1891–1902 reflect the tradetreaties with Austria-Hungary, Romania, and Russia. American grain was still paying the1887–91 rate.


lower. Under these circumstances, the achievement of a similar level of

self-suYciency to that actually attained in 1911–13 would have required

additional capital and labour. Using a joint marginal output coeYcient

of 0.85 for capital and labour (and assuming that the capital input rose

in proportion with labour), it can be calculated that the agricultural

labour force would have had to be 40% higher, so that agriculture

would have continued to account for 49% of the total labour force.

This would have had a major eVect on German industrial growth. It

also implies a much lower rate of rural–urban migration.10

A similar calculation shows that the cost of achieving full agricultural

self-suYciency would also have been high. To achieve the necessary

increase in production 16% more labour would have been needed in

agriculture. This would have meant that 40% of the labour force would

have still been occupied in agriculture in 1911–13 instead of 35%.

Again, this would have held back industrial growth.11

There was a cost to protection. Food prices were raised, though by

less than the price diVerences given in Table 7.3. Calculations using

price ratios from the 1905 survey by the British Board of Trade, shown

in Table 7.4, show that food prices were 14.5% above the British level

(weighted average of the Wgures in the table). According to the house-

hold budget surveys, these items accounted for 40% of total household

expenditure. So there was certainly an eVect on real wages from protec-

tion. But this was oVset by lower prices for milk and potatoes, presum-

ably reXecting lower German production costs in these sectors.

Germany faced some diYcult choices about the position of agricul-

ture within the whole economy, and it is not surprising that this was an

issue which was much debated. The outcome which was chosen was

one that involved a slow decline in self-suYciency, a rate of industrial

growth which was fast by the standards of the time, large-scale migra-

tion to the cities, and a level of agricultural protection which imposed

some costs on urban consumers.12 It was in many ways a sensible

10 Van Zanden (1991) suggests coeYcients of 0.35 for land, 0.5 for labour, and 0.15 forcapital (livestock). This gives a joint coeYcient on mobile factors (capital and labour) of0.65. This is probably too low. Additional investment could have raised the value of land(by drainage for example) even though there was little suitable land left uncultivated. Ajoint coeYcient of 0.85 is more reasonable.

11 Full self-suYciency is deWned as an increase of production suYcient to reduce netimports to zero. Germany would still have imported some products and exported others.There would also have been imports of fertilizers.

12 There was, however, a political price. The crisis in agriculture, even if it was a crisis ofadjustment, contributed to the radicalization of right-wing politics, Rettalack (1988), 101.


solution, which avoided the social costs of a complete agricultural col-

lapse, but still allowed urban living standards to increase.

3. ‘Few better farmers in Europe ’ : Technological

Progress in German Agriculture 1870–1913

In 1921 J. H. Clapham published The Economic Development of France

and Germany 1815–1914. Writing about German agriculture in com-

parison with France, and, by implication, in comparison with Britain as

well, Clapham was expressing opinions widely held by contemporary

observers of pre-1914 German agriculture. Germany was perceived as

the leader of ‘scientiWc agriculture’:

there were few better farmers in Europe than the best of the Junkers in the later

nineteenth century . . . after the decade 1840–50 Germany had little to learn

from any country in agriculture and forestry. In agricultural chemistry she was

undoubtedly the leader.13

In his writings Clapham placed great emphasis on the role of the

larger farms and estates of eastern Germany: ‘these were the Welds on

13 Clapham (1921), 206.

Table 7.4. German food prices as percentages of British prices, 1905

(percentage of total German household expenditure spent on each

category shown in brackets)

Wheat and rye flour 140 (2.3) Beef 122 (4.4)

Wheat and rye bread 131 (10.0) Mutton 137 (0.3)

Sugar 119 (1.4) Pork and lard 123 (8.4)

Potatoes 88 (2.6) Butter 105 (4.6)

Milk 75 (5.8)

Source: Price ratios from Board of Trade (1908); household expenditure proportions takenfrom Desai (1968), 122, who used the results of a household budget survey published inthe 1909 Reichsarbeitsblatt. Additional information on the consumption of different meatswas taken from Hoffmann (1966), 632. Price ratios for pork were also applied to lard;those for wheat flour were used for rye as well. The price ratio for bread was derivedusing the baker’s premium found by Peterson (1995), 271: 24% of the British price forwheat flour. This was used as an estimate of the absolute cost of producing bread fromflour in both countries. The Board of Trade survey provides a better estimate of theeffect of German tariffs than calculations based on the movement of prices insideGermany. World prices were falling in the late nineteenth century; German tariffs heldback this decline.


which the Wght for scientiWc agriculture was won’. But he also pointed

out that these diVered from the typical English estate of the time, which

was rented out to tenant farmers:

the German Empire was . . . a land of free land-owning peasants and powerful

cultivating squires . . . In a broad survey of modern German agriculture tenant

farming might almost be omitted . . . East of the Elbe where big estates were

common they were nearly all in hand. The typical eastern Junker lived on and

directed the management of an estate of perhaps 2000 acres of cultivated land . . . 14

ScientiWc agriculture involved, in particular, the application of the

discoveries of organic chemistry to agricultural practice. But it was also

recognized that knowledge of these new practices had to be diVused.

This involved a major expansion of agricultural education. By 1908

Germany had 26 Agricultural Mittelschulen, 45 Ackerbauschulen, 43

Agricultural Fachschulen, and, for those without access to a full-time

secondary school (or unable to aVord the time), there were Winterschu-len (204 of these in Prussia alone).15 Agricultural faculties had been

established at many universities. There was a network of research sta-

tions and experimental farms run by the provincial authorities and the

local Landwirtschaftsgesellchaften.16

The new agricultural science was based on an improved knowledge of

the nutrients required by plants, and the extent to which these would be

removed when the crop was harvested. Soil testing was used to establish

the extent to which nutrients could be supplied from resources already

present in the soil. DeWciencies could then be made up using a combin-

ation of animal manure, artiWcial fertilizers, and ‘green manure’: nitrogen-

ous plants which were ploughed in to provide food for subsequent crops.17

The story of one estate, Kittnau in Kreis Graudenz in West Prussia,

illustrates the extent of the changes:

14 The entrepreneurial abilities of the Junkers have been recognized by some authorswho are critical of their political inXuence, for example Rosenberg (1978), 94. On theother hand, Puhle (1986), 86, claims that the Junkers refused to invest or modernize theirestates, and had a ‘general disinterest in the nature of agricultural production’. Accordingto Schissler (1986), 36, ‘historians agree that the Junkers represent a great political,economic and social burden for Prusso-German history’.

15 Figures from Muller (1985).16 The German agricultural education system was surveyed in 1881–3 by the Royal

Commission on technical education. This part of the report was published in the Journalof the Royal Agricultural Society of England and Wales (1885), 126–64 and 518–46.

17 Valuable studies of late nineteenth-century agricultural technology and its eVect onproductivity include Evershed (1884), Dyer (1896), Rauchberg (1900), Sering (1932),Schremmer (1988), Haushofer (1957), Frauendorfer (1956), Rolfes (1976), Henning(1977), Perkins (1986), Reif (1994), and Fussell (1962).


Previously there had been little cultivation of root crops. In 1884 a sugar beet

factory was constructed at the railway station at Melno, 5 km from the estate,

and the manager, Oberamtmann Muller, immediately switched to intensive

sugar beet cultivation. Clover was replaced with a rotation using lucerne, and a

third of the farmed land was put into sugar beet. The arable land was drained, a

Fowler steam plough was bought and sugar beet was cultivated according to the

practices developed in Prussian Saxony. In order to move the sugar beet

promptly to the factory, around 5 km of roads were built at a total cost of

40,000 Marks. At the same time cereal yields also increased as a result of the

improved husbandry system.18

What evidence is there of a major change in the performance of

German agriculture? The Wrst place to look is movement of yields rela-

tive to other countries. Figure 7.3 gives average wheat yields in four

countries 1848–1913. In the 1850s German yields were at a similar level

1908189818881878186818581848

4

3

2

1

0.9

Germany

log scale

FranceDenmarkBritainGermany, eastGermany, rest

Fig. 7.3. Wheat yields 1848–1913, 1,000 kg per hectare, rolling five-year

averages

Source: Calculated from data in Mitchell (1981), from national sources, but then adjustedusing the 1905–9 Institut International d’Agriculture comparison of cereal yields: InstitutInternational d’Agriculture (1915), 28. The British figures were converted to metric tonsusing estimates of bushel weights from the 1879 Encyclopaedia Britannica.

18 Dade (1913), i.19. This is a collection of farm histories, providing a valuable descrip-tion of German agriculture in the period.


to France, some way below those in the leading countries: Britain and

Denmark (yields were also high in Belgium and the Netherlands). From

the late 1850s onwards, German yields increased faster than in other

countries, so that they caught up with the leading group, and overtook

Britain. The German performance was even more impressive when the

fact that Britain and Denmark were reducing their wheat acreage, while

it was expanding in Germany, is taken into account.

The Wgure also shows average yields for east-Elbian Germany, and

for the rest of Germany, for the years for which this is available. The

convergence of yields in eastern Germany with those in the rest of the

country is evident. Part of the improvement in German yields was due

to a more even performance within Germany itself.

Wheat provides an interesting comparison because it was a widely

grown crop. But rye was the more important crop in Germany. Figure 7.4

repeats the analysis for rye. Britain has to be omitted as the rye crop was

insigniWcant. This again shows that German yields caught up with those in

the leading country, Denmark in this case. German yields showed no

upwards trend until 1860. After that they started to rise, but slowly. There

1908189818881878186818581848

2log scale

1

0.9

0.8

Germany

France

Denmark

Germany, east

Germany, rest

Fig. 7.4. Rye yields 1848–1913, 1,000 kg per hectare, rolling five-year averages

Source: As Figure 7.3.


was a marked acceleration after 1890. Again, this was accompanied by a

strong process of convergence within Germany, although a gap between

the east and the rest remained at the end of the period.

Another basic indicator of agricultural conditions in the nineteenth

century was livestock intensity. Even though artiWcial fertilizers were

coming into use, animal manure was still the main source of plant

nutrients—nitrogenous fertilizers in particular were not much used. So,

high livestock intensity would not only be associated with a high level

of production of livestock produce, but it would also have a beneWcial

eVect on arable yields.

International comparison of livestock numbers (Table 7.5) shows that

overall livestock intensity rose by 39% in Germany between 1873 and

1913, which brought the German Wgure up to the British level. German

livestock intensity was some way above France, but well below the

Danish Wgure. Danish agriculture had reorientated away from cereal

exports and was now concentrating on the export of livestock produce.

The composition of the German livestock Wgures changed. Sheep

numbers fell sharply, and there was a large increase in pig numbers.

Table 7.5. Livestock numbers (in thousands) and livestock intensity

Horses Cattle Pigs Sheep LU/100ha

Germany

1873 3352 15777 7124 24999 45.0

1913 4558 20994 25659 5521 62.5

% change þ36.0 þ33.1 þ260.2 �77.9 þ38.9

France

1882 2838 12997 7147 23809 37.3

1913 3222 14788 7036 16131 39.9

% change þ13.5 þ13.8 �1.6 �32.2 þ6.9

Denmark

1871 325 1239 442 1842 54.6

1914 487 2463 2467 515 105.1

% change þ49.8 þ98.8 þ458.1 �72.0 þ92.4

Britain

1872 1258 5625 2772 27922 57.8

1912 1441 7026 2656 25058 64.9

% change þ14.5 þ24.9 �4.2 �10.3 þ12.3

Source: Data from Mitchell (1981), conversion to Livestock Units (LU) uses weights fromWagner (1896).


The livestock structure in 1913 resembled the Danish pattern, rather

than that of France or Britain, where sheep numbers remained high.

The more intensive livestock systems of Germany and Denmark had an

eVect on the value of manure as fertilizer. The manure of in-housed

cattle and pigs would be stored and spread on the Welds in the spring,

which reduced nutrient loss through leaching; sheep by contrast would

be left out on the Welds all year, which meant that part of the nutrients

would have leached away by the time the crops were Wnally planted.19

Going from these comparisons of livestock numbers and yields to

comparisons of productivity involves much greater problems of com-

parability. It is not a job for the faint-hearted. One of the more ambi-

tious exercises is Bairoch’s series of estimates of calorie production per

male agricultural worker reaching back to 1800. There are two problems

here. Female labour is ignored, so genuine variations in the male–

female ratio by country will aVect the results, and the aggregation of

production by caloriWc content can be criticized on the grounds that

calorie content is a poor guide to economic value. The calorie content

of a potato is greater than that of a bunch of grapes but the price of the

latter is much higher.

These objections apply most strongly to Mediterranean agriculture,

whose products—fruit, olives, wines—often have a high value relative

to their calorie content. Productivity in these countries will be under-

estimated. Conversely, it will be overestimated in countries which spe-

cialize in root crops. The calorie output per hectare of potatoes is much

higher than the equivalent yield of wheat, yet the value of output is not

so diVerent.

However, for all these problems Bairoch’s estimates are the only ones

available for a substantial group of countries in the early nineteenth

century, and some relatively crude aggregation is probably the only

option for this period. The results are given in Figure 7.5.

According to Bairoch, agricultural productivity in the United King-

dom was substantially greater than in any other European country

around 1800. It was a third higher than the Dutch level, the nearest at

that time. In the 1860s Britain was overtaken by Denmark, and around

1890 by Germany. Germany rose from Wfth position, level with France,

to be second only to Denmark in 1910. Of the other countries, relative

losers included Austria-Hungary, France, and (perhaps more surpris-

ingly) the Netherlands; the other relative gainer was Sweden.

19 Nutrient loss through leaching is now a major agricultural concern, on environmen-tal grounds, but it was recognized in the nineteenth century: see Voelcker (1883) on this.


Some of the German improvement is due to the ‘root-crops’ bias.

Germany had a proportionately larger acreage of potatoes and sugar

beet, and this would produce an overestimate of the value of German

agricultural output.20 The value of French ‘Mediterranean’ produce

would be similarly underestimated. The use of male labour as the de-

nominator also favours Germany. In the early twentieth century around

half the German agricultural labour force was female; only 10% of the

British was (as recorded). Thus, dividing by the male labour force raises

estimates of German productivity in 1910 by 73% relative to Britain. It

also raises it against other countries: by 70% against the Netherlands,

by 26% against Denmark, and 16% against France, for example.21

20 In 1913 19.7% of the German arable area was down to root crops: three times theFrench proportion, and eight times the British, Perkins (1981), 80; this had risen from14.8% in 1883 and 2.3% in 1800, Schremmer (1988), 50.

21 Calculated from estimates in O’Brien and Prados de la Escosura (1992).

18004

5

6

7

89

10

20

30

50

40

1810 1820 1830 1840 1850 1860 1870 1880 1890 1900 1910

Sweden

Switzerland

Netherlands

Britain

Denmark Germany

Belgium

France

Austria-Hungary

Fig. 7.5. Bairoch’s estimates of agricultural productivity 1800–1910 (calories

produced per male occupied in agriculture)

Source: From Bairoch (1989).


So, if Bairoch’s estimates produce a distorted picture how can they be

improved? The Wrst step is to express output in some unit of value, to deal

with the problems created by the calorie measure. The output of diVerent

commodities in the various countries should be valued at uniform prices,

and combined as an index of production measured in some common unit

of account. Generally the unit of account is some monetary measure of

given purchasing power (a purchasing power parity or PPP). However,

one author (Van Zanden) has chosen to express all prices in terms of

wheat equivalents (thus a cow might be worth the equivalent of four tons

of wheat, a ton of sugar beet half a ton of wheat, etc.).

Table 7.6 gives a range of estimates from diVerent sources, indicating

the unit of account, and the productivity measure, which are used by

each study. The table concentrates on the comparison between Britain

and Germany. This table shows once again that the use of male labour

as the denominator aVects all these results. The use of value added as

the numerator is also important. This is because German agriculture

was more ‘self-suYcient’, it relied less on inputs from other sectors. So

‘value added’ is a higher share of total output.

The fact that British agriculture made much greater use of employed

labour, which was predominantly male, is one explanation of the diVer-

ent rates of female participation between the two countries. However,

Table 7.6. German agricultural productivity compared to the United

Kingdom, 1910–13

Gross output per head Value added per head

Both sexes Men only Both sexes Men only

O’Brien/Prados de

la Escosura (PPP)

56/58 97/100 77/79 134/136

Bairoch (calories) 127

Van Zanden

(wheat equivalent)

77 133

Fremdling (PPP) 42 51

Broadberry (PPP) 67

Source: This is a reworking of a similar table in O’Brien and Prados de la Escosura(1992): their estimates are combined in a range (German weights and British weights);Broadberry’s estimate is added, from Broadberry (1997); and Bairoch’s estimate is revisedfrom 106 (which came from an earlier study) to 127, from Bairoch (1989). Quite whyBairoch made this revision is not clear.


the diVerence between Germany and the Netherlands was almost as

great, and this is more diYcult to explain. Institutional diVerences were

less important. So it may be that the German occupational censuses

were rather more diligent in the recording of the labour contributed by

wives and other female family members than surveys in other countries.

If this is right, then comparisons which use both sexes may understate

productivity in German agriculture.

Looking at the diVerent estimates, those of Fremdling use a small

number of prices to establish the PPP: just the prices for wheat and beef

for the whole agricultural sector. These are probably unfavourable to the

German side (potatoes, pork, and rye might have produced a diVerent

Wgure). The estimates using purchasing power parity calculations and

value added per head are the ones which should be preferred on a priori

grounds. Discounting the Fremdling estimates, this leaves those by

Broadberry and O’Brien/Prados de la Escosura. There should probably

be some allowance for the superior coverage of female labour

in Germany. This suggests that the most likely level of productivity in

German agriculture was around 75–80% of the British level in 1910–13.

As far as other countries are concerned, estimates by O’Brien and

Prados de la Escosura (Table 7.7) show Germany to be ahead of France

(on most measures) but behind Denmark. The comparison with the

Netherlands is heavily aVected by the inclusion or exclusion of female

labour. Allowing for some under-recording in the Netherlands would

put productivity in Germany ahead of the Dutch level in 1910.

Between 1890 and 1910 Germany seems to have made gains relative to

Britain, France, and the Netherlands, but not relative to Denmark (where

the increase was exceptionally high). Van Zanden has constructed an

Table 7.7. Production and productivity in selected countries, as

compared to the British level, 1890 and 1910 (UK ¼ 100)

Output per worker Output per male

worker

Output per hectare

1890 1910 1890 1910 1890 1910

Netherlands 82 89 89 92 192 237

Denmark 44 82 82 147 140 202

France 52 72 72 82 128 136

Germany 63 89 89 118 148 205


index of total productivity (combining land, labour, and livestock). This

shows that total productivity rose by 1.53% per annum in Germany in

1870–1910, which is just ahead of Denmark (1.31%) and substantially

more than in the Netherlands (0.82%), France (0.46%), and Britain

(0.19%).22 Broadberry’s estimates also show German productivity rising

from 55.7% of the British level in 1871 to 67.3% in 1911.23

So a tentative conclusion is that productivity growth in German

agriculture in 1870–1913 was higher than in most other countries with

the possible exception of Denmark. This was partly due to the use of

techniques already established in other countries. One study has oVered

the following view of the increase in German production per hectare

1870–1913: it was 50% due to increased fertilizer use (including both

artiWcial fertilizers and livestock manure), 15% due to improved breed-

ing, and 35% due to improved husbandry and reduced losses, including

harvest losses.24 Breeding was mainly new technology, although some

crops and improved breeds came from foreign countries. Husbandry

improvements, on the other hand, often involved the use of machinery,

and livestock systems, which had been largely developed outside

Germany: drills, reapers, the stall-feeding of cattle. The major German

contribution was the development of a husbandry system for lighter

soils: the Lupitz system.25

This section will conclude with an attempt to compare crop nutrient

supply from artiWcial fertilizers to that available from livestock manure.

The use of artiWcial fertilizers was clearly new technology, and an area

where Germany was an acknowledged leader. The Wgures presented in

Table 7.8 make use of Wgures produced by Schremmer for the nutrients

supplied by artiWcial fertilizers. He also provided estimates of the total

tonnage of animal manure, but not the nutrient content. The table gives

the estimates of nutrient value, using Wgures for the analysis of animal

manure given by the German scientist, Augustus Voelcker, who became

22 Van Zanden (1991).23 Broadberry (1997), 18; Broadberry makes use of the 1937 benchmark comparisons

made by Rostas (1948); these incorporate a crude, but possibly justiWable, adjustment ofthe German agricultural labour force (removing three out of four wives); Broadberry alsouses HoVmann’s estimates of the agricultural labour force 1870–1913, thus incorporating(as does everyone else who uses HoVmann in this respect) HoVmann’s upwards adjust-ment of the German agricultural labour force in the 1870s and 1880s to allow for asupposed under-recording of female labour in the census of 1882. This has the eVect ofreducing estimated productivity growth in German agriculture.

24 Estimates given in Muller (1895).25 Brouwer (1957) and Dyer (1896).


consulting chemist to the Royal Agricultural Society of England and

Wales.

These Wgures should be regarded as approximations. Schremmer’s

Wgures are themselves estimates. The nutrient value of manure depends

on factors such as storage practice, time of application, losses through

leaching, and the way that animals are fed. Margins of error are likely

to be high.

Keeping these reservations in mind, the estimates show that, by the

years just before the First World War, artiWcial fertilizers were supply-

ing around 40% of the potash and phosphate applied, but only 16% of

the nitrogen (an additional source of nitrogen came from the use of

clover and other nitrogen-Wxing legumes in the rotation). But they did

represent a large share of the increase since 1878–80: over 50% for

potash and phosphate, and 23% for nitrogen. Total nutrient applica-

tions rose twice as fast after 1878/80 as in the period 1800–1878/80,

and this acceleration was entirely due to the use of artiWcial fertilizers.So, the contribution of new technology was, in this area, a major factor.

4. Regional Patterns of Productiv ity Growth

A problem which has restricted discussion of agricultural developments

in Germany has been the lack of estimates of regional productivity

levels. HoVmann’s accounts provide national Wgures, but no attempt

was made to provide a regional breakdown. Thus, it was not possible to

Table 7.8. Plant nutrient supply, 1800–1914, average annual

applications divided by the total agricultural area

1800 1878–80 1891–3 1898–1900 1911–14

A. Estimates of nutrients available from animal manure

Nitrogen (kg/ha) 5 14 19 33

Phosphate (kg/ha) 4 12 16 28

Potash (kg/ha) 4 11 15 26

B. Nutrients available from artificial fertilizers

Nitrogen (kg/ha) 0.7 2.2 6.4

Phosphate (kg/ha) 1.6 10.3 18.9

Potash (kg/ha) 0.8 3.1 16.7

Source: Schremmer (1988) and Voelcker (1883).


consider issues such as whether the problems of eastern agriculture

were due to poor productivity rather than low prices. The best way to

create Wgures for regional productivity levels is to construct agricultural

accounts for each region: dividing estimates of total net value added by

the labour force (in full-time labour units to allow for the eVect of part-

time employment) produces Wgures for labour productivity.

It was considered that the provision of regional accounts would be of

value, not only to this study, but also to others working on German

agriculture, so three sets were prepared, for 1880–4, 1893–7, and

1905–9. The details of the calculations have been published separately.

The results are given in Table 7.9, which shows estimates of labour

productivity in German agriculture for 21 regional units for three Wve-

year periods: 1880–4, 1893–7, and 1905–9. The table also gives the

Table 7.9. Productivity in German agriculture: estimates from regional

agricultural accounts in 1913 prices, 95% confidence intervals in brackets

Net Value Added per FLU (Full-time Labour Unit) Annual % rates

of growth

1880–4

(Marks)

1893–7

(Marks)

1905–9

(Marks)

1880/4–1905/9

East Prussia 480 (454–506) 745 (703–786) 934 (877–992) 2.71 (2.87–2.54)

West Prussia 648 (615–680) 923 (879–967) 1078 (1022–1134) 2.07 (2.18–1.95)

Berlin/Brandenburg 731 (697–764) 1001 (957–1045) 1222 (1163–1280) 2.08 (2.19–1.97)

Pomerania 852 (811–893) 1175 (1121–1229) 1433 (1363–1503) 2.11 (2.22–1.99)

Posen 636 (607–664) 891 (850–932) 1179 (1124–1234) 2.51 (2.63–2.38)

Silesia 550 (521–579) 765 (728–802) 960 (911–1009) 2.26 (2.38–2.13)

Pr. Saxony/Anhalt 1089 (1046–1131) 1356 (1304–1407) 1388 (1330–1445) 0.98 (1.03–0.93)

Schleswig-Holstein 1145 (1072–1218) 1323 (1243–1403) 1709 (1600–1819) 1.62 (1.74–1.50)

Hannover/Old./Bruns. 779 (737–821) 1075 (1023–1126) 1136 (1075–1197) 1.53 (1.62–1.43)

Westphalia 579 (544–614) 865 (822–908) 834 (787–881) 1.48 (1.58–1.37)

Hesse-Nassau 524 (497–552) 798 (760–837) 769 (727–811) 1.55 (1.65–1.45)

Rhineland 535 (504–565) 761 (723–799) 757 (713–800) 1.41 (1.50–1.31)

Bavaria excl. Pfalz 510 (484–536) 691 (654–729) 667 (623–712) 1.08 (1.16–1.00)

Pfalz 480 (458–501) 791 (756–826) 699 (664–735) 1.53 (1.61–1.44)

Saxony 789 (744–834) 1052 (999–1105) 1395 (1316–1474) 2.32 (2.46–2.17)

Wurttemberg/Hoh. 594 (565–622) 641 (608–674) 661 (619–704) 0.44 (0.47–0.41)

Baden 517 (492–543) 631 (601–661) 635 (599–671) 0.83 (0.88–0.77)

Hesse 685 (654–716) 949 (908–990) 1016 (965–1067) 1.60 (1.68–1.5)

Mecklenburg 1267 (1211–1323) 1442 (1380–1505) 1716 (1632–1801) 1.23 (1.3–1.16)

Thuringia 750 (717–783) 1029 (984–1075) 1088 (1034–1142) 1.50 (1.59–1.42)

Alsace-Lorraine 596 (571–622) 609 (580–638) 676 (639–713) 0.51 (0.54–0.47)

Germany 672 855 982 1.53


annual rate of increase 1880/4 to 1905/9. As the calculations involve a

degree of uncertainty, conWdence intervals have been estimated both for

the levels of productivity and for the rates of growth. The regional

units are those deWned in Appendix 3A.26

The Wrst point which emerges clearly from Table 7.9 is the high rates

of productivity growth in eastern Germany. Productivity growth rates

of over 2.0% per annum are impressive by any standards. While this

does not prove that there had been underutilized resources in these

provinces, this is certainly consistent with this explanation.

In Chapter 4 regressions were presented which showed that agricul-

tural productivity growth had a positive eVect on migration. The table

shows that the reason for this is the high rate of increase in the east, in

regions which also had high out-migration. There is a contrast with

southern and south-western Germany, which was also agricultural, but

where productivity growth was slower. Out-migration was also less.

This is the eVect which is picked up by the regression results.

If there had been a labour surplus in eastern agriculture then it could

have caused both high out-migration, as some of those who were under-

employed left, and high productivity growth, as others became more

fully employed. A relationship between high productivity growth and

high out-migration would be one consequence of this.

Looking at the maps of regional productivity (Figure 7.6), it is evi-

dent that, in the Wrst period, there was a small group of central regions

with high productivity: Schleswig-Holstein, Mecklenburg, and Prussian

Saxony. By 1905–9 these regions were still amongst the leaders, but the

gap with regions to the west and east had closed. The south and south-

west, on the other hand, were lagging.

Table 7.10 provides a summary of the changes in the relative position

of the east-Elbian provinces. Their share in the total labour force was

relatively static, as was the share in livestock production. The factor

which increased the share of total value added, and thus produced faster

than average productivity growth, was the gain in the share of total

production of major roots and cereals (sugar beet, potatoes, wheat,

barley, spelt, oats, and rye). It was the gains made by the arable sector

which drove the process of convergence.

Table 7.11 provides a decomposition of the increase in the production

of major roots and cereals, comparing the east-Elbian provinces with the

rest of Germany. The major part of the shift in the balance of production

26 Details of the methods used to construct these accounts are given in Grant (2002a).


1,300−1,400

1,200−1,300

1,400−1,500

1,600−1,700 1,000−1,100

1,100−1,200

800−900

700-800

>1,700

600−700

1,500−1,600 900−1,000

Productivity in 1880-4Net valve added per full-time labour unit in 1913 Marks

Productivity in 1905-9Net valve added per full-time labour unit in 1913 Marks

1,000−1,100

900−1,000

1,100−1,200

1,200−1,300 700−800

800−900

600−700

500−600

>1,300

<500

Fig. 7.6. Agricultural productivity by region


in favour of the east came about as a result of a faster increase in yields.

There was a contribution from an increase in areas, and another from a

composition eVect (the substitution of higher yielding, or higher value,

crops for others which were less productive or less valuable). But the

main factor was the strong tendency for yields in eastern Germany to

converge on the higher levels of central and western Germany.

What was the cause of this convergence? Clapham suggests that sugar

beet cultivation played a crucial role in making producers aware of the

importance of artiWcial fertilizers. One source which conWrms this is a

survey of fertilizer production and use in Prussia published in the 1887

Landwirtschaftliche Jahrbucher.27 This brought together reports from six-

teen regional research stations. It showed that, typically, fertilizer use

was two or three times as high for sugar beet as for other crops. It also

27 Thiel (1887).

Table 7.10 East-Elbian share in total production and value added

1880–4 1893–7 1905–9

Production of:

Major roots and cereals 33.5 37.7 43.3

Other crops 31.6 27.7 27.8

Livestock produce 33.0 32.2 33.3

Total net value added 32.8 33.7 37.6

Labour force in FLU 32.7 32.7 32.2

Source: Calculated from accounts given in Grant (2002a).

Table 7.11. Decomposition of change in production of major roots

and cereals 1880/4–1905/9

East-Elbian

Germany

Rest of

Germany

Total % increase in production 118.5 48.0

Increase in production per hectare 95.5 39.7

Increase in areas 11.8 5.9

Decomposition of increase in production per hectare:

Composition effect 2.9 0.9

Weighted increase in yields 89.8 38.1

Note: Calculations use 1905–9 prices.


showed that most fertilizer was supplied from Wrms within the region.

This suggests that sugar beet may have stimulated the use of fertilizers

in other ways besides the ‘demonstration eVect’ mentioned by Clapham.

The introduction of sugar beet into a region would raise demand, and

this might then lead to the construction of additional fertilizer factories.

Having a fertilizer factory nearby would then reduce transport costs and

so lead to increased use on other crops.

Sugar beet cultivation expanded faster in the east. Sugar beet rose

from 0.6% of the eastern arable area in 1880–4 to 2.2% in 1905–9,

which was a larger increase than in the rest of Germany (1.7% to 2.5%).

So the sugar beet eVect is one which promoted convergence. But the

pattern of convergence was much wider than this. Sugar beet cultivation

was limited to areas with suitable soil and climate (Figure 7.7). Initially,

the centre of cultivation was Prussian Saxony, especially the Magde-

burg region. From there it spread west and east. But certain regions

8−10

6−8

4−6

3−4

2−3

1.5−2

1−1.5

0.5−1

<0.5

>10

Fig. 7.7. The sugar beet regions: sugar beet as a percentage of the total

arable area, 1907

Source: Data from agricultural part of the 1907 occupational census, SDR n.f. 212.


never had much. These included the regions of southern Germany,

which had relatively low productivity, but also some more successful

regions: Westphalia, Schleswig-Holstein, and the Kingdom of Saxony.

East Prussia, which recorded the fastest productivity increase, had little

sugar beet cultivation.

Arable production in Eastern Germany caught up from a position

well below the national average in 1880–4. Table 7.12 gives the results

of some calculations based on the statistical analysis of yield variations,

using statistical techniques similar to those used elsewhere in this study.

These show that the spread of sugar beet eVect did indeed promote

convergence. But only 20% of predicted convergence is attributable to

this eVect. There was a more general process of convergence which was

not driven by the spread of sugar beet.28 Sugar beet cultivation was

only part of the catch-up story. Improved techniques were coming into

use in regions outside the sugar beet areas as well.

Statistical analysis can be used to examine whether the presence of

larger estates contributed to increases in yields. Large estates might

have been better placed to take advantage of the new technologies:

more aware of scientiWc advances, having easier access to sources of

Wnance. There is documentary evidence which suggests that larger

farms were more likely to use artiWcial fertilizers. For example, a report

on conditions in a Gutsbezirk in Kreis Stolp in 1890–1 observed: ‘In this

area, as in the whole region, it is only the large farms and estates that

make use of artiWcial fertilizers and concentrated livestock feed.’29

Table 7.12. Explaining the convergence of arable yields, east-Elbian

Germany compared to the rest of Germany, 1880–4 to 1905–9

Value of arable production

per hectare

(average 1880–4 ¼ 100):

Convergence

effect

Sugar beet

effect

Predicted

gain

1880–4 1905–9

East 78.7 172.0 þ27.6% þ24.0% þ103.9

Rest 116.7 172.8 �21.7% þ12.5% þ43.0

28 This calculation uses the results of regressions given in Grant (2002a).29 Landwirtschaftliche Jahrbucher, xix, erg. 4 (1891). Every year this publication in-

cluded a series of reports on conditions in various parts of Prussia, which are a usefulsource of information on the state of agriculture.


However, the analysis of yields at the Kreis level provides a more

nuanced view. The larger farms and estates maintained an advantage in

wheat yields, which was a crop grown largely for sale in urban markets,

and thus better suited to the more market-orientated enterprises. But

when rye yields were examined it was found that the advantage which

the larger holdings had had in the 1870s had largely disappeared by

1900; the smaller and medium-sized farms had caught up.30 The new

techniques had become more widely diVused.

Even if the larger holdings had an advantage when the use of artiWcial

fertilizers is considered, this was compensated by the greater livestock

intensity on smaller holdings, which meant that these holdings were

deriving more nutrients from animal sources and so had less need of

artiWcial fertilizers. The livestock censuses show that the value of live-

stock per hectare was higher on smaller farms. They also show that the

total value of livestock kept was rising faster on smaller holdings. Be-

tween 1882 and 1895 the total value of livestock on all German holdings

of less than 20 hectares rose by 17.7%, but it only rose by 9.5% on

holdings of over 100 hectares.31

So, the emphasis on the leading role of larger farms and estates may be

misplaced. Even in the east there were many small and medium-sized

farms. There were three east-Elbian provinces where holdings of under

100 hectares accounted for over 60% of the total agricultural area: East

Prussia (60.5% in 1895), Brandenburg (64.8%), and Silesia (66.1%).

Technological change, leading to increased yields, on these farms was an

important component in the success of eastern agriculture in the period.

The main reason for the improved performance of the arable sector

east of the Elbe was that the new technology, in particular the use of

artiWcial fertilizers, was particularly valuable in areas where the natural

level of soil fertility was low, or areas with free-draining sandy soils. On

the sandy soils of eastern Germany increased nutrient supply was par-

ticularly beneWcial. These were free-draining, ‘hungry’ soils, where the

rapid leaching of nutrients meant that continuous fresh applications

were needed to maintain soil fertility. Table 7.13 gives the results of a

survey of soil quality in the original Prussian provinces in the 1860s.

The diVerence between Prussia east and west of the Elbe is very notice-

able. In the west the predominant soil type was a heavier loam, with a

higher clay content, and, in consequence, much less free-draining. This

30 Details of the statistical analysis are given in Grant (2002a); there is a further discus-sion of this issue in Grant (2002e).

31 Rauchberg (1900), 590.


Table 7.13. Soil quality in the Prussian Regierungsbezirke, from a

survey in the 1860s

Loam Sandy loam Sand Moor Grsrt

I. East of the Elbe

Konigsberg 22.4 57.8 16.8 2.9 29.6

Allenstein 6.2 49.1 39.0 5.7 15.2

Gumbinnen 22.3 50.6 19.0 8.1 24.1

Danzig 27.6 38.5 25.5 8.3 24.7

Marienwerder 14.5 33.6 48.4 3.5 23.3

Potsdam 11.6 44.8 33.3 10.3 34.6

Frankfurt 12.7 27.7 50.6 9.1 33.9

Stettin 4.9 52.0 30.2 12.9 35.7

Koslin 8.9 36.2 47.1 7.8 21.7

Stralsund 1.4 70.3 15.4 12.9 58.9

Posen 11.6 49.9 32.8 5.7 25.2

Bromberg 7.7 42.5 39.2 10.6 26.0

Breslau 48.3 24.6 25.1 2.0 53.8

Liegnitz 32.0 28.8 36.3 2.9 42.6

Oppeln 31.2 33.5 33.5 1.8 36.6

Average 17.9 40.9 34.7 6.5 31.3

II. Other Prussian provinces

Magdeburg 37.0 20.1 35.7 7.2 61.4

Merseburg 66.2 11.4 22.1 0.4 89.7

Erfurt 69.8 25.4 4.7 0.1 60.7

Munster 29.2 16.5 44.5 9.8 53.3

Minden 55.0 12.3 29.2 3.5 59.1

Arnsberg 93.3 3.6 3.1 0.0 52.9

Koblenz 90.1 9.9 0.0 0.0 53.5

Dusseldorf 41.4 38.4 16.0 4.2 96.2

Koln 68.4 23.3 7.5 0.8 88.7

Trier 65.6 22.3 12.0 0.1 38.0

Aachen 71.5 18.2 5.4 4.9 78.4

Average 60.8 17.3 18.8 3.1 65.3

Note: Percentage distribution of the land area (excluding areas under water) by soil type,together with the average level of Grundsteuerreinertrag (Grsrt) for cultivated ground, inMarks per Hectare.Source: Data from Meitzen (1868).

accounted for over 90% of the area in Koblenz and Arnsberg. In the east

the predominant types were the lighter ‘sand’ and ‘sandy loam’. Outside

Silesia, heavier soils were comparatively rare.


The land tax assessment (Grundsteuerreinertrag) was closely linked to

soil quality. The Wnal column gives the average assessment for each

Regierungsbezirk. The level east of the Elbe was less than half the aver-

age for the remaining Prussian provinces, conWrming the poorer quality

of the eastern soils.

Writers on agricultural matters, inside Germany and outside, con-

sidered that the major achievement of German agriculture in the period

was the raising of yields on the lighter, sandy soils. The ‘Lupitz

system’, developed by the Mecklenburg farmer A. Schultz-Lupitz, at-

tracted the attention of English agriculturalists.32 There was govern-

ment support for research into yields on lighter soils. Work at

experimental farms, and Weld trials conducted by the local Land-

wirtschaftsgesellchaften, demonstrated the value of fertilizers to local

farmers. The results were some spectacular gains. Average rye yields in

Kreis Uelzen in the Luneburg Heath rose 205% between 1882–4 and

1910–14, from a position 29% below the national average to one 15%

above.33 This meant that moorland areas, and areas which had ex-

tremely light sandy soils, which had previously been left uncultivated,

could be brought into production.

ScientiWc research in the early part of the century, in particular the

discoveries of Justus von Liebig, had led to a greater understanding of

plant nutrient requirements and the possibilities of providing these

through artiWcial fertilizers. There was, however, no simple relationship

between the discoveries of Liebig and their application to the poorer soils

of northern Germany. To begin with, practical, and cost-eVective, means

of supplying soluble nutrients to plants had to be found. The Wrst suc-

cessful ‘artiWcial fertilizer’, superphosphate produced by the application

of sulphuric acid to mined phosphate, was developed by Gilbert and

Lawes in England, making use of Liebig’s suggestion that the value of

bonemeal could be improved by treatment with sulphuric acid.34

32 Brouwer (1957) describes the work of Schultz-Lupitz; his system was introduced toEnglish readers in an article in the 1896 Journal of the Royal Agricultural Society ofEngland and Wales, Dyer (1896).

33 Figures from Schneppe (1960), who also gives details of government-sponsoredresearch on the light soils of the Luneburg Heath. It was a general complaint of Germanagriculturalists that the quality of German soil was poor. See, for example, the remark byDr H. Gerlich at a meeting of the Royal Statistical Society in 1899, Journal of the RoyalStatistical Society, 62/4 (1899). Sandy soils have advantages and disadvantages, highfertilizer requirements being undeniably one of the latter. The author of this study farmson soils of this type in central Sussex. See also Thiel (1887) for details of fertilizer use inthe Prussian regions.

34 Frauendorfer (1956). There is a tendency in writings on nineteenth-century agricul-tural technology to focus on developments within the boundaries of the state to which the


But then these ideas and new products had to be incorporated in a

practical farming system, and this took years of experimentation.

Schultz-Lupitz, the son of a pharmacist whose university education

made him aware of the importance of Liebig’s discoveries, was inspired

by this to buy a 250-hectare farm in 1855. But it was only in 1879–80

that his experiments with diVerent fertilizers and rotations produced a

system which consistently raised yields. The value of Liebig’s discover-

ies lay in the conWdence they gave to innovative farmers, and to re-

searchers at experimental stations, that ways could be found to raise

fertility on poorer soils. Without this conWdence, Schultz-Lupitz would

never have engaged in twenty-Wve years of initially fruitless work.35

A second major advance in the period was improvements in yields

through breeding. Again, German successes in this area attracted atten-

tion in England.36 New strains of wheat and rye were developed. But

perhaps the greatest success was the raising of the yield and sugar con-

tent of sugar beet. This last was increased from 8.5% in 1870 to 16% in

1914, an important contribution to the success of this industry.37

New productive knowledge was an important part of the success of

German agriculture in 1870–1913. Traditional agriculture was ‘trans-

formed’: new techniques were introduced, as were new crops and new

inputs. But, equally, there was a contribution from old technology as

well. Much of the increase in total plant nutrients came from increased

livestock numbers, deWnitely an old technology. The convergence of

yields within Germany may have been due to new technology, but the

rise in average German yields relative to other European countries

owed a lot to rising livestock intensity.

One question which arises is how this relatively favourable view of

German agriculture can be reconciled with the evidence of high debt

levels and Zwangsversteigerungen (forced sales of agricultural property).38

writer belongs, and which were therefore published in the writer’s own language: oneexample is Fussell (1962). Frauendorfer’s article is an exception. It gives full weight todevelopments in Britain, France, and Germany, and is good on interactions betweenresearchers in these countries. The conclusion which emerges is that, while France andGermany led in basic scientiWc research, British research on the practical application ofthese ideas was of considerable importance. Britain did not lag in the development offertilizer technology.

35 Brouwer (1957).36 Evershed (1884).37 Muller (1985).38 ZdKSB (1905), p. xlii; Zwangsversteigerungen were reported annually in the ZdKSB

in the 1880s and 1890s.


But a distinction can be drawn between crises of agriculture, which aVect

the whole sector badly enough to bring about a fall in the total area

under cultivation, and crises which aVect individual farmers who have

over-borrowed or paid too much for assets whose value has subsequently

fallen. In crises of the latter type, land does not go out of production,

as the assets of bankrupt farmers are bought up by other farmers, who

are less indebted or more eYcient. The ‘crisis’ of German agriculture

in the 1890s was of this latter type. There was a genuine agricultural

crisis in Britain in the 1880s and 1890s, and again in the 1930s. In

these periods land did go out of production. This is the acid test. Un-

doubtedly, some estate owners did conform to the stereotype found in

many works on east-Elbian agriculture, with an addiction to ‘ruinous,

late-feudal, conspicuous consumption’ which led to increased indebted-

ness.39 But east-Elbian estates considered as a whole did not deserve this

criticism.

Germany 1870–1913 is an excellent example of Schultz’s view of

agricultural transformation. Germany is also a good example, one might

almost say a model, of how this can be done: improved agricultural

education, the setting up of research stations, investments in scientiWc

research, trial plots and other demonstrations organized through local

farming societies, good interaction between researchers and innovative

farmers. In all these areas German agriculture was a leader in European

agriculture in 1870–1913.But the consequence of this was that the declining share of the home

market available to German farmers, after imports had been deducted,

could be supplied by a smaller labour force. So, this also contributed to

out-migration. And the more modern farming techniques reduced the

variability in crop yields, which meant that institutions which provided

insurance against crop failure were less necessary. So, improved tech-

nology led to institutional change as well.40

5. Rural Underemployment , Labour-Saving

Technology , and M igration

There is no evident labour-saving bias to the technological changes dis-

cussed in the last section. Indeed, many of the changes in German

39 Puhle (1972), 16.40 The decline in the variability of crop yields is discussed in Grant (2002c).


agricultural practice should have raised labour requirements: the area

under root crops was rising and these required more labour, livestock

numbers were rising, and there was a shift from sheep (less labour inten-

sive) to stall-fed cattle and pigs.41 What was impressive was the ability of

German agriculture to deliver a large increase in food production,

1870–1913, with a relatively small increase in the labour force. This is

consistent with the view that there was initial underemployment.

There was one area in which labour-saving technology was intro-

duced, and that was equipment for harvesting and threshing. Some

authors, including Max Weber, have argued that the introduction of

steam-powered threshing machines reduced out-of-season labour re-

quirements and led to out-migration. However, the statistical analysis

showed that steam-powered threshing machines were not associated

with reduced agricultural employment. This section will look brieXy at

the reasons for this Wnding.When considering the eVect of machinery it is important to exclude

the eVect of farm size. Larger farms do use more machinery, but, in this

period, they were also intensifying production and introducing more

labour-intensive crops. The question of the eVect of farm size on labour

requirements was the main issue examined in the previous chapter.

To exclude the farm size eVect, a standardized index of access to

steam-threshing was constructed, which used results from the 1907

census for the use of this machinery by size of farm. These results were

then weighted by the average size distribution for the whole country,

not the regional distribution. The index therefore gives the answer to

the hypothetical question, ‘What percentage of farms would have used

steam-threshing machines if the size distribution had been the same as

the national average?’ The results are shown in Figure 7.8.

There are some considerable diVerences. The range is from an index of

12 to one of 68 (100 would indicate that all farms used steam-threshing).

There was a clear regional pattern. The use of steam-threshing was

highest in central Germany, in Hesse, Brunswick, and Franconia, and in

parts of Westphalia and Hanover. It was not high in the east, using this

standardized measure.

For a study primarily concerned with migration the important point

to note from Figure 7.8 is that few of the areas identiWed as having high

41 Figures given in Muller (1985) give the labour requirement of a sugar beet rotationas 97% higher than a Norfolk four-course rotation, and 346% higher than a three-Weldsystem.


use of steam-threshing are areas of high out-migration. So it is not

surprising that migration regressions, or regressions with the change in

the agricultural labour force as the dependent variable, do not Wnd that

steam-threshing displaced labour and led to migration.

By and large, there was not a major labour-saving eVect from changes

in agricultural technology in the period. The livestock sector remained

almost entirely unmechanized apart from mechanical milk separators.

All feeding was still done by hand. Most arable improvements increased

yields rather than displacing labour. Labour would only have been

displaced if the area under cultivation had fallen in consequence.

The rises in stock numbers shown in Table 7.7 should have required

a substantial amount of additional labour. The fact that this was not

required suggests that there had been underemployment. It is hard to

Wnd any improvement in technology which might have enabled the

number of pigs to increase by 260% without using more labour if

labour had been fully utilized in 1873.

10−20

20−30

30−40

40−50

50−60

60−70

Fig. 7.8. Standardized index of the use of steam-powered threshing machines,

by Regierungsbezirke, 1907 (100% use ¼ 100)

Source: Calculated from data in SDR n.f. 212.


Contemporary reports speak of underemployment, particularly

amongst family members. One report, on conditions in a Gemeinde in

Silesia in 1886, mentions that farmers are seeking to rent more land to

make more use of family labour ‘which otherwise would not be fully

utilized’.42 The same report also recorded that the cost of poor relief in

the Gemeinde had risen because of the decline in cottage industry.43

This is supported by the regression results given in Chapter 6, whichshowed that the decline of the agricultural population was higher in

areas where there had been cottage industry. The decline in this source

of employment created underemployment in some areas.

A direct indicator of underemployment might be the ratio of popula-

tion to agricultural land in rural areas. If underemployment leads to

migration there could be a positive relationship between this variable and

migration. This relationship cannot be found at the Kreis level, but there

are considerable variations between Kreise in land quality and in

the availability of non-agricultural employment. Data on migration at the

level of the Gemeinde, or village, are not generally available from

the oYcial statistical sources, but a study of migration in the 1880s,

published in 1890 in the Prussian Landwirtschaftliche Jahrbucher, pro-

vided data for 720 villages in seven rural Kreise. The analysis made use of

the indirect method for estimating rates of migration by decade (compar-

ing the natural rate of increase from recorded births and deaths with the

actual increase as shown by the census returns). The econometric analy-

sis is given in Appendix 7A. It shows that population density, the rela-

tionship between the total population and the agricultural area, had a

strong eVect on migration: 30% of the recorded variation in rates of out-

migration in these rural areas could be explained by this factor alone.

Table 7.14 provides a summary of the results by showing the pre-

dicted rates of out-migration for various Gemeinden. The predicted rate

for a village in the top decile ranked by population density is 7.8 times

that of a village in the bottom decile; the predicted rate for a village in

the top quartile is 4.6 times that for one in the lowest quartile.

This simple analysis is suYciently strong and clear to make it evident

that there was overpopulation and underemployment in parts of rural

Germany in the 1870s and 1880s, and that this was a major inXuence on

migration. There was surplus labour. The validity of this aspect of the

Lewis Model is conWrmed.

42 Landwirtschaftliche Jahrbucher, xvi, erg. 2 (1887), 54. 43 Ibid. 76.


It is extremely diYcult to produce an estimate of the scale of the

rural labour surplus in Germany in the period. Much depends on the

deWnitions used. Are underemployed family members treated as part of

the surplus? Is low productivity employment, in cottage industry or on

family smallholdings, regarded as a form of underemployment? Modern

development economists, using surveys of work patterns, have come up

with a wide range of results. Estimates of underemployment in develop-

ing countries have been as high as 70% (for Peru). But a more typical

result is around 20–37%. Much depends on the methods used.44

Rural underemployment is a diVerent concept from formal, measur-

able unemployment in societies which pay unemployment beneWt. It is

not static. It is aVected by changing social and technological conditions.

So it is inevitably diYcult to deWne and measure.

In nineteenth-century Germany the amount of surplus labour was

aVected by changes in the terms of trade, in particular by the fall in

imported grain prices to 1897 and the subsequent recovery, and by

changing conditions within agriculture, by institutional developments, and

by new agricultural technology. The introduction of new technology

meant that a rising urban population could be fed by a rural workforce

which was static or falling, thus making it possible for the demographic

surplus in rural areas to move out of agriculture into industry. Migration

out of agriculture was, therefore, part of a dynamic process, as agriculture

adjusted to the new conditions created by industrialization, and was in

turn transformed by the eVects of industrialization.

Table 7.14. Summary of results from the analysis of migration from

720 rural Gemeinden, 1880–9: predicted rates of out-migration by

decade per 100 head of resident population

Ranking of Gemeinde by

population density

Population density

(population relative to

total agricultural area)

Predicted rate of

out-migration

Top decile average 4.06 16.99

Top quartile average 2.68 12.49

Average all Gemeinden 1.27 6.61

Lower quartile average 0.47 2.72

Lower decile average 0.37 2.18

Source: All data from Kaerger (1890).

44 The 70% Wgure is from a CIA study, Todaro (1997), 244.


Appendix 7A. Econometric Analysis of M igration in

Rural Gemeinden

The data set was derived from an article in the 1890 Landwirtschaftliche

Jahrbucher and analysed in the regressions shown in Table 7A.1. There

were seven Kreise represented, mainly in the rural east. The article

looked at Die Sachsengangerei, the pattern of migration from rural east-

ern Kreise to work on the sugar beet Welds of Provinz Sachsen (Prussian

Saxony).

The results show a consistent eVect from population density. The

relationship is not a linear one, as is shown by the signiWcance of the

quadratic term in column b. The predicted eVect of rising population

density is large. A village with population density 70% greater than the

average is predicted to have a rate of migration 42% above the average;

one with twice the average population density is predicted to have a rate

of out-migration 81% above the average (using the coeYcients from

column b).

As a check on the results the Wnal column gives results obtained

when the regression equation includes a full set of Kreise dummies.

Table 7A.1. Analysis of migration from rural Gemeinden, 1880–9:

dependent variable is rate of out-migration per 100 head of resident

population

No Kreise dummies used Kreise dummies used

a b c

Constant 2.20 0.20 �0.78

POPDENSITY þ3.10 þ5.48 þ4.20

(16.27) (12.23) (10.11)

POPDENSITYSQD �0.332 �0.241

(7.83) (4.70)

N 720 720 720

S 6.20 6.06 7.31

R2 0.269 0.302 0.470

adj. R2 0.268 0.300 0.462


RSS 27760 26311 20002

Source: All data from Kaerger (1890).


This removes any eVect from the location of the Kreis, and other factors

operating at this level. There is a reduction in the size of the estimated

coeYcients, as would be expected, but the results remain signiWcant.

Variations in population density at the village level were important

factors in migration.


8

Migration and Urban Labour Markets

1. Introduction

Lewis’s description of his model as a model of economic development

with unlimited supplies of labour has, naturally, tended to focus atten-

tion on labour supply. Yet, as pointed out in Chapter 1, the assumptions

he made about the supply of capital to industry are also important. His

model could be described as one in which there is a shortage of capital

instead of one in which labour is abundant.

The assumption that ‘only capitalists save’ is an extremely powerful

one in the context of a developing economy where an increase in the

overall rate of investment is required. EVectively, if capitalist savings

are a function of the level of capitalist income and the anticipated rate

of return on new investments, then one of these (or both) must rise if

there is to be an increase in the rate of investment. If borrowing from

the rest of society (through the banking system) is precluded, then it is

likely that there will be a shift in the distribution of income in favour of

capitalists under this assumption.1

A more realistic form of the assumption is that ‘only capitalists

invest’ and that they use savings out of their own incomes, or those of

their family and friends. This has two implications: Wrst, that know-

ledge of potentially proWtable investment opportunities is not widely

diVused—other groups may save, but they are unable to achieve the

high rates of return available on new industrial investments. Secondly,

the cost of Wnancial intermediation is high. Bank Wnance for industry is

either unobtainable or prohibitively expensive, and industrialists have

1 This is not a watertight theoretical proposition. It is possible that the prospect ofhigher returns on new investment might induce capitalists to raise savings out of anunchanged share of national income, and that the fact that the rate of return on newinvestment had risen (which would tend to raise the proWts share) might be balanced by afall in the rate of return on existing investments as a result of the introduction of newtechnology: a Schumpterian ‘creative destruction’ process. However, the assumption ‘onlycapitalists save’ does make it considerably easier to construct models in which there is ashift in the distribution of national income in favour of capitalists during the early stagesof industrialization. Kaldor (1956) and Pasinetti (1962) model this eVect.

little access to pools of savings that may be available elsewhere in the

economy.

However, Wnancial institutions and corporate structures are not

static, and these points, which may be a reasonable description of con-

ditions in the early stages of industrialization, should have less relevance

in a mature economy. The modern corporation, or joint-stock company,

brings together technically proWcient managers, who have knowledge of

investment opportunities but lack substantial savings of their own, and

investors who have savings but not specialized knowledge. A developed

banking sector, with access to integrated capital markets, will bring

down the cost of intermediation and draw upon additional sources of

savings.

In Germany in 1870–1913 there were important changes in corporate

structure and in the Wnancial system which enabled industry to draw

upon savings from outside the limited group of industrial capitalists,

and their families and friends, which had often been the main source of

Wnance for new industrial projects in the early part of the century. The

‘labour surplus’ phase of industrial development was brought to an end

as much by this as by changes within agriculture or the diVerent phases

of the demographic transition.2

This chapter will consider the pace of change, paying particular at-

tention to the role of the Wnancial markets. There were diVerent phases

in the development of the Wnancial sector which had important conse-

quences for the availability of funds for industrial investment. The

analysis is set in the context of a general discussion of the relationship

between migration and conditions in urban labour markets. For there

were other changes in the structure of the urban, or industrial, sector

which aVected migration. The most importance of these was the eVect

of Germany’s external trade position. As Germany became more inte-

grated in the world economy, the importance of the external sector rose

and the German economy became more specialized.3 Changes in the

international terms of trade, and in the structure of German exports,

2 Lindenlaub and Kohne-Lindenlaub (1991) provide an analysis of the early Wnancingof the Krupp enterprise. The view that there was a capital shortage in Germany has beenattacked by Borchardt (1982). The evidence he presents is, however, consistent with theview that there was no shortage of savings, but that there was a requirement for insti-tutional developments to turn these into industrial investments. Walter Rathenau alsocriticized the view that there was a capital shortage in Germany, Pogge von Strandmann(1988a).

3 Germany had always been inXuenced by foreign trade, but the transport improve-ments of the nineteenth century greatly increased the importance of the trade sector.

254 Migration and Urban Labour Markets

had consequences for urban labour markets, and in turn aVected the

Xow of migration.

This chapter also considers some of the consequences of migration

for urban labour conditions: the extent to which it stabilized urban

markets (reducing registered unemployment), the eVect on wage vari-

ations, the eVect of skill transfers, and the impact on the relative

position of women. The question of the eVect on overall income distri-

bution is considered in Chapter 9.

2. Structural Changes in the German

Economy 1870–1913

Table 8.1 gives Wgures for aspects of the German economy, comparing

Wve-year averages, and bisecting the whole period. The early 1890s is animportant dividing point; after this the growth rate of per capita GDP

was 0.5% higher than in the earlier period.

In this second period the rate of investment was also signiWcantly

higher, and this results in a faster rate of increase in the capital stock

per industrial worker (including mining). The acceleration was a

marked one: the rate of increase is 74% higher in the second period.

The importance of foreign trade was also rising. In the Wrst period the

increase in export volumes was only slightly faster than the increase in

GDP; in the second period it was considerably faster. In both periods

Table 8.1. Structural change in the German economy (all Wguresuse 1913 prices)

1870/4–1890/4 1890/4–1909/13

Annual rate of increase of GDP 2.2% 3.2%

Annual rate of increase of per capita GDP 1.3% 1.8%

Annual rate of increase of total capital

per industrial worker

1.5% 2.6%

Annual rate of increase of exports 2.4% 5.1%

Annual rate of increase of exports of

manufactures

3.6% 5.8%

Note: The export-orientation of the pre-1914 economy was high even by the standardsof the post-1945 period: Abelshauser (1983), 148 and 193.Source: All Wgures are calculations from HoVmann (1965).

Migration and Urban Labour Markets 255

the increase in the export of manufactured goods (Wnished and semi-

Wnished) was faster than the increase of total exports.

The division into two periods is a simpliWcation of a rather more

complex pattern, in which Xuctuations in the international economy

also played an important part. Figure 8.1 gives Wgures for total net

investment, and net industrial investment, as percentages of net domes-

tic product. There were Wve main periods. The Wrst was a strong boom

in the early 1870s, dominated by railway investment. This was followed

by a sharp recession which depressed investment into the 1880s. The

third period, which lasted to the mid-1890s, saw a moderate recovery;

boom conditions were not restored until the so-called Elektroboom of

the late 1890s.4 The Wnal period, after 1900, was one of considerable

Xuctuations, but less violent than either the boom of the 1870s or the

4 Hentschel (1978) uses the term ‘Elektro-boom’.

18700

10

20

Rolling five-year averages

1874 1878 1882 1886 1890 1894 1898 1902 1906 1910

Industrialinvestment

All investment

Fig. 8.1. Net investment (all sectors) and net industrial investment as

percentages of net domestic product, 1870–1913 (all Wgures use 1913 prices)


Elektroboom. They were the normal trade cycle Xuctuations of an indus-

trial economy. The level of total investment remained substantially

higher than in the period before 1895.5 Even in recessions it did not

sink back to the level of the 1870s.

One important source of structural change was the development of

the Wnancial sector. The collapse of the boom of the 1870s did consider-

able damage to the German Wnancial sector. Share prices fell 64%between 1872 and 1877, which greatly reduced the ability of companies

to raise money from the stock market. In 1871–3 a total of 2,781 million

Marks was raised from company Xotations; in 1874–9 just 284 millions.6

This increased the importance of bank Wnance, but the banking sector

had its own problems, stemming from the large number of bank Xota-

tions and formations during the boom. In 1870–4, 103 banks were

founded; 29 of these failed in 1873–4, accounting for a third of all

liquidations (by capital size) during the crash.7 There were further

liquidations and consolidations during the rest of the decade. Of the 36

banks Xoated in 1871 only 15 survived to 1880, of the 35 Xoated in 1872

just 10, and not one of the Wve Xoated in 1873.8

The consolidation of the banking system which followed gave the

German Wnancial system a distinctive character. Between 1870 and

1888 the Deutsche Bank, the largest bank, absorbed 71 other banks.9

The other major joint stock banks, or Kreditbanken, also made acquisi-

tions. By 1890 there were two other banks of similar size (with issued

capital of 75–80 million Marks): the Disconto-Gesellschaft and the

Darmstadter Bank. The Dresdner was slightly smaller (issued capital of

60 million Marks). Together with the Berliner Handelsgesellschaft and

the Schaaf’hauschen Bankverein (50 million and 36 million Marks re-

spectively) these six joint stock banks dominated industrial banking, as

far as the larger companies were concerned.10 Their inXuence was par-

ticularly strong in the new issue market.

HoVmann has produced Wgures for the Wnancing of investment

which, given the gaps in the available Wgures, are best regarded as

illustrative rather than deWnitive. In particular, little is known about the

5 Many of these phases reXected changes in international conditions. Internationaldevelopments are described in Milward and Saul (1977).

6 Wehler (1995), 554–5.7 Tilly (1986), 124.8 Pohl (1976), 26; these Wgures refer to stock market Xotations, Tilly’s to foundations.9 Wehler (1995), 628; Bohme (1966), 347.

10 Jeidels (1905), 66.


contribution made by companies which were not Aktiengesellschaften

(joint stock companies whose accounts are available). Table 8.2 uses the

available Wgures, and expresses them as percentages of net domestic

product.11

The table shows that the rise in industrial investment accounted for

most of the increase in total investment after 1895. The increase of 3

percentage points was also matched by a similar rise in recorded sources

of Wnance, and this, in turn, can be shown to have resulted from the

increased contribution of two sectors: the Aktiengesellschaften and the

banking sector.

The Wnal section of the table breaks down the contribution of the

banking sector. This shows that the rise in the position of the Kredit-

banken, from 19% of the total banking sector to 29%, was an important

factor in the overall increase. But smaller banks and Sparkassen were

also important sources of Wnance, particularly to smaller and medium-

sized companies. They raised funds from smaller investors, which were

11 Calculated using data from HoVman (1966), 812–13; HoVmann’s title for his table is‘The Wnancing of domestic net investment’, though there are enough uncertainties tomake this a rather optimistic title. HoVmann separates the activities of railway companiesfrom other joint-stock companies, so these are included in ‘other sources’. One justiWca-tion for this is that a number of railway companies passed into state ownership in thisperiod, resulting in large payments to shareholders, and making it diVicult to disentanglethe private and the public sector in this industry.

Table 8.2. Investment and sources of Wnance, 1870–94 and

1895–1913; all Wgures are expressed as percentages of net domestic

product (current prices)

1870–1894 1895–1913

All net investment 11.51 14.66

of which: net industrial investment 3.63 6.49

All recorded sources of Wnance of which: 7.12 10.5

Banks 3.04 5.56

Aktiengesellschaften 0.77 1.61

Other sources 3.35 3.34

Contribution of banking sector

Kreditbanken 0.58 1.61

Other banks 2.46 3.95

Source: All Wgures from HoVmann (1965).


then placed in the Wnancial markets, thus contributing to the overall

pool of available funds.

As before, the division into two periods is a simpliWcation which ob-

scures more complex trends. Figure 8.2 gives the available Wgures for

sources of Wnance. The ‘all sectors’ series shows a gradual recovery from

the trough of the mid-1870s, reaching a plateau of around 10% of NDP

after 1900. But the Wgures also show the process of structural change: the

rise in the importance of the banking sector, and the slow revival of the

contribution of the Aktiengesellschaften, from a negative position in the

late 1870s. These two sectors dominate the period of the Elektroboom.

These Wnancial statistics show how damaging the consequences of

the 1870s boom and recession were to the structure of the German

capital markets. Recovery was a long, slow process. They also show the

importance of the institutional changes which eventually overcame this

legacy. Somewhere around 1895 the process of recovery and rebuilding

1870-2

0

2

4

6

8

10

12

14

1875 1880 1885 1890 1895 1900

Aktien gesellschaften

sectors

Other financial

Banks

All sectors

1905 1910

Fig. 8.2. Sources of Wnance 1870–1913

Note: All Wgures are rolling Wve-year averages expressed as percentages of net domesticproduct, current prices.


was complete. The German economy moved out of a ‘capital shortage’

phase and into a phase in which buoyant investment played a leading

role in economic growth.12

A second major area of structural change lay in the external sector.

This was inXuenced by similar technological developments to those

which lay behind the Elektroboom: the growth of technologically ad-

vanced industries in which Germany had a comparative advantage.

This contributed to the acceleration of economic growth after 1895, and

thus to the ending of the ‘labour surplus’ period.

There was a shift in the composition of German exports (Figure 8.3):

exports of manufactured goods (Wnished and semi-Wnished) rose relative

12 The development of the German banking sector has been a subject of debate sincethe work of Riesser (1911) and Jeidels (1905). The views of Gerschenkron (1962) providedfurther stimulation. More recent contributions include Pohl (1976, 1984, and 1986), Neu-berger and Stokes (1974), Tilly (1986), Edwardes and Ogilvie (1995), Fohlin (2002),Feldenkirchen (1982), and Wellhoner (1989).

19101906190218981894189018861882187818741870

6000

5000

4000

3000

2000

1000

0

Finished goods

Agricultural goods

Semi-finished goods

Raw materials

Fig. 8.3. German exports by category, in million Marks (1913 prices)

Source: All Wgures calculated from HoVmann (1965), 530–5.


to other exports. In 1870–4 manufactured goods accounted for 52.5%

of exports; by 1890–4 this was 65.3%, and 74.2% in 1909–13.

The composition of the most important sector, Wnished goods, also

changed (Figure 8.4). Exports of Wnished goods in 1913 can be divided

into Wve roughly equal categories: textiles (1,150 million Marks), metal

goods excluding machinery (1,135 million Marks), chemicals and related

products (1,042 million Marks), machinery (1,025 million Marks), and

other Wnished goods (1,007 million Marks). Of these, other Wnished

goods showed the slowest rate of growth, but was the largest category in

the 1870s. Textile exports rose rapidly between 1870 and 1887, but then

remained relatively static until 1900. The fastest rates of growth were

recorded by the other three categories. Between 1894/6 and 1911/13 the

annual rates of increase were: 8.5% for chemicals and related goods,

8.7% for metal goods excluding machinery, and 15.3% for machinery.

The rapid growth of exports was made necessary by an adverse shift

in the German terms of trade after 1895. This is shown in Figure 8.5.

Other industries

Textiles

MachineryMetal goods

Chemicals

19101906190218981894189018861882187818741870

1200

1000

800

600

400

200

0

Fig. 8.4. German exports of Wnished goods, by category, in million Marks

(1913 prices)


HoVmann’s index of export prices divided by import prices fell by 22%

between 1894/6 and 1911/13. This was partly due to the recovery of

world food prices, but another factor was a fall in the prices of German

manufacturing exports relative to the cost of raw material imports,

which was steeper (32% over the same period).13

Although there was a general rise in raw material costs in this

period, German industry was aVected more than others. HoVmann’s

index of German manufactured export prices divided by the prices of

imported manufactures fell by 30% during the period 1895–1913.

Foreign-produced manufactures rose in price while the prices of

13 The adverse shift in the terms of trade was one reason why German wages did notconverge, or did not converge strongly, with those in Britain and the United States. Thereare a range of wage indices for Germany in the period, and some show rather strongerwage growth. The main studies are Kuczynski (1945), Grumbach and Konig (1957),Gommel (1979), Desai (1968), Bry (1960), Orsagh (1969), and Hohls (1995).

Manufactures - raw materials

All exports - all imports

19121910

19081906

19041902

19001898

18961894

18921890

18881886

18841882

1880

160

150

140

130

120

110

100

90

80

Fig. 8.5. The terms of trade of the German economy 1880–1913

Note: 1913 ¼ 100; export prices divided by import prices and manufactured export pricesdivided by raw material import prices.Source: From HoVmann (1965), 548.


German-produced goods were static. The average cost of raw material

imports rose by a half between 1895 and 1913. The cost of imported

food rose by a third.

From the German point of view this was a serious development.

Calculated over the whole 1870–1913 period, the income elasticities of

food imports and imports of raw materials were 1.31 and 1.58 respect-

ively.14 The combined volumes of these two items would be expected to

rise 45% faster than the rate of growth of national income. As these

categories accounted for 73.5% of Germany’s imports in 1894–6 it is a

fairly simple calculation (assuming that other categories of imports rose

in line with GDP) that, faced with the adverse price movements experi-

enced in 1895–1913, German exports would have to rise 80% faster

than the growth of overall national income to maintain balance of pay-

ments equilibrium. The required rate of increase for 1895–1913 was

5.7% per annum.15

Maddison’s Wgures for the period show that world trade grew at an

annual rate of 3.9% in 1870–1913.16 So the implication of this calcula-

tion is that Germany would have to make substantial gains from other

exporters to maintain a rate of growth of per capita GDP of around

1.8% per annum. Given the resources available to the German econ-

omy, it is clear that the main target would have to be Britain.

This provides a rather diVerent perspective on the Anglo-German

trade rivalry of the period before 1913. From the German point of

view, Germany’s economic success in 1870–1913 consisted of a rate of

growth which still left per capita GDP 24% below the British level in

1913 (having been 41% behind in 1870).17 German wages in 1905 were

83% of the British level at current exchange rates, and between 71%

and 75% of the British level in real terms, according to the Board of

Trade survey of that year. Phelps-Brown and Browne have pointed out

that the original Board of Trade wage comparison was too heavily

weighted towards craftsmen’s wages, which were higher relative to the

unskilled in Germany. They used Desai’s indices, which are more

broadly based, and this brought the German level down to 75% of

14 Calculated by dividing the rates of increase of these import categories, 1870/4 to1909/13, by the overall growth of German GDP, Wgures from HoVmann (1965).

15 The actual annual rate of increase of exports 1894/6 to 1911/13, 5.4%, did not quitematch this, which was one reason why Germany had a deWcit on visible trade in 1911–13.

16 Maddison (1994).17 Maddison (1995); this is an upwards revision of the Wgures for Germany in Maddi-

son (1991 and 1994). According to Prados de la Escosura (2000), 26–7, the gap was only12%.


the British at current exchange rates, and 63–67% at equivalent pur-

chasing power.18

It is also notable that there was little tendency for German wages to

converge on the British level, despite Germany’s faster rate of growth.

Williamson’s comparison shows real wages rising 47.2% in Germany

1870/4 to 1909/13, and 46.3% in Britain.19 Throughout this period

German exporters had a substantial competitive advantage over their

British rivals. Per unit labour costs were between 72% and 79% of the

British level in 1905.20

Yet German exporters may well have needed an advantage of this

order to make the required gains at Britain’s expense. Taking market

share away from an already established rival is not an easy business.

German exporters had to keep prices low. Between 1895 and 1913

prices of German manufacturing exports were reduced by around 20%

relative to the domestic price level. The need to gain market share by

oVering lower prices meant that German wages had to be held down.

As a result the impressive performance of the German economy was not

fully reXected in real wage increases.

There is a wider implication to this point. The argument that

developing economies are likely to encounter deteriorating terms of

trade is a well-established (though controversial) view in development

economics. It is a possible alternative to the Lewis Model. In the Lewis

Model, urban wages are held down by the availability of under-

employed rural labour. If the terms of trade are deteriorating, then

wages will be held down in the export-orientated sectors by the eVects

of falling export prices.21

There is a way out of this dilemma, although it is one which rela-

tively few developing economies have managed to take, and that is to

‘leapfrog’ up the technological ladder: to establish a strong position in

new industries rather than cutting prices to gain an increased share in

existing ones. This is what Germany achieved, and Figure 8.4 showed

18 Phelps-Brown and Browne (1968). Their index covers 18 manufacturing sectors,together with transport and mining, so it is more than just an index of manufacturingwages.

19 Williamson (1995), 164–7; this is an index of unskilled wages (mainly buildinglabourers), not those paid in export-orientated industries.

20 This combines Broadberry’s Wgure for relative manufacturing labour productivity in1905 with the wage Wgures produced by the Board of Trade survey, and by Browne andPhelps Brown: Broadberry (1997a), Phelps-Brown and Browne (1968), and Board ofTrade (1908).

21 The arguments are outlined in Todaro (1997), 478–80.


that much of German export growth was due to successes in new indus-

tries: chemicals, steel, electrical equipment, machine tools.22

But, although the weight of these newer industries was rising (chem-

icals, machinery, and metal goods accounted for 30.1% of exports in

1911–13, against 5.6% in 1870–2), Germany also pushed up exports in

categories where price advantages were probably more important:

exports of textiles, other Wnished goods, and semi-Wnished goods all

rose in 1895–1913. The increase in German exports was due to a com-

bination of successes in new technology and cost advantages in older

industries.

There are two ways of looking at the economy of Wilhelmine Ger-

many. One is a static analysis, looking at indices of real GDP per capita

or wages at purchasing power. These comparisons show that Germany

in 1913 was not a particularly rich country by the standards of the time,

although progress had been made since 1870. The other is to look at

potential: the structure of industrial output and, in particular, the pat-

tern of exports. Here, Germany had established a strong position in a

number of industries with excellent growth prospects in the twentieth

century. The potential for strong export-led growth, eventually realized

by the Federal Republic after 1945, was already in existence by 1913.23

Moreover, the move up the technological ladder made it possible for

Germany to escape from the terms of trade dilemma shown in Figure

8.5. Instead of competing on price in markets where British Wrms had

established positions, Germany was now able to oVer new products,

which Britain did not produce, and also compete on quality, by incorp-

orating new technology in goods such as electrically-driven machine

tools. This meant that export growth could be sustained without cutting

22 There is an argument, put forward by Paul Krugman, in Brezis et al. (1993), thatleapfrogging is an automatic process. This argument does not survive close scrutiny of therelatively small number of cases where leapfrogging has occurred. There were, forexample, strong reasons why the dyestuVs industry should have located in Britain notGermany (as it initially did in the 1860s). Britain had an ample supply of the mainfeedstock (coal tar) and a dominant position in the industry which was the main customer(textiles). Such strong positive externalities, running from an existing technology to a newone, generally mean that a more advanced economy has a considerable advantage in a newtechnology compared to a less advanced economy. Germany’s success in newer industriesowed much to Germany’s strong position in science, education, and training.

23 This is the converse of the British position. Static analysis of the British economy in1913 shows that it was still doing well, but the composition of exports was ill-suited torapid growth in the twentieth century: something that no amount of apologetics for thelate Victorian economy can obscure. Britain was ‘over-committed’ to industries with poorgrowth prospects, Richardson (1965).


prices, and holding down wages. The prospects for a sustained increase

in real wages were much better as a result.

3. M igration and Industrial Structure

The rapid growth of exports in certain sectors had eVects on migration.

These industries were not evenly distributed over the whole country but

relatively concentrated. Table 8.3 gives Wgures for the four main regional

groupings, and also for the three advanced industrial regions: Berlin-

Brandenburg, the Kingdom of Saxony, and Rhineland-Westphalia.

Table 8.3. Percentage distribution of the non-agricultural occupied

population, and the population occupied in three export-orientated

industries, 1895 and 1907

Metal-

working

Machinery Chemicals All three

industries

Total

non-

agricultural

occupied

population

1895

East 13.8 14.9 9.7 13.8 17.7

Centre 26.1 34.6 29.1 28.9 29.0

Berlin/Brandenburg 9.6 12.4 12.3 10.7 10.9

Saxony 8.7 12.9 6.6 9.8 9.9

West and north-west 39.3 28.2 36.8 35.8 32.6

Rhineland-

Westphalia

24.9 11.9 15.7 20.2 16.1

South and south-west 20.8 22.2 24.4 21.6 20.7

1907

East 11.9 10.9 8.3 11.2 16.0

Centre 26.2 37.6 28.9 31.2 29.8

Berlin/Brandenburg 11.1 15.4 12.3 13.0 12.2

Saxony 8.0 13.5 6.8 10.2 9.8

West and north-west 42.1 32.9 41.0 38.2 35.0

Rhineland-

Westphalia

28.2 16.6 17.6 22.5 18.4

South and south-west 19.8 18.6 21.8 19.4 19.1

Source: Figures are calculations from data in Kaelble and Hohls (1989).


These three advanced regions accounted for 40.7% of the total occupied

population in the three export-orientated industries in 1895, and this

had risen to 45 percent by 1907. Metal-working and machinery were

more concentrated than chemicals (possibly for geographical reasons).

The east lost ground between 1895 and 1907. Its share of these three

industries fell, as did its share of the total non-agricultural occupied

population. The failure of eastern industry to keep pace with the rest of

the country was a major factor in the high rate of migration out of this

region. The south did rather better. It had a high share of employment

in chemicals so its share of the export-orientated industries was rela-

tively high, though this fell between 1895 and 1907.24

Figure 8.6 gives the change in the regional pattern of non-

agricultural employment, 1895–1907. There is an evident shift towards

25−30

30−5

20−5

10−15

40−5

35−40

60−70

70−80

<10

>80

15−20

50−60

45−50

Fig. 8.6. Percentage increase in non-agricultural employment by region,

1895–1907

Source: Calculated from data in Kaelble and Hohls (1989).

24 Hohorst, in Pollard (1980), 231, gives estimates of regional income levels whichconWrm that the east was falling behind.


the centre and the west. In part this reXects the growth of cities: these

areas had more large cities and these were the areas where employment

in industry and services was increasing most rapidly. But it is also

noticeable that there were a number of regions in the west (Stade,

Luneberg, and Oldenburg, for example) which did not contain large

cities, where non-agricultural employment rose faster than in eastern

regions, such as Konigsberg, Breslau, or Stettin, which did.

Again, the south and south-west has a diVerent pattern, containing

both regions where there was little increase in non-agricultural employ-

ment (Upper Alsace, Lower Bavaria, and Jagstkreis in Wurttemberg)

and others where the increase was over 40%—Mittelfranken, Mann-

heim, and Lorraine (there were none in the east). This was one reason

why most migration in the south was short or medium-distance; the

east was the main source of long-distance migration because there were

less opportunities to Wnd non-agricultural work within the region as a

whole.

There were two trends which did not help the east. The Wrst was the

tendency for regions to become more specialized and for industries to

become more concentrated. External economies, internal and external

to the Wrm, meant that industrial growth was highest in regions which

were already relatively industrial.25 The second was vertical integration:

the tendency for processing industries to locate in areas close to sources

of primary production. Again, this could be both internal and external

to the Wrm.

Both these eVects can be demonstrated using the occupational cen-

suses. Table 8.4 provides a summary of the results of a comparison of

the censuses for 1895 and 1907 (the full results are given in Table 8B.1

in Appendix 8B). The analysis shows that there was a tendency towards

concentration both in the industrial and in the service sectors. Regions

which had above-average levels of non-agricultural employment had

higher rates of increase for employment in these sectors compared to

regions with below-average rates of employment in industry and ser-

vices. The estimates suggest that a region with 70% of the occupied

population employed outside agriculture would experience a 46% rise

in non-agricultural employment; one with only 40% would have a rise

of 34%.26

25 Steller (1903) gives Wgures for the formation of machinery Wrms which show that theeast was losing ground in this area.

26 Using the coeVicients given in the Wrst column of Appendix Table 8B.1 (OLSestimates).


Analysis of results for individual industries identiWed three which

were becoming more concentrated: metal-working, machinery, and

clothing. This suggests that an ‘agglomeration eVect’ was in operation

in these industries. The miscellaneous, ‘other industries’, sector was

also moving in this direction. Only one sector showed a tendency to

move towards dispersion—utilities (though the result is not a secure

one). In this sector a tendency towards dispersion might be expected, as

a result of the spread of electriWcation and gas production, and wide-

spread investments in improved water supplies.

Taking the machinery sector as an example, the results imply that a

region with 0.5% of the occupied population in this industry in 1895

would experience a 78% increase in employment; one with 2.0% would

have an increase of 120%. There would be a pull of migrants towards

the towns and cities associated with these ‘concentrating’ industries.

A similar approach can be used to examine the evidence for increased

vertical integration. There was strong association between linked indus-

tries, such as mining and metal production: employment in one was

Table 8.4. Summary of results of statistical analysis of trends in

employment shares for 81 regions, 1895–1907: trends towards

concentration and dispersion


Sectors or industries showing a trend towards concentration:

The industrial sector as a whole

The service sector as a whole

Metal working

Machinery

Clothing

Miscellaneous industries


Sectors or industries showing a trend towards dispersion:

Utilities

III. Other industries showing no clear trend to concentration or dispersion

Mining and quarrying

Metal production

Chemicals

Textiles

Food and drink

Construction


high in regions where the other was also present. The question con-

sidered was whether these links were getting stronger. If so, then em-

ployment in the ‘downstream’ industry would be rising faster in areas

where the ‘upstream’ industry was already established. The strongest

evidence for this was found in the machinery sector: this industry was

locating in areas where metal-working was already established, and also

in areas where high employment in utilities indicated good supplies of

power and other services. There was some evidence that metal-working

Table 8.5. Summary of results of statistical analysis of trends in

employment shares for 81 regions, 1895–1907: trends towards vertical

integration (linked pairs)

I. Groups of industries showing a trend towards increased vertical integration

Secure results:

Machinery—metal-working

Machinery—metal-working—utilities

Less secure results:

Metal-working—metal production

II. Groups of industries showing no trend towards increased vertical integration

Mining—metal production

Clothing—textiles

Table 8.6. Summary of results of statistical analysis of movement of

insured workers 1909–1913


Characteristics of the recipient regions which increased inwards migration:

High employment in industry and services relative to agriculture

High employment in services relative to industry


Characteristics of the recipient regions which increased inwards migration:

High employment in chemicals

High employment in metal-working

High employment in construction

Characteristics of the recipient regions which decreased inwards migration:

High employment in utilities


was expanding more in areas where employment in metal production

was already high.

The implications of industrial structure for migration are considered

in table 8.6, which summarises the statistical analysis given in Appensix

8B section ii. This is based on the recorded movements of insured

workers in 1909–1913. The analysis shows that the likelihood of migra-

tion into a given region was influenced by the availability of non-agri-

cultural employment, particularly in the service sector, which seems to

have attracted migrants more than industry.

Industries which were found to have a particular appeal to migrants

in the period included 2 export-oriented industries – chemicals and

metal-working – which were expanding rapidly in the period just before

the First World War, and construction, which had always employed

large numbers of transient workers. On the other hand, it seems that

jobs in utilities were not easily obtained by migrants, as this sector is

associated with lower migration.

4. M igrat ion and Economic Cycles

The argument put forward earlier was that there was a change in the

nature of migration, from a destabilizing, labour-surplus pattern up to

1895 to a stabilizing pattern after this date, more characteristic of

modern mature economies. Migration then acted as a buVer, absorbing

part of the eVects of short-term economic Xuctuations on employment

in urban labour markets. But there was still substantial rural–urban

migration, which helped to reduce the strain on urban labour markets

in recessions. This may help to shed some light on a diVerent question

which has been much debated: why was recorded unemployment rela-

tively low before 1914?

Town migration Wgures provide a source for annual movements. The

discussion in Chapter 4 showed that these are not without their prob-

lems: there were gaps in the coverage, diVerences in deWnitions and in

legal requirements to register a change of residence. From 1898 onwards,

comparability was improved, and annual Wgures were published in the

Statistisches Jahrbuch Deutscher Stadte. For the years 1904–12 it is pos-

sible to produce a series covering 25 cities with a total population of 7.5

million. All the major cities are included with the exception of Munich.

These Wgures are given in Table 8.7, together with labour exchange

Wgures for 23 of these cities, and two series for economic conditions in


the whole economy. The Xuctuations of the economy can be followed in

the second column, giving the annual change in industrial production.

There was a period of expansion in 1904–7, followed by a brief reces-

sion in 1908, and a recovery in 1909–12. The Wnal two columns show

how labour market conditions responded to these Xuctuations. In the

recession there was both a rise in applicants per vacancy, and in un-

employment.

The way that migration was aVected by changes in economic condi-

tions is quite clear. There was a sharp fall in net gains in 1908–9 from

the much higher rates in 1904–7. It is noticeable that the response of

migration was at least as great as the change in recorded unemploy-

ment. Statistical analysis (reported in Table 8B.4 in Appendix 8B,

Section (iii)) shows that a change of one percentage point in unemploy-

ment produced an equivalent change in migration. The eVect of migra-

tion was to halve the change in unemployment (assuming that all those

deterred from migrating would otherwise have become unemployed).

However, this may underestimate the importance of migration. The

unemployment Wgures are derived from trade union sources, and so

Table 8.7. Summary of gains and losses through migration 1904–1912:

25 cities with full sets of migration Wgures, together with the change

in total industrial production, a comparison of labour exchange

applications to the number of open places, and recorded

unemployment

Net gains

as % of

population

% change in

industrial

production

Ratio of

applicants

to vacancies

Recorded

unemployment

1904 2.50 þ4.2 1.26 2.1

1905 2.73 þ3.7 1.03 1.6

1906 2.96 þ4.3 0.95 1.2

1907 2.14 þ7.8 1.26 1.6

1908 0.26 �0.9 1.79 3.0

1909 0.78 þ4.4 1.65 2.9

1910 1.24 þ5.0 1.30 1.9

1911 1.28 þ6.1 1.15 1.9

1912 1.11 þ7.2 1.21 2.0

Source: Calculated using data from SJDS (1906), 280 and 482–3, and subsequent issues;unemployment from Bry (1960), 325–6; industrial production from HoVmann (1965),451–2.


record numbers unemployed as a percentage of union members. This

may be a reasonable approximation for unemployment amongst the

working population. But the migration variable used in the regressions

is net moves relative to the total population. If participation rates

amongst migrants are above average, particularly in the ‘volatile com-

ponent’ of migration, those who move in response to changes in eco-

nomic circumstances, then the ‘unemployment-relieving’ eVect will begreater. As we know that migrants are predominantly young and of

working age, it is indeed probable that participation rates will be above

average.

A few calculations illustrate the importance of this point. If the par-

ticipation rate amongst ‘volatile’ migrants had been 60%, against a

German average in 1907 of 46%, then the ratio between the change in

unemployment and the change in migration (of those available for em-

ployment) would be raised from 1 : 1 to 1.37 : 1. If the participation rate

had been 80%, the ratio is increased to 1.83 : 1.

Figure 8.7 shows the eVect of adding the change in migration to the

change in unemployment. Adjustment 1 assumes the participation rate

of migrants to be 60%; adjustment 2 makes it 80%. The combined

eVect is to greatly increase the volatility of the whole series.

It seems clear, from these Wgures, that migration in and out of the

urban sector made an important contribution to the reduction of

recorded unemployment in the period before 1914. There are two add-

itional points which follow from this. The Wrst is that this implies the

existence of a pool of underemployed labour in agriculture. At times

when the economy was expanding rapidly, this might be reduced to

quite a low level, but in recessions it would rise to substantial propor-

tions if the reduction in migration recorded for these 25 cities applied

to all moves into the non-agricultural sector.

The second is that a decision by an unemployed migrant to return to

his or her district of origin is not without its costs. There is the cost of

the journey. In addition, there may be advantages to remaining in a city

even if unemployed. It would be easier to look for job opportunities. It

is likely that, when conditions improve and new vacancies appear, the

Wrst to obtain these jobs will be the unemployed still resident in the city

or industrial district, not those who have returned to a parent’s farm or

smallholding.

If this is correct, then the implication is that, if unemployment ben-

eWts were to be made available to workers who had previously been

unable to claim (as in fact happened in the 1920s and 1930s), then, even


if these were kept quite low, a substantial number of the unemployed

would prefer to remain in the cities rather than migrate back to rural

areas. The payment of beneWt would change the position from one in

which there was a substantial, though Xuctuating, pool of hidden rural

underemployment, to one in which there was overt unemployment in

cities.

5. Some Consequences of M igration

(i) Migration and wage convergence

The consequences of migration are, on the whole, likely to be favour-

able for the migrants themselves, even if they may be not so good for

those already resident in the recipient cities who face increased compe-

1905−4

−2

0

2

4

6

1906 1907 1908 1909 1910 1911 1912

Change in unemployment

Plus migration

Plus migrationadjusted (1)

Plus migrationadjusted (2)

Fig. 8.7. Percentage changes in unemployment plus migration, 1904–12

Note: Annual changes in the unemployment rate in % points, relative to the workingpopulation, showing the eVect of allowing for changes in the rate of migration underdiVerent assumptions about the participation rate of migrants.


tition for jobs and housing. The ‘congestion eVect’ on urban amenities

and infrastructure may also be a cost not included in the calculations

made by individual migrants. But the transfer of labour from relatively

unproductive activities, or from regions with high underemployment,

should beneWt the economy as a whole.

The available wage and migration data are not suViciently good to

make it possible to measure the eVect of migration on urban wages by

econometric analysis. It is possible to look at some general trends in

wages, in particular the question of wage convergence. Table 8.8 pro-

vides an analysis based on the surveys of ‘typical’ day-labourer rates

(Ortsubliche Tagelohne) made for insurance purposes. The Wrst part of

the table gives regional averages (weighted by the city population).

These show that wages in eastern cities were well below those in other

Table 8.8 (a) Day-labourer rates (Ortsubliche Tagelohne) in 85 cities,

various years; weighted averages (by population); German average ¼ 100

1884 1892 1902 1912

Men

East 77.9 86.6 86.5 85.5

Centre 102.2 107.8 106.7 109.7

West and north-west 110.8 110.3 106.4 104.8

South and South-west 101.0 100.3 106.2 109.7

Women

East 73.6 73.8 79.1 78.8

Centre 100.8 101.7 99.9 108.0

West and north-west 113.4 113.5 109.9 106.7

South and south-west 105.2 102.3 109.3 110.6

Source: Calculated using data from SJDS (1913), 826.

Table 8.8 (b) CoeVicients of variation for day-labourer rates in

85 cities

1884 1892 1902 1912

Men 0.172 0.149 0.133 0.138

Women 0.198 0.169 0.170 0.166

Note: The coeVicient of variation is the standard deviation divided by the mean: it istherefore a standardized measure of dispersion.


parts of the country, although the gap did fall. The period of ‘catch-up’

was 1884–92 for men and 1892–1902 for women.

In 1884 the highest wages were in cities in the west and north-west,

but this was gradually altered, as this region tended to lose ground to

the centre and the south. Tying this in to migration is not so easy. The

main migrant Xow was from the east to the centre and the west, so it

probably contributed to the convergence of eastern wage rates, but there

was no obvious connection with the diVerent trends of wage rates in the

centre and in the west.

Wage dispersion (as measured by the coeVicient of variation—part

(b) of the table) tended to fall, both for men and women, although most

of the fall occurred in 1884–92. This indicates that the German labour

market was becoming more integrated: regional diVerences were eroded

by the migration of labour from low wage areas. This occurred despite

the relatively uneven pattern of economic development shown in Figure

8.6, which should have caused wage dispersion to rise. The migration

process was suViciently powerful to overcome this eVect.

The wage Wgures produced by these surveys have the advantages of

being contemporary estimates, and produced using a common method-

ology, but they are not based on actual data: the local insurance oVices

were asked to provide Wgures for what they considered to be the ‘typ-

ical’ day-labourer rate for the locality. It would be desirable to supple-

ment this evidence of convergence with data from some other source,

recording actual payments to some category of workers. There are vari-

ous possible series, but relatively few that cover a substantial number of

regions or cities over a decade or more. One possible series gives the

earnings of miners (hard coal, lignite, and iron ore) in nine categories

covering eight diVerent regions, 1889–1913 (Figure 8.8). This series

shows a consistent fall in the coeVicient of variation for above-ground

workers, but the position with regard to below-ground workers was less

clear. Of the two, it is probably the above-ground series which gives the

more representative picture of general labour market conditions. Below-

ground workers were often on piece rates which meant that geological

problems would aVect earnings. The diViculties experienced by some

Silesian coal mines (an area where earnings were low), for example, may

have held back convergence.

Another series gives the weekly earnings of unskilled building

workers in eight cities (Figure 8.9). This also shows a strong process of

convergence. However, convergence was much weaker for the more

skilled categories, carpenters and masons (although data is only available

for four cities).


18890.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

1892 1895 1898 1901 1904 1907 1910 1913

Below ground

Above ground

Fig. 8.8. CoeVicients of variation for the daily earnings of miners,

1889–1913, nine categories

Source: Calculated using data from Bry (1960), 342–5.

189910

20

30

40Marks

1901 1903 1905 1907 1909 1911 1913

BerlinDresdenElberfeldFrankfurt am MainKielLubeckNurembergRostock

Fig. 8.9. Weekly earnings of unskilled building workers, eight cities, 1899–1913

Source: Calculated using data from Kuczynski (1945).


The evidence of convergence is particularly striking when contrasted

with the results given in Section 3 of this chapter, which showed a

process of divergence with regard to non-agricultural employment: the

east was not keeping up with the more advanced parts of the country. If

wages were converging while industrial structure was not, then the

most likely explanation is that this was caused by migration, if no other

force promoting convergence can be identiWed.However, at the end of this period there were still major disparities

in wage rates between diVerent parts of Germany. Figure 8.10 gives the

25−30

30−35

20−25

10−15

40−45

35−40

>10

15−20

60−70

70−80

>80

50−60

45−50

Fig. 8.10. Day-labourer rates in urban areas in German regions, 1912

(weighted averages of male and female rates)

Note: Regional averages were calculated by weighting the rates for each city by its share ofthe total urban population in each region (where the number of cities in the region isgreater than one). Regional units which did not have a major city were amalgamated withnearby regions (Koslin with Stettin for example).Source: Data for 94 cities from SJDS (1913), 826.


Ortsubliche Tagelohne (day-labourer rates) for urban areas throughout

Germany in 1912. The Wgures are weighted averages of male and female

rates, using the share of each sex in the total occupied population for

the whole of Germany (the male share was 66%). Figure 8.10 conWrms

the point that urban wages in the east were below levels in the rest

of the country, and it shows that this applied just as much to the indus-

trial areas, such as Silesia, as to the other, more agricultural, eastern

regions. It also shows that wages in many southern cities were high.

Day-labourer rates in Freiburg, Mannheim, Stuttgart, and Munich

were as high as anywhere in Germany.

As in the earlier analysis of rural wage rates, the inXuence of the

Russian–Polish border is strong. There is a clear east–west gradient,

although high rates were also found in the centre, in the Berlin area,

and in Saxony. Relating this to the discussion in earlier chapters, it

provides a reminder of the point that German labour market conditions

were inXuenced by Germany’s position in central Europe, and the prox-

imity of Russian Poland in particular.

(ii) Migration and skill endowments

There is one possible consequence of migration which is less favourable

for the region of origin: if migration depletes the skills available in a

region signiWcantly faster than the overall rate of out-migration then the

region will become an unattractive location for advanced industries

with high requirements for skilled labour. A vicious circle could de-

velop: a lagging region loses skilled workers and then lags even more as

a result.

The German occupational census of 1907 is well suited to the analy-

sis of this question, as it recorded the place of birth of respondents, as

well as classifying them by the skill requirements of their occupations.

Consequently, it is possible to compare the numbers born in a given

region, of a particular skill level, with the numbers resident at the time

of the census. Regions which experience out-migration will have a ratio

of numbers born to numbers resident of less than 1. The question is: is

this ratio lower for the higher skill categories?

Table 8.9 provides an analysis of this type for the main regional group-

ings. The results show that the east had a net loss of around a quarter of

the men born in the unskilled category, but a lower Wgure of just 10% of

women. However, female participation rates fell with age, so this may

just reXect the fact that many unskilled female migrants withdrew from

the labour market after migration. Net losses for the east in the skilled


category are about the same for men and women, which means that they

are no higher than the unskilled for men, but rather higher for women.

In general the table shows that the east did not lose disproportionate

numbers of skilled or entrepreneurial male migrants, with the exception

of the ‘managerial and technical’ category, where there were high losses

both of men and women. So the question revolves around two issues:

how important were the high losses in this category? and can the high

losses of skilled and entrepreneurial women relative to unskilled be

disregarded?

Looking at the absolute numbers, it is clear that the most important

ratio in the more skilled categories is the ratio for skilled male workers:

there were more of these than any other category. The fact that the east

did not lose disproportionate numbers of these leads to the conclusion

that the eastern economy did not experience a major loss of economic

potential through migration. On the other hand the Wgures do suggest

that for white-collar workers of both sexes, and for entrepreneurs and

the self-employed in the service sector, the east was a relatively un-

attractive place to work. This would have contributed to migration. It

Table 8.9. Ratios of numbers born to numbers resident, by skill

category and region, 1907 occupational census


north-west

South and

south-west

Entrepreneurs and

self-employed: industry

Men

Women

0.803

0.845

1.077

1.148

1.062

1.044

0.995

0.993

Entrepreneurs and

self-employed: services

Men

Women

0.713

0.706

1.134

1.200

1.075

1.059

0.996

0.980

Managerial and

technical: industry

Men

Women

0.587

0.622

1.129

1.153

1.070

1.021

0.950

0.960

Skilled workers:

all sectors

Men

Women

0.740

0.744

1.069

1.185

1.090

1.024

1.015

0.964

Unskilled workers:

all sectors

Men

Women

0.746

0.903

1.273

1.202

1.152

1.035

0.973

0.996

Source: Calculated from 1907 occupational census, SDR n.f. 210.


would also have had some detrimental consequences for the growth of

non-agricultural employment.

Turning to the other regions: the centre and west made net gains in

all categories. It is noticeable that, in both regions, the highest net gains

were in the ‘male, unskilled category’; the educational and training

system within these regions was supplying most of the skills needed

from those born within the region. The south had net losses in all

categories except skilled male workers. So the migration balance was

slightly favourable to the ratio of skilled to unskilled in the south. The

eVect was not so large: migration raised the ratio of skilled to unskilled

male workers from 0.0835 to 0.0871.

In general, therefore, the migration pattern of the more skilled was

not massively diVerent from that of the less skilled, and skill transfers

were not a major contribution to the economic success or failure of the

diVerent regions.

(iii) Migration and the position of women

The analysis in Sections (i) and (ii) has shown up some diVerences

between the position of men and women. According to the Wgures in

Table 8.9, the gap between male and female wages was larger in the

east: female pay was 78.8% of the national average in eastern cities in

1912, while the Wgure for men was 85.5%. The analysis of skill trans-

fers has indicated that opportunities for white-collar employment, an

important area for women, were better outside the east. So, it would

appear that migration out of the east should have led to an improve-

ment in women’s prospects.

There are two important problems with this view. The Wrst is the

eVect of migration on women’s place in the household economy. Analy-

sis of the budgets of rural households in 1910–14 shows that the house-

hold economy contributed as much as 44% of total household income

in some rural areas.27 This cannot be exactly equated with the female

contribution, but it does give an idea of the potential size of the contri-

bution which could be made by a married woman who did not have

formal employment. There is good evidence that migrant households

preferred, if possible, to replicate this position in industrial areas, by

obtaining smallholdings. This was one factor which contributed to the

27 Mittler (1929), 210–15.


large rise in small agricultural holdings in industrial regions.28 There

was an increase of 72,975 in the numbers of holdings of less than 2

hectares in Westphalia in 1882–1907.29 However, this was only possible

in industrial areas where agricultural land could be found in close prox-

imity. This was often the case in mining areas but it was rare in the

larger cities.

The second point is that the gap between male and female earnings

was higher in urban areas than in the countryside, and it was higher still

in the major cities. Table 8.10 gives Wgures calculated from the 1892 and

1902 surveys of Ortsubliche Tagelohne. The second part also shows that

the gap between rates for men and women in cities was tending to rise up

to 1902, although there was some reduction in 1902–12.

In summary, migration had, on the whole, negative consequences for

the economic position of women. Although some may have beneWted

Table 8.10. The relative position of women: ratios of male to

female wages (Ortsubliche Tagelohne)

(a) Average ratios for diVerent localities in Prussia (1892 and 1902 surveys

combined)

Cities Towns Rural

Male to female ratio 1.646 1.532 1.519

Source: Data from Neuhaus (1904).Note: Neuhaus (1904) gives, for all the Prussian Kreise, Wgures for rural and urban rates.These have been divided between ‘cities’ (urban areas with a population of at least 50,000in 1912) and ‘towns’ (other urban areas). The averages given in the table are unweightedaverages following this classiWcation.

(b) Average ratios for 86 German cities

1884 1892 1902 1912

Unweighted 1.591 1.618 1.679 1.657

Weighted by population 1.585 1.639 1.696 1.665

Source: Calculated using SJDS (1913), 826.

28 The smallholding could support the family in times of unemployment or sickness.‘A small piece of land is eVectively a savings bank for the worker’, Wygodzinski in Sering(1899–1910), i. 140.

29 Dillwitz (1973), 98; see Tenfelde (1977), 114–18 for the importance of smallholdingsto the Ruhr miners.


from the improved opportunities to achieve white-collar employment,

or to become self-employed in the service sector, these were a minority.

Wages for women in cities were lower relative to men than in the

countryside, and there was a loss of economic status as a result of

reduced opportunities within the household economy.30

However, life expectancy Wgures show a rather diVerent picture

(Table 8.11). Here, it seems that men were the main losers. At a typical

migrant age of around 20, men experienced a loss of nearly 4 years in

life expectancy by moving to a city. This was 3.5 years more than the

loss experienced by women. Possible reasons for this include the more

dangerous nature of the work done by men in urban areas, and the fact

that more women were employed as domestic servants, which may have

involved better housing or residence in areas with better sanitation. On

this point, therefore, it does not appear that women were the main

losers from migration.

6. Conclusions

The main theme of this study is the movement of labour out of

agriculture into urban labour markets, and the conditions and the

30 Kocka (1990), 465 gives Wgures for the gap between male and female earnings: Ritterand Tenfelde (1992), 215 also show that there was a gap in status at the workplace as well.On the other hand, mortality Wgures, as given in Appendix 5B, show that male lifeexpectancy fell by more when rural and urban rates are compared.

Table 8.11. Comparison of life expectancy in major Prussian cities to

expectancy in the rest of the province, calculated from mortality tables

for 1893–4

At birth At 1 year At 20 years

Men: Major cities 34.7 47.7 56.9

Rest of province 40.0 51.8 60.8

DiVerence �5.3 �4.1 �3.9

Women: Major cities 39.9 52.2 62.4

Rest of province 42.9 53.5 62.8

DiVerence �3.0 �1.3 �0.4

Source: Calculated from tables in Kucyzinski (1897); Wgures for individual cities givenin Appendix 5B.


circumstances under which a ‘labour surplus’ is Wrst created and then

released. The preceding chapters have, therefore, concentrated on the

rural sector—on technological, demographic, and institutional develop-

ments within this part of German society. But the urban context is also

important. If the urban sector is under pressure—from capital shortage

or from adverse movements in the international terms of trade—then

the ability of the urban economy to absorb the rural surplus, without

putting downwards pressure on wage rates, is impaired.

From this point of view, there were two main phases in the develop-

ment of the industrial sector in this period. The Wrst lasted from the

1870s to the 1890s, and was characterized by favourable movements in

the terms of trade for the export sector. However, it was also a period

when the German capital markets were suVering from the after-eVects

of the Grunderzeit boom of the 1870s. This restricted access to capital at

a crucial time in German economic development.

The second phase began in the 1890s and lasted until the First

World War. In this period the capital markets still experienced periodic

crises, but these were relatively quickly overcome. Investment levels

were high, and only temporarily aVected by cyclical downswings. On

the other hand, the export sectors were adversely aVected by the move-

ment of the terms of trade. The recovery of world food and raw mater-

ial prices from the trough of the 1890s had a part in this, but German

industry was also having to cut prices to gain a share on world markets:

the price of German manufactures was falling relative to those of other

countries.

In the long run, Germany overcame this problem by moving up the

technological ladder, by developing new products rather than compet-

ing in established markets. This process was already under way before

1914. The Wgures for ‘high technology’ exports make this clear. This

created the prospect for rapid export growth without adverse move-

ments in the terms of trade, as would be the case for the Federal

German Republic after 1945. If world trade had not been disrupted by

the First World War and its aftermath, then the Kaiserreich might well

have moved into this third phase, which would have made it much

easier to achieve real wage improvements.



and Sources

(i) Appendix Table 8B.1

Dependent variables For each industry: % change in numbers oc-

cupied, calculated from occupational census

data given in Kaelble and Hohls (1989).

Independent variables As above, % of the total occupied popula-

tion occupied in each industry.

(ii) Appendix Table 8B.2

As Table 8B.1, industry variables as indicated.

(iii) Appendix Table 8B.3

Dependent variable Logistic transformation, Yjk ¼ log (r=1� rjk), where rjk is the average annual

numbers of moves from region j to region k,

1909–13 divided by the occupied population

in region k in 1907. Movement data from

Reichsarbeitsblatt (annual issues) 1910–14;

occupied population from Kaelble and

Hohls (1989).

LOGDISTANCEkj The distance in kilometres between each

region measured from maps, being the

smallest distance between the frontiers. The

distance between contiguous regions was

entered as 10 kilometres, this being the value

which was found to reduce the coeVicient on

a contiguity dummy to zero.

POPREGORIGINj The total population of the region of origin

1907, data from Kaelble and Hohls (1989).

INDUSTRY%k As above, the percentage of the total occu-

pied population occupied in industry, 1907,

in the recipient region.

SERVICES%k As above, for services.

CHEMICALS%k As above, for chemicals.

MACHINERY%k As above, for Maschinenbau.


METALWORK%k As above, for Metallverarbeitung.

CONSTRUCTION%k As above, for construction.

UTILITIES%k As above, for Versorgungsleistung (gas, water,

and electricity).

REGORIGIN%AGj As above, for agriculture in the region of

origin.

(iv) Appendix Table 8B.4

Dependent

variable (MIGt)

Data from SJDS (1907), 69, and subsequent

issues, combining annual results for 25

cities: total net moves in was divided by total

population. The selected cities were those

with complete sets of data for the period.

UNEMPLOYMENTt The annual average rate of unemployment

amongst trade union members, data from

Bry (1960), 325–6.

APPL=VACANCIESt Data from SJDS 1904–13, combining results

for 23 cities: total numbers of registered ap-

plicants at labour exchanges divided by the

number of vacancies, annual Wgures.

INDPRODUCTIONt Industrial production, data from Hofmann

(1965), 451–2.

Appendix 8B. Econometric Results

(i) Analysis of employment change by industry, 1895–1907

The changing structure of German industry had a direct eVect on

patterns of migration. As industry became more concentrated and more

export-orientated, migrants were drawn towards these areas. The ten-

dency to concentration can be demonstrated using the occupational

censuses. Table 8B.1 gives the results of a series of equations estimated

using the form Yki ¼ aþ bXki, where Yki is the percentage increase of

employment in industry k in region i in 1895–1907 and Xki is the

percentage of the total occupied population in region i occupied in

industry k in 1895. Any tendency towards concentration should pro-

duce a positive estimate for b.


There was a problem with heteroscedastic errors in some of the

estimated equations. So alternative estimates were produced using a

two-stage weighted least squares procedure. There are few major diVer-

ences between the two sets of results, but the 2S WLS results are

somewhat clearer, due to the elimination of some nearly signiWcant, but

in fact rather spurious, results.31

31 The 2S WLS procedure follows the procedure set out in Greene (2000), 514–15.A weighting variable was calculated, which was wi ¼ 1=(Xi)

z for each industry, the value zbeing obtained by an interactive maximum likelihood procedure. This was then used inWLS estimation of the equation. The reason why this problem arose was that there weresome industries where employment was low and highly variable. In metal production, forexample, employment was often in single Wgures: 1 person was employed in Bromberg in1895, but 9 in 1907, an increase of 800%. Using OLS produced absolute errors stronglyrelated to the explanatory variable.

Table 8B.1 Regressions of change in employment 1895–1907 on the

percentage of the total occupied population employed in each industry

in 1895, estimated using ordinary least squares and a two-stage

weighted least squares procedure (81 regions)

Equations estimated using OLS Equations

estimated

using 2S WLS

constant b t-ratio R2 adj. R2 b t-ratio

All industry 18.04 0.406 (2.74) 0.087 0.075 0.574 (3.91)

All services 2.40 0.797 (4.26) 0.187 0.177 0.895 (4.00)

Mining and

quarrying

93.33 �2.74 (0.42) 0.002 0.000 �0.04 (0.00)

Metal

production

274.57 �70.4 (1.65) 0.035 0.022 �4.57 (0.92)

Metal-working 16.02 4.12 (3.50) 0.134 0.124 4.12 (3.50)

Machinery 64.16 28.1 (3.18) 0.113 0.102 40.2 (4.18)

Chemicals 82.69 �28.2 (1.14) 0.016 0.004 �1.24 (0.10)

Textiles 2.15 0.71 (1.17) 0.017 0.000 0.18 (0.54)

Clothing �9.65 4.11 (3.68) 0.146 0.135 5.44 (3.33)

Food and drink 20.46 1.29 (1.20) 0.018 0.005 1.16 (1.01)

Construction 52.00 �1.79 (1.22) 0.018 0.006 �0.47 (0.35)

Utilities 170.32 �164.3 (1.41) 0.025 0.012 �87.3 (1.73)

Other industry 10.26 1.47 (4.00) 0.169 0.158 1.34 (4.04)


There is good evidence of a move towards concentration both in

industry and services. Analysis of the service sectors (not reported)

showed that this result was heavily aVected by the domestic service

sector.32 Within the industrial sector there are four categories which

show clear trends towards concentration: metal-working, machinery,

clothing, and ‘other industries’. The sector with the best evidence of a

move towards dispersion is utilities, although this is not a result sign-

iWcant at the 5% level.

The occupational censuses can also be used to look for evidence of

vertical integration. This analysis is reported in Table 8B.2. Correlation

analysis revealed a number of industries which were linked geographic-

ally. The highest value in 1895 was 0.83 for the correlation of employ-

ment in mining and metal production. But these industries were already

so well integrated that there was little scope for further movement in

this direction. Column a of Table 8B.2 shows that there was no sign-

iWcant eVect on the change in employment in metal production from

the percentage employed in mining in 1895.

Column e shows that there was little eVect from textile employment

on the change in the numbers occupied in the clothing industry: the

coeVicient on textile employment is positive but not signiWcant. Simi-

larly, the results in column b indicate that there was some eVect from

employment in metal production on the increase in employment in

metal-working, but the coeVicient was not signiWcant.The machinery industry was the one which showed the strongest

inXuence from other industries. As column c shows, the increase in

employment in this industry was inXuenced by the presence of metal-

working in 1895. In column d other variables are added. These results

indicate that the strongest eVect on the rate of growth of machinery

employment came not so much from the percentage employed in ma-

chinery in 1895, but rather from the presence of metal-working and

from employment in utilities. This last eVect may well reXect the in-

creased importance of electrical power, as the electrical industry was

included in this category. The machinery industry was moving to the

areas most advanced in the use of electricity.

32 Which suggests a ‘middle-class agglomeration’ eVect. However, this result is notunimportant: domestic service was an important source of employment opportunities forfemale migrants, and the availability of such opportunities was therefore a factor in migra-tion decisions by women.


(ii) Industrial development and migration in 1909–13

The eVects of industrial developments on migration can be studied

using data provided by the various Landesversicherungsanstalten—the

oVices which administered accident and sickness insurance. When an

insured worker left one region for another, a Quittungskarte was ex-

changed between the regions. Total numbers of Quittungskartenaustauch

were published annually in the Reichsarbeitsblatt from 1910 onwards.

Table 8B.2. Regressions of change in employment 1895–1907 on the

percentage of the total occupied population employed in related

industries in 1895, estimated using OLS (81 regions)

Metal

production

Metal-working Machinery Clothing

a b c d e

Constant 281.12 17.01 30.84 31.12 �10.08

MINING95 �17.55

(0.63)

METALPROD95 �30.52 þ2.63 þ9.38

(0.40) (1.10) (1.45)

METALWORK95 þ3.47 þ14.96 þ9.96

(2.65) (4.36) (2.71)

MACHINERY95 þ18.27 þ8.76

(2.20) (1.01)

UTILITIES95 þ295.72

(3.96)

CLOTHING95 þ4.02

(3.57)

TEXTILES95 þ0.20

(0.78)

N 78 81 81 81 81

S 398.80 20.09 56.48 51.45 14.09

R2 0.040 0.148 0.287 0.424 0.153

adj R2 0.014 0.126 0.269 0.393 0.131

F-test of regression 1.56 6.76 15.73 13.97 7.04

RSS 11926363 31491.7 248817 201211 15480.2

Note: t-ratios in brackets. As before there were heteroscedasticity problems with columna; however, no corrective mechanism was found which produced signiWcant results forthis equation, so the results are not reported.


This provides a more complete picture of annual movements than any

other published source. Even though data are only available for Wve

years, 1909–13, it still gives a valuable insight into the operation of the

German labour market in the years just before the First World War.

One reservation is that the data covers only workers who were already

insured. Younger migrants, who had not been insured, would not be

included.

In Table 8B.3 the linear log-odds model, used in Chapter 5, is ap-

plied to the Quittungskartenaustauch data. Here, the focus is on the

characteristics of the recipient region (k), and so the dependent variable

is a logistic transformation of the number of recorded moves into region

k from region j, divided by the total occupied population in that region

(from the 1907 occupational census).

The Wrst column gives the basic model. The probability of a move

into a given region is inXuenced by distance, by the size of the popula-

tion of the region of origin, and by the amount of non-agricultural

employment in the recipient region. The much higher coeVicient on

service employment conWrms that this sector was an important source

of opportunities for migrants. The percentage occupied in agriculture

in the region of origin was also entered, but this variable was not found

to be signiWcant and so was dropped.33

Columns b, c, and d show that, entered separately, the export-orien-

tated industries have a positive eVect on the probability of inward mi-

gration. Due to collinearity, when entered together only chemicals and

metal-working have signiWcant coeVicients. The Wnal column shows the

results of a step-by-step procedure, in which all eleven industries and

seven service sectors were entered. The procedure identiWed four in-

dustries as having a signiWcant eVect on migration (no service sectors

survived the procedure). These are chemicals, metal-working, and con-

struction, with positive coeVicients, and utilities, with a negative coeVi-

cient. The attraction of work in construction to migrants is not

unexpected: it oVered easy entry into a relatively unskilled urban occu-

pation. Moreover, the transitory nature of the work meant that in any

given period a number of workers would be on the move, as projects in

one area came to an end and others started up elsewhere. The negative

coeVicient on utilities suggests that few jobs in this sector were open to

migrants.

33 This might show that agricultural regions were no longer major sources of migrants,but as many migrants out of agricultural regions would not have had insurance cards, thisconclusion may not be justiWed.


(iii) Migration and trade cycles 1904–12

Annual data for movements into and away from major cities can be

obtained from the published series of town migration Wgures produced

as a result of the system of MeldepXicht, compulsory registration after a

change of residence. For 1904–12 a series can be produced covering 25

cities, which includes all the major cities with the exception of Munich.

Table 8B.3. Regression results for linear log-odds migration model:

1909–13

a b c d e

Constant �5.685 �6.104 �5.830 �5.822 �7.723

LOGDISTANCEkj �0.454 �0.455 �0.451 �0.453 �0.451

(38.4) (38.9) (38.1) (38.4) (38.8)

POPREGORIGINj þ0.378 þ0.380 þ0.379 þ0.379 þ0.380

(16.29) (18.09) (17.97) (17.94) (18.32)

INDUSTRY%k þ0.0170 þ0.0138 þ0.0162 þ0.0164 þ0.0267

(5.62) (4.49) (5.37) (5.44) (5.21)

SERVICES%k þ0.0637 þ0.0633 þ0.0630 þ0.0571 þ0.0799

(15.94) (16.03) (15.83) (11.91) (6.35)

REGORIGIN%AGj �0.0003

(0.14)

CHEMICALS%k þ0.0762 þ0.0699

(3.11) (2.74)

MACHINERY%k þ0.0356

(2.46)

METALWORK%k þ0.0553 þ0.0510

(4.56) (4.00)

CONSTRUCTION%k þ0.0495

(4.38)

UTILITIES%k �0.7230

(2.34)

N 930 930 930 930 930

S 0.9393 0.9289 0.9344 0.9362 0.9169

R2 0.713 0.720 0.716 0.715 0.728

adj R2 0.712 0.718 0.715 0.714 0.725

F-test of regression 460.1 474.6 466.9 464.3 307.9

RSS 815.2 797.2 806.7 809.9 774.2

Note: t-ratios in brackets.


Although there are only Wgures for nine years, it was still possible to

produce satisfactory econometric results (Table 8B.4) once Wrst diVer-

ences have been used (to deal with the high autocorrelation of residuals

shown in column a).

Various diVerent dependent variables related to economic conditions

were tried. Column b uses the change in unemployment; column d uses

the change in industrial production. The coeVicients on both are sign-

iWcant. The best results are those given in column c, possibly because

the applicants/vacancy ratio was obtained from the cities themselves,

and thus recorded conditions in these cities rather than conditions in

the economy as a whole. The results in column d are the weakest, and

the Durbin Watson statistic is within the inconclusive range

(d�L ¼ :773, d�U ¼ 1:332) for a sample of this size.

Table. 8B.4. Regression analysis of migration into major cities

1904–12


MIGt DMIGt DMIGt DMIGt

a b c d

Constant 4.335 �0.187 �0.190 0.790

UNEMPLOYMENTt �1.320

(�3.88)

D UNEMPLOYMENTt �1.046

(�5.78)

DAPPL=VACANCIESt �2.616

(�7.44)

DINDPRODUCTIONt�1 �0.223

(�2.31)

N 9 8 8 8

S 0.568 0.341 0.274 0.636

R2 0.683 0.848 0.902 0.471

adj R2 0.638 0.822 0.886 0.383

F-test 15.08 33.4 55.29 5.34

RSS 2.259 0.698 0.449 2.424

DW 0.55 3.08 2.91 1.01

Note: t-ratios in brackets.


9

Industrialization, Migration,

and Inequality

1. Introduction

The analysis contained in the preceding chapters has provided ample

evidence of the existence of labour surplus conditions in Germany in

the late nineteenth century. The sources of this surplus were various:

population pressure as a result of the demographic transition; the

release of underemployed labour in agriculture and cottage industry;

the increased use of seasonal migrant labour in agriculture; techno-

logical change which raised agricultural productivity; altered social rela-

tions inside agriculture as a result of changed economic circumstances.

The combined impact of these changes was to create a period when

labour supplies were highly elastic. High levels of rural–urban migra-

tion provided the motor for rapid economic growth. But this also meant

that Germany faced social and economic problems which were unpre-

cedented in the experience of European industrialization.

This amounts to a qualiWed endorsement of Max Weber’s views

about the origins of the agricultural labour surplus. The main qualiWca-

tion is that the release of surplus labour was brought about by other

factors besides Weber’s mechanism involving the decline of the patri-

archialische Arbeitsverfassung due to the spread of capitalist values. But

the basic point, that there is surplus labour within agriculture and that

this surplus is released as a result of changes associated with industrial-

ization, is endorsed.

Comparison with other countries shows that the German experience

was unique. No other European country had such a rapid transition to

an urban industrial society. The rate of growth of the major cities was

unusually high; the numbers migrating internally was also above the

levels in other countries. Only the United States went through a similar

experience, but this was in a very diVerent context, where most

migrants were immigrants.

Intertwined with these problems were other sources of diYculty. As

a late-industrializing country, with a massive need to import food and

raw materials, Germany had to expand exports rapidly, gaining share at

the expense of other countries. This meant that wages had to be kept

down. Germany’s position in central Europe made it a natural destin-

ation for migrants from other countries in central and eastern Europe.

This connected Germany to other ‘labour surplus’ countries and pro-

longed the labour surplus period in German industrialization. The re-

quirements of industrial growth also placed strains on the German

capital markets, resulting in the Wnancial crisis of the 1870s. A ‘capital

shortage’ phase coincided with the period of labour surplus.

The implications of the labour surplus period are set out in the

propositions of the Lewis Model: growth will be high as under-

employed labour is transferred to productive occupations; wages will

tend to lag behind productivity in the advanced or industrial sector; the

proWts share will rise; the beneWts of industrialization will go dispropor-

tionately to the owners of capital. Inequality in the urban sector will be

high and remain so until the labour surplus phase is over, when the

‘turning point’ is reached.

As noted in Chapter 1, the Lewis Model is one possible explanation

of the Kuznets Curve: the prediction that inequality will rise during the

initial stages of industrialization, and then fall as the economy reaches

maturity. Bringing together the views of Weber, Lewis, and Kuznets

produces a coherent model of economic development: the impact of

industrialization and the spread of capitalist values brings about the

release of surplus labour from agriculture; this holds back wage

increases in industry and raises the proWt share; which in turn leads to a

rise in inequality in the economy as a whole.

This chapter examines the evidence of a rise in inequality, making

particular use of Prussian tax statistics. This is an unusually valuable

source, which makes it possible to study the evolution of income

inequality in an industrializing society over a long period, something

which is rarely possible even with modern statistical materials.1

1 An important point is that there were no conXicts long enough to aVect the distribu-tion of income in the 1815–1914 period, something which makes it hard to draw generalconclusions from the movement of inequality in the major industrial countries in thetwentieth century.

294 Industrialization, Migration, Inequality

2. The Kuznets Curve and German H i story

The idea that, during industrialization, the relationship between growth

and inequality follows the path of an inverted ‘U’, with inequality Wrst

rising and then falling, has long been associated with Simon Kuznets,

whose 1955 article gave it wide currency. However, the idea that the

distributional consequences of industrialization might not be favourable

was certainly not a new one. Apart from its role in the socialist critique

of capitalism, it was also a concern of the German historical school, and

discussion of income distribution was an important component in the

German debate over Manchestertum: the pros and cons of laissez-faire

and free trade economics. Evidence that the income distribution was

worsening fuelled fears that the lower and middle classes were not

beneWting from industrialization to the same extent as the rich.

Kuznets was inXuenced by the same evidence, drawn from Prussian

tax records. The Soviet economist Prokopovitch had used these in a

study of income distribution, and, at the time of the 1955 article, this

was the only empirical support that Kuznets had for the initial rise in

inequality.2 As a result of Prokopovitch’s work, Kuznets was aware of

the diVerence between rural and urban inequality in nineteenth-century

Prussia. He found a similar pattern in studies of inequality in the

United States and India.3 This led him to a more general statement of

the relationship between inequality and economic development.

Migration had a central role in Kuznets’s own explanation of his

‘inverted U’ relationship. Kuznets’s model had three parts:

1. There are two sectors A (rural) and B (urban); incomes are higher

in B; there is migration from A to B;

2. Income inequality is higher in B;

3. Income inequality in B will tend to fall.4

Kuznets showed, using mathematical examples, that under certain

assumptions the Wrst part, on its own, could produce a ‘Kuznets

Curve’. This eVect was intensiWed by the second part. But Kuznets also

thought that income inequality in the three countries he was consider-

ing (Britain, Germany, and the United States) had in fact stabilized

while the Wrst two components were tending to cause it to rise, and he

2 Prokopovitch’s work was published by Keynes in the Economic Journal, Prokopovitch(1926).

3 Kuznets (1955), 7; the German Wgures came from Prokopovitch (1926).4 Kuznets (1955), 7–8 for parts 1 and 2, p. 17 for part 3.

Industrialization, Migration, Inequality 295

attributed this to the third component (for which he had no direct

evidence).

The Lewis Model provides an explanation for parts 2 and 3 of Kuz-

nets’s model: it explains high inequality in sector B, and it also predicts

that this will fall once the labour surplus phase is over.

Later Kuznets made a cross-section study of inequality in developing

countries which appeared to support his hypothesis. However, this

result has been disputed, and more recent studies of contemporary

evidence from developing countries have tended not to show an

inverted U-shaped curve, so that one participant at a conference was

led to declare: ‘the Kuznets curve is a Wction’.5

If the Kuznets Curve is a contemporary Wction, then this leaves some

questions. Does the historical evidence for the curve stand up? And, if it

does, how could it be that the relationship held for some industrializing

countries in the nineteenth century and not for their twentieth-century

successors? Is the Kuznets Curve ‘historical fact, but contemporary

Wction’? Is it just a ‘peculiarity of German history’?

Since Kuznets wrote his article some additional evidence of high

inequality during the early stages of industrialization has emerged. Lin-

dert and Williamson concluded, with reference to the United States in

the mid-nineteenth century, that ‘there apparently was a Kuznetsian

fall from grace as America embarked on the path of modern economic

growth’.6 Charles Feinstein has reinforced the pessimistic view of

working-class living standards during the British industrial revolution:

‘the majority of the working class . . . had to endure almost a century of

hard toil before they really began to share in any of the beneWts that

they had helped to create’. In the British context this was a continuation

of a pattern of high inequality already established in agriculture.7

In the next section of this chapter, new estimates of inequality,

derived from Prussian tax statistics, are presented which show that

Kuznets’s ‘inverted U’ hypothesis is relatively well supported for the

1820–1914 period. The peak of the curve was reached around 1900.

Thereafter inequality tended to decline. So, the Kuznets Curve is not a

Wction as far as nineteenth-century Germany is concerned.

5 Dani Rodrik, the discussion was published in Bruno et al. (1996), the remark is onp.159. For examples of the studies which led to this remark, see Anand and Kanbur(1993), Ravaillion and Chen (1997), and Dollar and Kraay (2000).

6 Lindert and Williamson (1980), 284.7 Feinstein (1998).


If the Kuznets Curve has some basis in historical fact, then this

suggests that it is one possible outcome of the process of industrializa-

tion, but not a relationship set in stone. Given favourable circum-

stances, some countries can avoid the initial rise in inequality; others,

facing less favourable conditions, may not experience the subsequent

fall postulated by Kuznets. Which leaves the question, what are these

favourable or unfavourable factors? In Germany’s case, what were the

reasons for the existence of a Kuznets relationship in the 1820–1914

period?

3. The German Debate

Germany, as one of the second wave of industrializing powers, could

observe the eVects of industrialization on British society and draw con-

clusions. There were horriWed reactions from German visitors ranging

from Friedrich Engels to Richard Wagner, who saw London as ‘the

fulWlment of Alberich’s dream—Nibelheim, world dominion, activity,

work, everywhere the pressure of steam and fog’.8

The German debate began once it was clear that Germany was

industrializing, and facing similar problems. The eVect of industrializa-

tion on income distribution was one aspect of this debate. Gustav

Schmoller, writing in 1895, summarized the position as follows:

Optimistic free-traders and defenders of the current social order have certainly

asserted that economic freedom and the current organisation of society will, on

their own, guarantee an increasingly equal distribution of income, while the

historical school, the more pessimistic critics of our time, and above all the

socialists have, in the main, come to the opposite conclusion.9

As this quote indicates, concern over the movement of the income

distribution was not limited to the left. A wide range of opinion feared

that industrialization would lead to widening inequality unless govern-

ment action was taken to prevent this. Considerable impetus was given

to the pessimistic view by an article by Ernst Engel, the director of the

Prussian Statistical OYce, published in the Zeitschrift des Koniglich

Preußischen Statistischen Bureaus in 1875. This contained an analysis

of the Prussian tax statistics which appeared to show that the beneWts of

8 Following a boat trip down the Thames to Greenwich in 1877, C. Wagner (1978–80).9 Schmoller (1895), 1067–8.


industrialization were going, in the main, to the higher-income groups.

Engel speciWcally referred to the way that the data seemed to support

Schmoller’s concerns: ‘There is no doubt that the concentration of

wealth has increased.’10

The historical school placed much more emphasis on the role of insti-

tutions, particularly the state, in the determination of income inequality.

Schmoller argued that there was a tendency for income inequality to

rise but that state action could counteract or even reverse this. This

was a small, but still signiWcant, breach with free market economics.

Although the historical school was not a large group, they were very

active, founding journals and producing surveys (including many

used as sources for this study), so they had a considerable eVect on

opinion.11

The view that the income distribution was worsening was challenged

by Soetbeer, who argued that the lower classes were in fact doing better

than Engel had thought. Engel was using income tax Wgures at a time

when the Prussian tax system was a hybrid, with the lower-income

groups paying a Klassensteuer, a tax which was intended to reXect ability

to pay, but which may have been less representative at times of rising

money wages. Soetbeer made a more systematic analysis of the income

distribution, which was less aVected by changes in the tax system. Adolf

Wagner, writing in 1904 but using tax data from 1822 onwards, took

the view that both the upper- and the lower-income groups were doing

well, but at the expense of the middle classes.12

In 1913 HelVerich made an analysis of developments since 1896

which came to the conclusion that, ‘Contrary to the much asserted view

that society is developing in a plutocratic direction, the lower-income

classes are gaining their full share of the increase in the national

income’.13 HelVerich had made a full analysis of the national income

and its distribution. His work was one of a series which made use of the

data produced by the revised Prussian tax system, which had been in

operation since 1892. This was showing a declining trend in inequality,

at least amongst taxpayers. It had also stimulated a discussion of the

10 Engel (1875), 142.11 For discussion of Schmoller’s views, and the role of the historical school, Dumke

(1991) and Grimmer (1999).12 Soetbeer (1882) and Wagner (1904). Soetbeer was a statistician who published on a

number of issues, including monetary reform.13 HelVerich (1915), 128. Karl HelVerich was a director of the Deutsche Bank and

secretary of the Treasury during the First World War.


statistical problems involved in the construction and analysis of income

distributions.14

Despite this progress, there are serious problems with the work of

these authors. One is the lack of an agreed standard for the measure-

ment of inequality. A second is a tendency to underestimate gains by

those below the income tax threshold. Both Soetbeer and HelVerich put

in unchanged Wgures for average incomes in this group. Yet, if an

income distribution is shifted upwards, so that the numbers falling

below some Wxed threshold falls, the average income of this group will

tend to rise unless there is some balancing shift within the exempt

group.15 A third problem is that comparisons are often between single

years, and over short time periods. A much longer view, less dependent

on estimates for single years, is needed.

4. The Pruss ian Tax Data : Problems and Opportunities

There are three main problems with the use of tax data to examine

income inequality. First, part of the population is normally exempt

from the tax, due to income levels below the tax threshold, or some

other reason for exemption. So, coverage is inevitably incomplete. Sec-

ondly, as the income assessment is linked to the tax payment, there is

an obvious incentive to minimize declared income and conceal sources

of income wherever possible. Some downwards bias must be antici-

pated. Thirdly, the tax system tends to change over time, which has

eVects on coverage, on incentives to evade, on deWnitions of income, on

procedures for checking returns, and so on.

These are serious problems, but not insuperable. There are two ways

of dealing with the Wrst problem. One is to divide the income received

by diVerent groups of taxpayers by total personal income, obtained

from national income accounts. This is an approach used in modern

studies by Atkinson and Piketty.16 It has been applied to the Prussian

income tax data by Geisenberger and Muller.17 There are, however, a

14 See Bresciani (1907) for a discussion of the statistical issues; this article anticipatesmany of the techniques used in modern studies, including this one. Other studies ofincome (and property) inequality include Kuhnert (1906 and 1916) and Biederman (1918).Constantine Bresciani (or Bresciani-Turroni) was an Italian economist who studied inBerlin in 1902–7.

15 A point which can be established by simulation using a log-normal distribution.16 Piketty and Saey (2001), Atkinson (2002).17 Geisenberger and Muller (1972).


number of problems with this approach. The main one is that it can

only be used for periods for which national income accounts are avail-

able, so the Geisenberger and Muller study starts in 1874. The second

is that the income side of the national income accounts is generally the

least reliable. Data on physical production is normally available from an

earlier stage than information on wages, rents, and proWts. HoVmann’s

estimates of income from capital and land in the nineteenth century

were derived using methods which have come in for criticism.18 More-

over, for a series which is derived from Prussian tax statistics, the

divisor should be based on income accounts for Prussia, and these are

problematic, as there is no information on inter-regional trade or trans-

fers of property income (post-1871 Prussia contained 61–62% of the

total German population).19

An alternative approach (which was the one used to prepare the

estimates given in this study) is to Wnd other sources of information on

the incomes of those excluded from the tax system. In nineteenth-

century Prussia, the most important low-income group was that con-

taining the landless agricultural labourers. Information on these, and on

incomes in low-paid non-agricultural occupations, can be used to con-

struct an overall distribution. This in turn can be used to derive a Gini-

coeYcient, or other measure of general inequality, whereas the Wrst

approach only yields a measure of the share of top taxpayers in total

income.

The second problem, the problem of evasion, is inherent in the use

of income tax data.20 It is a serious source of possible error, particularly

in cross-section comparison with other countries (where enforcement

procedures may be diVerent). It is a problem in a time series for a single

country if there are changes in the proportion of income which is not

declared, as a result of improved methods of oversight, for example. It

is clear from contemporary comment on the tax system that there were

18 Problems with HoVmann’s Wgures are discussed in Fremdling (1988 and 1995).19 The best income distribution Wgures are those derived from the Saxon tax statistics,

Jeck (1970); this source has also been used by Hentschel (1979) and Kiesewetter (1988).However, Saxony was a small and not very typical part of Germany, which industrializedearly and had a relatively equal land distribution. Given the importance of Prussia, with62% of the population in 1907, it is clear that any study which attempts to draw conclu-sions for Germany as a whole must make use of the Prussian data. More recently Germanstudies have tended to look at other aspects of inequality, such as social mobility—seeKaeble and Federspiel (1990) and Kocka (1983)—and there has been little work onincome distribution. Kaelble (1986) surveys work up to that date.

20 For a more detailed examination of the pros and cons of tax data as a source ofinformation on income distribution, see Atkinson (2002).


changes of this type.21 The reform of the Prussian tax system in 1891–2

led to a more rigorous examination of tax returns. It is important to

make some allowance for these changes.

The tax system was not constant in Prussia in the period, but evolved

in stages from a Klassensteuer, under which households paid a Wxed

charge, reviewable every three years, based on an assessment of ability

to pay (which took account of property owed, and indebtedness as well

as expected income), to a modern income tax system. Comparability

between the diVerent systems is not to be expected.

These last two problems were addressed by producing estimates of

income inequality for periods when the tax system was not subject to

major changes (1822–46, 1853–73, 1874–91, and 1892–1914), and then

looking for contemporary estimates or other data sources to chain the

series together. This proved to be possible for 1873–4 and 1891–2, but

not for 1846–53. During this period the Klassensteuer went through a

series of changes: urban areas which previously paid a diVerent tax were

brought into the system and the top bands were changed to a system

more explicitly linked to income. It was not possible to Wnd evidence

from other sources which could cover this period.

In addition to selecting an approach which should minimize potential

sources of error, an attempt has been made to quantify the eVect of

those errors which remain. By constructing diVerent variants for each

period, based on alternative sets of assumptions, it was possible to

examine the sensitivity of the Wgures to the assumptions used in the

central estimates, and Wnd extreme values produced by a range of pos-

sible assumptions. These variants were then used to calculate variance

estimates which were, in turn, employed to construct conWdence inter-

vals for the central inequality measures.22 Appendix A contains the

resulting estimates and conWdence intervals.23

The approach is one which is appropriate for a data source which has

serious deWciencies by modern standards. It generates a range of prob-

abilities as well as a single set of point estimates. Taking as an example

the weakest period, 1822–46, the tax system in these years was only

21 Meisel (1911) gives data on the number of returns rejected or challenged under thepost-1892 system. On average, a challenge to an assessment led to an increase of 25–30%in the amount of taxable income. See also Wagner (1880), 612, on the pre-1892 system.

22 This follows the approach set out in Feinstein and Thomas (2001).23 A full discussion of the issues involved in these calculations, together with details of

the calculations and references to the sources used, has been published separately in Grant(2002b).


indirectly related to income and coverage was not universal. Neverthe-

less, the data does provide certain pieces of evidence. The numbers in

Stufe 1, the highest band, rose by 60% between 1822 and 1846, while

the total population in the areas subject to the tax rose by only 39%.

The numbers exempt from the tax on the grounds of having insuYcient

property or income rose steadily, from 36.6% of the population in 1822

(including dependants) to 41.5% in 1846.24 These two factors strongly

suggest a hollowing out of the distribution, with numbers rising at each

extreme.

Moreover, there is evidence from other sources which suggests that

real wages for those at the bottom of the distribution were falling in this

period. Table 9.1 gives an index derived from Wgures for earnings in the

Prussian state forests, deXated by local food prices.

Taking these factors into account, the balance of evidence strongly

suggests a worsening of the income distribution in the period. But by

how much? This is less easy to say. The variance calculations produce a

95% conWdence range for the increase in the Gini-coeYcient of 12.2% to

16.2%, assuming that the central estimates are subject to uncorrelated

24 The Wgures of those exempt have been adjusted to allow for the eVects of anextension of exemptions to those aged 14–16 and over 60 in 1831.

Table 9.1. Rural wages in Prussia 1820/9–1875/9

(post-1815 boundaries)

1820–9 104.6

1830–9 101.1

1840–9 88.7

1850–9 81.9

1860–9 91.4

1870–4 91.3

1875–9 100.0

Note: Index based on earnings of male day-labourers in thePrussian state forests (1875–9 ¼ 100), deXated by local food priceindices.Source: Earnings in Prussian state forests come from Eggert (1883);this gives the results produced by a circular sent to all Oberforsterin November 1879 asking for the movement of wages since 1800.The survey was not repeated. Prices were taken from ZdKPSB(1907), 84–92.


errors. But examination of the highest and lowest variants (considering

the possibility that there are correlated errors produced by incorrect

assumptions), and applying variance calculations to these estimates,

produces a much wider range: an increase of between 4.3% and 21.4%.

This demonstrates that the statement that inequality worsened in the

period can be made with a high degree of certainty, even though the

exact scale of the increase is less easy to determine.

Thus, for all these caveats and notes of caution, the Prussian tax data

does present a major opportunity, provided that the data is handled

with caution. For 91 years, 1822–1913, the Prussian population paid

some form of income-related tax. Coverage was, at times, remarkably

high (74% in 1891). The period covers a society going through a pro-

cess of transformation from an essentially rural, pre-industrial economy

in the 1820s to a position as one of the world’s most advanced industrial

economies in 1913. Moreover, and in contrast to the twentieth century,

there were fewer unexpected shocks to the income distribution: world

wars with concomitant rises in taxation and prices, major world depres-

sions on the scale of the 1930s, unanticipated periods of high peacetime

inXation as in the 1970s. It is a unique opportunity to study the evolu-

tion of inequality in an industrializing society.

5. The Kuznets Curve in N ineteenth-Century Pruss ia

(i) Estimates for Prussia

Figure 9.1 gives the base estimates chained together to produce an

index. Because of the gap between 1846 and 1853 there are two base

years, 1846 and 1853. The estimated Gini-coeYcients are converted to

indices with these years set to 100.25

The shape is approximately that of the Wrst part of Kuznets’s

‘inverted U’. There was a downwards movement in inequality after a

peak in the early 1900s, though the extent of the fall was not large (a fall

of just 5% in the average Gini-coeYcient estimated for 1910–14 com-

pared to 1903–7). The main downwards phase of the Kuznets Curve for

25 The Gini-coeYcient is a measure of the extent to which the income distributiondiVers from perfect equality, all divergences being given equal weight. It is, in eVect, anaverage of the absolute values of all deviations from the average income level, expressed ona scale of 0 to 1, where 0 is perfect equality (all incomes are the same) and 1 is perfectinequality (one person has all the income).


Germany, as for other industrialized countries, came in the twentieth

century. Modern estimates give a Gini-coeYcient for Germany in the

1990s of 0.25, which is approximately the same level as the chained

estimate (using 1892–1914 as the base estimates and making no allow-

ance for changes in 1846–53) for Prussia in the 1820s.26 So the Prussian

or German Gini-coeYcient appears to have risen 35–40% in the nine-

teenth century, and fallen by a similar amount in the twentieth. This is

in line with Kuznets’s hypothesis.

In the Wrst period, 1822–46, industrialization was just beginning, and

Prussia was still a predominantly rural society. In these years, the main

inXuences on the level of inequality came from developments within the

rural sector. These include the eVects of the Prussian land reforms, the

consequences of a relatively high level of population growth, leading to

a period of quite severe demographic pressure, and the eVect of ruralexpansion in the east, bringing into cultivation land in remote areas

which were not well suited to market-orientated peasant agriculture. In

addition, agricultural prices fell in the 1820s and 1840s, which would

have disadvantaged the sector as a whole, without necessarily leading to

a worsening of inequality within agriculture.

26 Atkinson et al. (1995), 40.

1912190218921882187218621852184218321822

130

120

110

100

90

80

Fig. 9.1. Estimates of inequality from Prussian tax data 1822–1914: indices

of estimated Gini-coefficients (1846/1853¼100)


The second period saw the onset of a period of more rapid growth in

Germany. The start of more widespread industrialization is generally

dated to the 1850s. The tax statistics analysed by Engel seemed to show

that the fruits of industrialization were not being equally distributed.

Engel’s analysis was probably too pessimistic. The tax system in this

period was a hybrid. The richer taxpayers were subject to an income

tax, which was responsive to increases in earned and unearned incomes;

the less well-oV paid a revised version of the Klassensteuer. This was a

period in which, according to available evidence, wages were rising in

industry and agriculture. This may have led to a lag in the adjustment

of Klassensteuer rates to improved earnings. More importantly, it means

that any estimate of the overall income distribution should allow for

some improvement in the incomes of those below the tax threshold.

The base estimates for this period show an increase in the Gini-

coeYcient of between 4.3% and 6.0%, but the systematic error variants

show a much wider range, from 20.2% to 3.5%. This uncertainty is

largely caused by the question of the extent to which the incomes of the

exempt population moved up with other incomes. However, this period

is also identiWed as one when inequality most probably moved upwards.

The fruits of the early stages of German industrialization were not

evenly distributed.

In the next period, 1873–82, there was a change in the terms of trade

which was unfavourable to agriculture. The widening gap between agri-

cultural and non-agricultural earnings had the eVect of pushing up

inequality. As the exempt population contained large numbers of land-

less agricultural labourers, the assumptions made with regard to this

group have a signiWcant eVect on the level of overall inequality. The

base estimates show an increase in the Gini-coeYcient of between

11.6% and 13.3%. The systematic error variants show a range from

13.6% to 5.7%. The lower Wgure comes from a variant which is based

on a more optimistic view of the movement of real earnings in agricul-

ture in the period.

So the Wrst three periods studied, covering 1822 to 1892, less the

omitted six years in 1846–52, all show strong evidence of rising inequal-

ity, even when allowance is made for possible sources of error or bias.

The upwards phase of the Kuznets Curve is relatively established for

nineteenth-century Prussia. There were a number of diVerent factors at

work, but the general trend was towards greater inequality.

The Wnal period, 1892–1914, can be divided into two: a Wrst part inwhich inequality may have continued to rise, and a second one in which


inequality began to fall, albeit slowly and uncertainly. The break came

around 1906. Given the nature of the data, there are considerable diY-

culties with annual movements in the inequality measures. The main

problem is the estimation of the movement of the income levels of the

exempt. There are a number of possible wage series, but these show

diVerent patterns of cyclical movements.27 Thus, while the base esti-

mates for the period 1892–1906 show an increase in the Gini-coeYcient

of 5.5–6.0%, the range of possible variants runs from þ8.3% to

�1.3%. The period is best regarded as one of relative stabilization,

with Xuctuations around a core level which may have still been moving

upwards. The exact timing of the peak is uncertain, but it probably lay

in the early 1900s.

In this period, the recovery of the German capital markets, combined

with the ending of the main period of labour surplus migration out of

agriculture, produced an improvement in conditions in urban labour

markets. The increase in real wages still lagged behind productivity,

but this owed more to the adverse movement of the terms of trade. As

German industry forced its way onto world markets, it had to keep

prices low, and this restricted wage rises.

From 1906 onwards there was a slight decline. The base estimates for

1906–14 show a fall of 3.1–3.4%, with a range for the other variants of

between �5.4% and �2.7%. There was a cyclical rise in inequality in

1913–14 which reduced the overall decline to 1914.28 But, making some

allowance for these Xuctuations, the evidence for this period indicates

that the downward phase of the Kuznets Curve had begun.

Although the timing of some of these movements can be uncertain,

the overall trends are clearer. Throughout most of the nineteenth cen-

tury inequality was rising. In the 1890s this upwards movement came to

an end, and, after a period of Xuctuation, with no strong trend upwards

or downwards, inequality began to fall, although the fall before 1914

was a slight one.

(ii) Comparison with other countries

To get some idea of the scale of inequality in Prussia in the period, it is

necessary to make comparisons with other countries. This is the only

27 The main wage series are those given in HoVmann (1965), Desai (1968), Bry (1960),and Hohls (1995).

28 The tax system worked in arrears, so the assessments for 1914 were not aVected bythe outbreak of war in that year (as would have been the case with a PAYE system).


way to create a standard of measurement, to decide whether a given

Gini-coeYcient is ‘high’ or ‘low’.

The peak Gini-coeYcient for Prussia in 1906 is estimated to have

been 0.34 (Table 9A.5 in Appendix 9A). The World Bank data set

measuring inequality in modern economies shows that only seven out of

the 70 developing economies had lower Gini-coeYcients than this. So,

even at the 1906 peak, Prussian incomes were more equally distributed

than in most developing countries today.29 On this basis the overall

level of inequality in Prussia in the period appears to have been rela-

tively low for a country going through the disruptions of industrializa-

tion. However, the increase over levels in the earlier part of the century

may well have been a factor creating resentment and pessimism over the

future distribution of the rewards of industrial progress.

Another point of comparison is the experience of other European

countries in the late nineteenth century. Figure 9.2 compares the Gini-

coeYcients estimated for Prussia with estimates for Britain given by

Charles Feinstein. The Prussian estimates are chained backwards from

the estimates for 1892–1913 (which are the most reliable). ConWdence

intervals are also shown (from Appendix 9A). It is not possible to apply

the same procedure to the British Wgures, but, given the nature of the

data, it can be assumed that the conWdence intervals would be larger

than the Prussian ones, even those given for the 1822–46 period.

It appears that the movement of inequality in Britain and Prussia

followed diVerent paths. Inequality in Britain was consistently high,

and not much aVected by industrialization. Estimates produced by

Peter Lindert show a rise in inequality in the late eighteenth century,

but Britain was already a highly unequal society before industrializa-

tion.30 The estimated Gini-coeYcient for 1913 (0.48) would put Britain

in the top half of the World Bank data set, with a higher level of

inequality than 45 of the 70 developing economies.

This is not surprising given the much less equal distribution of land

ownership in Britain. Industrialization brought about a transfer of

population out of an agricultural sector which was itself highly unequal.

The fact that inequality did not rise as a result of industrialization was

due to the fact that inequality was just as high inside agriculture as

outside.

29 Deininger and Squire (1996), table 1; the calculation excludes Eastern Europe andthe former Soviet economies. There are massive problems of comparability in an exerciseof this type. For a discussion of some of these see Atkinson and Brandolini (1999).

30 Lindert (1994).


The Prussian experience was very diVerent. Inequality in the early

part of the century was relatively low, reXecting the more equal distri-

bution of land ownership. But industrialization was accompanied by an

increase in overall inequality. There are a number of possible reasons

for this. There was an increase in the number of landless labourers in

the early part of the century, at a time when real wages in agriculture

were falling. This could have been responsible for part of the increase

in inequality in the 1822–46 period. There was a rise in the urban–rural

wage gap in the 1870s, which would have increased overall inequality in

this period. Both these factors were, however, relatively short-term

eVects. The longer trend needs a diVerent explanation.

19071892187718621847183218171802

0.6

0.5

0.4

0.3

0.2

0.1

Prussia

Upper 95% C I

Lower 95% C I

Britain

Gini-coefficients

Fig. 9.2. Estimates of inequality from Prussian tax data, 1822–1914 (including

calculations of upper and lower 95% confidence intervals), compared with

figures for Britain

Note: The 1846–53 gap is covered by assuming that the Gini-coefficient for 1846 was thesame as that for 1853. An important point to note is that the probability of a movementfrom the upper limit of the 95% range to the bottom is 1/1600, which means that theconfidence intervals for rates of change are tighter than might be assumed from thisgraph: see the calculations presented in Appendix 9A.

Source: British figures from Feinstein (1988), 723.


The most likely cause was a shift of population from a relatively

equal rural sector to a less equal urban or industrial sector. There are

two sources of information on inequality in the urban sector. One is the

tax data for the period after 1896. A breakdown into two sectors is

available for a number of years. These have been used to produce

estimates of inequality which are given in Table 9.2. Some caveats

should be noted. These are based on a much cruder breakdown than

that used for the estimates given in Appendix 9A, just six tax-paying

classes plus the exempt. Consequently margins of error are likely to be

rather larger than those given in the appendix. In addition, the same

Wgures for household size by income class are used for both sectors,

which is a possible source of bias.31

The results show that urban Gini-coeYcients were 28% higher than

rural on average. So, a shift of population from one sector to another

would have an eVect on overall inequality.32 The table also shows that

inequality in both sectors fell substantially in 1900–13. This, however,

would have been partially balanced by the eVect of the continuing shift

of population into the less equal urban sector.

A second source of evidence comes from budget surveys in urban

areas. Results from these are given in Table 9.3. Again there are caveats

to note. These are based on household frequency distributions with no

information about household size. This has the eVect of raising the

Table 9.2. Gini-coeYcients estimated for rural and

urban sectors from Prussian tax data, 1896–1913

Urban Rural

1896 0.429 0.351

1900 0.438 0.349

1905 0.396 0.321

1910 0.362 0.261

1913 0.355 0.265

Source: Income tax Wgures from Biedermann (1918); income of the exemptcalculated using HoVmann’s Wgures for wages in industry and agriculture.

31 Household size was deWnitely larger in agriculture, but this would only be a sourceof bias if the diVerence was not the same for all income classes.

32 Overall inequality is not, however, just a weighted average of the two sectors. Thesize of the gap between the sectors is also important.


estimated Gini-coeYcient by around 0.08 (many low-income house-

holds had few dependants). The surveys were undertaken at diVerent

times, using methods of recording and selection which could have

varied considerably. This may well explain some of the surprising

jumps in inequality shown for Hamburg, for example.

Allowing for these reservations, the results still show a pattern of

inequality substantially higher than the estimates given for the rural

sector. They indicate that rural–urban migration would have been a

powerful force driving up overall inequality. And, more broadly, they

support the view that Germany went through the ‘labour surplus’ phase

of the Lewis Model in this period, and that this was driving up inequal-

ity in urban areas. These were the circumstances under which the

urban voters started to turn to the Social Democratic Party.

6. The Kuznets Curve : Contemporary F ict ion?

H i storical Fact? German Peculiarity?

If the Kuznets Curve was a product of historical circumstances, which

may be speciWc to Germany, this leaves a question: what was it that

produced the relationship in this period and not in others?

Two sector models, connected by migration, have a certain value in

the elucidation of points such as the importance of migration costs and

the signiWcance of expectations. But they also obscure other points,

such as the role of technological change within sectors. They focus

attention on changes in the relative importance of the diVerent sectors,

not on changes within sectors.

Looking at the process of development in the more advanced econ-

omies over the past 200 years, this can be summarized as an alteration

Table 9.3. Gini-coeYcients calculated from household budget

surveys

Hamburg 1867–8 0.651 Leipzig 1875 0.643

Hamburg 1873–4 0.467 Leipzig 1885 0.613

Hamburg 1881–2 0.720 Leipzig 1900 0.605

Hamburg 1890–1 0.534 Dresden 1880 0.559

Hamburg 1900–1 0.496 Breslau 1880 0.542

Breslau 1900 0.595

Source: Calculated from survey results given in Kuczynski (1908).


of the production function for the economy as a whole. The pre-indus-

trial production function was one in which, by deWnition, the level of

productive knowledge was low, and the stock of tangible capital was

also low. Physical eVort played an important part in the production

process, and the other major factor was the quantity of suitable agricul-

tural land. It might also be noted that the productive knowledge which

was available was rarely covered by patent rights, and so there were few

‘owners’ of such knowledge able to charge for its use: medieval har-

nesses for draught animals were major technological improvements but

nobody held a patent.33

The modern industrial or post-industrial economy is one in which

agricultural land has little importance in the overall production func-

tion. Physical eVort is also relatively unimportant. The stock of tangible

capital is much higher. There has also been a massive increase in

the level of productive knowledge, and a substantial part of national

income now goes to those who possess this knowledge, as skills, patent

rights, experience gained through learning-by-doing, or unique capabil-

ities possessed by Wrms and other institutions.34

The modern debate about income distribution mainly revolves

around the distribution of skills and the relative rewards received by

skilled and unskilled labour. Ways to improve the overall income distri-

bution include the provision of training for those at the bottom of the

income distribution, and the achievement of high and uniform stand-

ards in education. The modern German vocational training system,

which had its roots in the pre-1914 period, is generally seen as a par-

ticularly good example of these measures.35 German vocational training

was already admired and studied by other countries in this period. So,

if the wider diVusion of skills is indeed a major factor tending to bring

about a more equal distribution of income, then this factor was already

operating in pre-1914 Germany.

There are some reasons to expect that skill acquisition and diVusionmay lag during the earlier stages of industrialization and migration has a

part in this. Vocational training typically takes place immediately after

the end of formal education, which was around the age of 14 in

Germany in the period. Relatively few migrants had arrived at their

destination by this age. Only 13.9% of those arriving in Berlin in

33 Mokyr (1990), 36–8.34 The role of human capital in the movement of the income distribution is discussed

in Atkinson (1996a).35 Nickell (1996).


1881–5 were aged less than 14 at the time of arrival.36 Migrants arriving

later would Wnd it more diYcult to obtain formal vocational training.

They would be too old to be accepted as apprentices. Employers might

be more likely to oVer training to the children of resident workers than

to ‘footloose’ migrants. Moreover, from the point of view both of em-

ployer and of employee, the prospective rate of return on investment in

vocational training would fall with age, as the anticipated period of

future employment was reduced.

Migrants could receive training prior to migration, but this would be

less valuable. Most vocational training involves a practical element,

training ‘on the job’. This is only possible in the vicinity of the industry

concerned: shipyard apprenticeships will only be available in towns

where there are shipyards. Even more general courses are less likely to

be available in sparsely populated rural areas. The establishment of a

course on electrical engineering is only possible when there are a suY-cient number of prospective students in the catchment area to make the

course a viable proposition. Such agglomeration eVects will tend to raise

skill levels in cities and the areas around them; rural migrants will not

have the same opportunities.

Kuznets made a similar point, though with less emphasis on skill

acquisition:

The very fact that after a while, an increasing proportion of the urban popula-

tion was ‘native’, i.e. born in cities rather than in the rural areas, and hence

more able to take advantage of the opportunities of city life in preparation for

the economic struggle, meant a better chance for organisation and adaptation, a

better basis for securing greater income shares than was possible for the newly

‘immigrant’ population coming from the countryside or abroad.37

The consequence of this is that there will be a delay after the onset of

industrialization before skill diVusion starts to aVect the distribution

of income. As for the distribution of rewards to holders of other forms

of productive knowledge, such as patent rights, in the modern economy

this knowledge is more likely to be held by corporations than individ-

uals, and so this is bound up with the broader question of the distribu-

tion of capital ownership. However, it can be said that the initial

establishment of property rights over certain forms of productive know-

ledge, through patent laws and laws on copyright, would tend to increase

36 Calculated from Berlin, Statistisches Amt (1887), table V.37 Kuznets (1955), 17; skill acquisition is also an important component in Williamson’s

model, Williamson (1985).


inequality unless there are oVsetting factors (the increase in the import-

ance of this source of income might be at the expense of another form

of income which was even less evenly distributed, such as agricultural

rents).

In modern societies the ultimate ownership of capital is relatively

widely spread, with institutional investors such as pension funds, with

large numbers of beneWciaries, holding a high percentage of total equity.

Institutional improvements which widen the range of savings options

available to the less wealthy, and which, by bringing down the cost of

intermediation, reduce the gap between the rate received by small

savers and the rate of return on industrial investments, will tend to

decrease inequality. The arguments presented in the last chapter sug-

gested that only a small group of relatively wealthy investors were able

to share in the high proWts available from industrial investments in the

early stages of industrialization. As the Wnancial system developed, a

greater range of investors came to share in these beneWts.

If the mature industrial economy is one in which a wider diVusion of

skills and improved savings opportunities are tending to reduce in-

equality, it does not necessarily follow that, in the absence of these

factors, a country in the early stages of industrialization will experience

a rise in inequality. The rising importance of capital may give a labour-

saving bias to the development of technology in the early stages of

industrialization, as argued by JeVrey Williamson, but the eVect of thisfactor could be negated by a decline in the share of rents, if land

ownership is highly concentrated, as agriculture becomes less import-

ant.38 Much will depend on the starting position: the distribution of

income in the pre-industrial economy. This in turn will be heavily

inXuenced by inequalities in land ownership.

Kuznets’s view was that ‘inequality in the percentage shares within

the distribution for the rural population is somewhat lower than in that

for the urban population’.39 His evidence came from a comparison of

income distributions in India, from US data for the period before the

Second World War, and from Prokopovitch’s study of Prussian tax

data. But these studies may demonstrate the importance of institutional

factors in patterns of land ownership. The United States, for example,

had a relatively equal rural income distribution partly as a result of

38 Williamson (1985), 86–90; Williamson’s model is discussed in Feinstein (1988).39 Kuznets (1955), 7–8.


successive Homestead Acts which had favoured the establishment of

medium-sized family farms.

Other factors which aVect land ownership include climate, infrastruc-

ture, and geographical isolation. There are certain tropical crops which

are associated with ‘plantation’ production systems, due to economies of

scale favouring larger units.40 Where public sector provision of infra-

structure is poor, so that farmers have to provide this themselves, the

cost of this will be a problem for all except the largest estates. This

eVect is intensiWed by geographical isolation. The current expansion of

large-scale soya-bean production in the Brazilian Mato Grosso is a good

example of the eVect produced by a lack of infrastructure when com-

bined with geographical remoteness. The example of US agriculture

shows that soya beans can be successfully produced on small and

medium-sized units, given favourable circumstances.41

The state has an important role in this, both because it sets the

legislative framework and because it is responsible for infrastructure

investments. The United States in the nineteenth century was suY-

ciently democratic to be able to carry through a land settlement policy

broadly favourable to the interests of small and medium-sized units; the

modern Brazilian state does not appear to have either the desire or the

capacity to do this.42

Even in conditions favourable to large-scale production the state can

alter the playing Weld so as to prevent the establishment of larger units.

British colonial policy in Fiji was hostile to the cultivation of sugar cane

on plantations, with the result than many Indian immigrants brought in

as indentured labourers were eventually able to set up as small farmers

(on leased land). The much-disputed current EU banana import policy

is an attempt to protect small-scale producers on Caribbean islands

from competition from plantations in central America.

Societies make choices which aVect land ownership, and these are

often choices which reXect the balance of political power. Two eight-

eenth-century constitutional monarchies, Britain and Denmark, made

diVerent choices, which resulted in very diVerent patterns of land

ownership. The range of crops grown was similar (the Danish shift to

40 Engerman and SokoloV (1997) make this point.41 Article on the Mato Grosso in the Financial Times, Survey of Brazil (3 Oct. 1994), 16.42 In the Brazilian case the problem may well be that the cost of dealing with the social

problems created by one of the world’s most unequal income distributions leaves littlemoney for infrastructure investments in remote regions. In which case, this is an exampleof the way that high inequality can breed further inequality.


livestock production came later); neither country was geographically

remote. In the British case, the political system produced an outcome in

which the interests of land-owners were placed above all other consider-

ations.43 The Danish reforms were more beneWcial to the peasantry.

It has also been argued that a more equal distribution of land, and a

more equal starting point, will promote skill acquisition and thus pre-

vent a rise in inequality in the urban or industrial sector.44 Putting a

child into secondary education or some form of vocational training is a

considerable investment for families in poorer countries: there is the

cost of feeding and clothing, there are often fees to be paid, but most

important of all there is the loss of potential earnings, a valuable add-

ition to the family budget. With high initial inequality, and borrowing

constraints on poorer families, rather fewer will be able to aVord this

investment.

Statistical results which support these views are inXuenced by the

fact that, since 1945 there have been a number of ‘high growth–high

equality’ cases, countries with relatively equal land distributions which

have achieved high growth rates: Japan, South Korea, Taiwan, and the

Irish Republic are some examples.45 There have been others with high

levels of inequality in land ownership, which have had diYculty sus-

taining high growth rates even when these have been achieved for short

periods: Brazil is a good example.

Consideration of the four ‘high growth–high equality’ cases reveals

that in all these countries there was a reallocation of property rights in

agriculture at a time when some exceptional event made it possible to

achieve this without it being seen as a precursor of further measures. If

there was an expectation of repeated expropriations, this would aVect

investment in agriculture and in other sectors where there might be

political pressure for a reallocation of property rights.

These events include defeat, or near-defeat, in war, or in a civil war,

and the end of colonial rule. Prussia is a good historical example. Defeat

by Napoleon at Jena made it possible to carry out a land reform without

creating the expectation that similar measures detrimental to the free

exercise of property rights would be introduced at some future date.

These ‘Jena-like’ windows of opportunity appear rarely, if at all. In

their absence, property rights tend to evolve slowly. Sometimes,

43 The importance of starting points was also emphasized by Williamson (1991).44 Perrson and Tabellini (1994), Galor and Tsiddon (1994), and Clarke (1995); see also

Bernabou (1996) for a survey of the arguments.45 There is an interesting study of the Taiwan case in Fei et al. (1979).


technological changes, such as those which (arguably) made the enclos-

ure of common Welds advantageous, can create conditions where suY-

cient gains can be made to compensate some owners for the loss of their

property rights, while reallocating these rights to others. In general, the

declining relevance of feudal rights and obligations made it necessary

for these to be recast in a more modern form, but this did not deter-

mine the eventual outcome.

These factors produce a more complex mechanism than the one

outlined by Kuznets, and one which is heavily inXuenced by historical

circumstances. There are now many possible paths, not just one. Alter-

native paths include some which are radically diVerent. A country with

a extremely high level of inequality in land ownership might experience

little or no increase in inequality during the early stages of industrializa-

tion simply because the starting position was so high.46 If political

problems, or low skill acquisition, then holds back economic growth,

this country may never break through to economic maturity and the

downward phase of the Kuznets Curve. It could become stuck in a high

inequality–low growth trap: an initial pattern of high inequality being

perpetuated through low skill acquisition.

A second country with a highly equal land distribution may avoid the

Kuznets Curve for completely diVerent reasons. The transition to the

downward phase could be so fast that the initial rise in inequality would

be insigniWcant. If high skill diVusion is established right at the onset of

industrialization then this factor could outweigh other forces tending to

increase inequality. In which case there would be no Kuznets Curve:

inequality would start low and remain low.

Kuznets explained high inequality in his sector B, the urban or

industrial sector, using an argument which contains elements of the

Lewis Model:

It seems most plausible to assume that in earlier periods of industrialisa-

tion . . . [sector B’s] income distribution was more unequal than that of the

agricultural population. That would be particularly so during the periods when

industrialisation and urbanisation were proceeding apace and the urban popula-

tion was being swelled, and fairly rapidly, by immigrants—either from the

country’s agricultural areas or from abroad.47

46 Lindert’s Wnding that inequality in Britain started high and remained high mayindicate that Britain was in this category, Lindert (1994).

47 Kuznets (1955), 16.


He also saw the concentration of savings as an important factor tending

to raise inequality in sector B.48 Lewis himself endorsed the use of his

model as an explanation of high inequality: ‘the elasticity of supply of

unskilled labour is so high (whether in the dual or the classical models)

that it dampens real wages . . . The fruits of growth are concentrated in

few hands.’49

This study has looked at the labour surplus assumption of the Lewis

Model in some depth. The conclusions are broadly favourable to the

view that there is a labour surplus phase in industrialization, though

much depends on circumstances: in the German case the inXuence of

geographical location, in particular the proximity of Russian Poland,

has been shown to be considerable.

However, there are two points which have emerged which have

important implications for the Kuznets Curve. The Wrst is demo-

graphic. In Germany the onset of the fall in birth rates characteristic of

the ‘demographic transition’ came early: there was a downward trend in

the Princeton Im index (the index of births within marriage allowing

for the age distribution of married women) from the late 1870s

onwards. This was led by the urban areas. In 1880–1 marital fertility in

cities of over 20,000 inhabitants was 12.9% below the Wgure for rural

areas; by 1905–6 it was 31.4% below, having fallen by 25.6% compared

to the fall of just 5.5% in rural areas.

But this fall was itself a continuation of a pattern of behaviour

already present in rural society in Western Europe: the restriction of

population growth principally through delayed marriage. The main

diVerence was that the population in the urban areas began to marry

earlier and to restrict fertility within marriage. This, however, was not

unusual, as Protestant rural areas followed a similar pattern of behav-

iour; in Catholic areas, by contrast, people married later. The situation

in many contemporary developing economies is very diVerent. As

Todaro put it: ‘LDC (less-developed country) birth-rates today are

substantially higher than they were in pre-industrial Europe. This is

largely because of early and almost universal marriage in contemporary

Third World countries . . . since the initial level of birth-rates in West-

ern Europe was generally low as a result of either late marriage or

48 Ibid. 6–8.49 In Ranis and Schultz (1988), 14; Williamson commented (at the same conference):

‘the upswing of the Kuznets curve is not an unambiguous fact of history’, ibid. 23. Taylor(1979) and Taylor and Anda (1988) explore the distributional implications of the LewisModel.


celibacy, overall rates of population growth seldom exceeded the 1%

level even at their peak.’50

The result has been that many development economists take a rela-

tively pessimistic view of the likely pace of population increase: Hayami

writes (in 1997): ‘the probability of low-income countries escaping

quickly from the strong pressures of exogenously given population

growth appears to be relatively slim over the next couple of decades’.51

If the Kuznets Curve is indeed ‘historical fact but a contemporary

Wction’ then this very diVerent pattern of demographic behaviour is

probably part of the explanation. The labour surplus phase in the in-

dustrialization of Western Europe was curtailed, partly by the rapid

decline of the urban birth rate, and partly by the fact that birth rates

were already controlled in rural areas. Hence, the more rapid transition

to the downward phase of the Kuznets Curve.

The point is also of interest for the light it sheds on household

motivation and behaviour. Take, for example, the Wnding reported in

Chapter 5, that birth rates were lower in areas of partible inheritance.

This suggests a concern for the well-being of subsequent generations

and a willingness to adapt behaviour to secure an improvement. Too

many children threatened the viability of the family holding. So

numbers were restricted. This may not be entirely altruistic; parents

may have considered that their own interests, in having children able to

provide support in their old age, were better served by having relatively

few children who were better-oV. But it also suggests that, if a move to

a city brought improved educational and training opportunities, then

similar considerations would lead to a decision to restrict family size so

that available resources were suYcient for children to make use of these

opportunities.

The second point concerns the capital side of the Lewis Model.

DiVerential saving rates, especially in the form used by Lewis, ‘only

capitalists save’, create a powerful mechanism which may well drive up

inequality in an economy which does not draw on international sources

of capital. However, once the possibility of large-scale international

capital Xows is introduced into the equation, then the domestic savings

rate becomes less relevant. With perfect capital mobility, and zero

transactions costs (for turning international savings into domestic in-

vestment), the whole Lewis Model disappears, as does any concern

50 Todaro (1997), 196 and 200. 51 Hayami (1997), 60.


with the rate of population growth and the agricultural labour

surplus.52

There was certainly a degree of capital mobility in Western Europe

in the mid-nineteenth century, but the largest Xows were those associ-

ated with the international bond market: there was little direct invest-

ment in industry. In the last quarter this began to change, and there

was much more direct investment in those countries which began to

industrialize in this period: Italy, Russia, and Austria-Hungary. The

electrical industry, which also emerged in this period, was dominated

by a small group of German and American Wrms which had subsidiaries

in diVerent countries. But this was a relatively new development.

The position facing contemporary developing countries is not the

same. There is a large pool of international capital available for invest-

ment in all sectors of the economy. Provided it is able and willing to

oVer conditions favourable to the interests of foreign investors, a suc-

cessful developing economy should be able to attract suYcient invest-

ment to avoid Wnding itself with surplus labour, even with high

population growth and large-scale migration out of agriculture.53

Given these factors, it is not surprising that modern studies do not

Wnd a Kuznets relationship between economic growth and inequality.

The more successful developing economies have managed to avoid the

rise in inequality that Kuznets predicted. The less successful have failed

to make the breakthrough to economic maturity and the downward

phase of the curve. The Kuznets Curve is one path among many

possibilities.

7. Sources of Inequality in Germany 1870–1913

(i) General comments

From the discussion in the previous section it appears that there are

ways whereby a society going through the early stages of industrializa-

tion can avoid a rise in inequality. How well does nineteenth-century

52 The consequences of diVerential savings were examined by Kaldor (1956) and Pasi-netti (1962).

53 This is not, however, something that is either easy to achieve, or costless to thedeveloping country. Dependence on external Wnance can lead to instability as a result ofchanging sentiments in international Wnancial markets, which is one reason why successfuldeveloping economies such as Singapore or South Korea have taken steps to encouragedomestic savings.


Germany measure up to these standards? Was the rise shown in Figure

9.1 inevitable?

Germany started with a relatively equal distribution of income, due

to a more equal distribution of land ownership, which was generally

more egalitarian outside Prussia, with some exceptions such as Meck-

lenburg. Historians have criticized the land reforms of the early nine-

teenth century on the grounds that they left a relatively unequal

distribution of land ownership east of the Elbe and also that there was a

rise in the numbers of landless labourers. But a broader comparison

shows that the reforms, coupled with the previous policy of

Bauernschutz which protected and entrenched peasant rights, were at

least partially successful.54

Unlike the most successful of the modern developing countries,

Germany was not able to avoid a rise in inequality during industrializa-

tion. In comparison with these countries, Germany suVered from two

main disadvantages: there was little foreign direct investment (by

modern standards), and Wnancial institutions were relatively under-

developed. In particular, the system of universal banks closely tied to

industry, which Germany pioneered and which has reappeared in other

countries such as Japan or South Korea, was then relatively unknown.

In consequence, there was a period when industrial investment was

held back, more by institutional problems than by a general shortage of

capital, which ran from the 1870s to the mid-1890s.The shift of the terms of trade against agriculture in the 1870s would

not, under reasonable assumptions about factor immobility, have led to

a permanent rise in inequality. However, it had a powerful short-term

eVect, and it was particularly unfortunate that, just as the movement of

the international terms of trade required an increased rate of migration

from rural areas into cities, the rate of industrial investment should be

held back by the institutional problems mentioned in the last paragraph.

It also coincided with a period when surplus labour was being released

from agriculture as a result of institutional changes, such as the decline

of the contractual labour system studied by Weber.

German conditions were also inXuenced by geography, in particular

the proximity of Russian Poland, which had a considerable inXuence on

the structure of rural wages. By restricting wage growth this factor

54 Historians who have analysed the agrarian reforms include Conze (1969), Dickler(1975), Harnisch (1986), Koselleck (1967), Lutge (1963), Plaul (1986), Perkins (1986), andSaalfeld (1963).


contributed to overall inequality even in areas where the numbers of

migrants were small. Competition from migrants from eastern and cen-

tral Europe also aVected German migration to the United States, but it

was fortunate that this eVect was mainly felt after 1895, by which time

the ‘labour surplus’ period was over. During the period up to 1895, the

safety valve of emigration to the United States was open. It reduced the

pressure on urban labour markets inside Germany, and thus helped to

prevent an even steeper rise in inequality.

Given these factors, it is probable that high levels of inequality were

inevitable in the urban, or industrial, sector in 1870–95. Nothing much

could have been done to avoid this. Higher levels of agricultural protec-

tion could have held back the rate of decline of rural employment, but

at the cost of a signiWcant reduction in industrial growth, and there

would also have been a detrimental eVect on urban wages. Examples

from modern developing countries show that, under similar circum-

stances, the introduction of measures to raise urban wages creates

‘urban bias’, and a division of the urban labour market between a rela-

tively well-paid formal sector, and a low-paid, or underemployed, infor-

mal sector. Some measures might record this as an increase in

inequality.

The same point applies to the activities of trade unions. In this

period, trade unions were better organized in the more skilled sectors,

so their inXuence, in so far as they did aVect wage levels, would have

tended to widen divisions within the urban labour market.

The deliberate restriction of internal migration through government

action, a practice followed in a number of developing countries (notably

China), would have had the eVect of widening the overall income distri-

bution by holding a number of potential migrants in low-paid rural

employment. It is a policy which is only likely to appeal to governments

which give little weight to the interests of those living in rural areas.

The factors that tend to bring the initial rise in inequality to an end

have been identiWed as higher investment and increased skill diVusion.

In both these areas the German record can be described as exemplary:

it was one which other countries have sought to emulate. Both the

vocational training system, and the Wnancial system as it developed in

the 1890s, were well suited to the requirements of the German economy

in this period, and these were the forces which brought about the

decline in inequality, in combination with the recovery of German

agriculture after 1895, which reduced the urban–rural gap and also

decreased the pressure on urban labour markets.


(ii) Analysis by simulation

The Wrst year for which separate distributions for both the rural and

urban sectors are available is 1896. These can be used to simulate the

eVect on overall inequality of diVerent scenarios. Between the 1820s

and the 1890s there was a total increase in inequality of around 40%;

this is the change that has to be explained.

Table 9.4 presents simulations which show that, using Kuznets’s

assumptions, migration will push up inequality (scenarios C and E) but

that this eVect can be reversed if more realistic assumptions are made,

so that migrants come from the lower part of the rural distribution and

move to the lower part of the urban distribution (scenarios D and F).

So migration, in itself, is not a very convincing explanation of the

Kuznets Curve.

Table 9.4. Simulations using 1896 rural and urban income

distributions

I. Using 1896 sector weights: 40% urban, 60% rural

Scenario EVect on overall inequality

A All rural incomes fall 10% Gini-coeYcient rises 2.6%

B Incomes of poorest 50% in rural

sector fall 10%

Gini-coeYcient rises 3.8%

C 10% of rural population moves to

urban sector—no eVect on sector

distributions (Kuznets process)


D 10% of the poorest half of the rural

population move to bottom half of

the urban sector

Gini-coeYcient falls 2.1%

II. Using circa 1850 sector weights: 10% urban, 90% rural

E 10% of rural population moves to

urban sector—no eVect on sector

distributions (Kuznets process)


F 10% of the poorest half of the rural

population move to bottom half of the

urban sector

Gini-coeYcient falls 3.8%

G Overall eVect of move from

10% urban/90% rural to 40%

urban/60% rural (Kuznets process)



However, the Kuznets process as a whole, the shift from a predomin-

antly rural society to one in which the urban sector is substantially

increased, will increase overall inequality, as scenario G shows. But the

eVect only accounts for around a third of the increase in inequality

which actually occurred between the 1820s and the 1890s. Other factors

which might have contributed are simulated in scenarios A and B: a

shift in the terms of trade which has an adverse eVect on agriculture

raises inequality, but the eVect is greater if it is strongest on those at the

bottom of the rural distribution (scenario B).

In general, the strength of the rise in inequality shown by the analysis

of the Prussian income tax data is such that it was almost certainly

brought about by a combination of adverse factors. These were:

1. a shift of population from a more equal rural sector to a less equal

urban sector (Kuznets process);

2. increased inequality in the rural sector (due to a rise in the

numbers of landless labourers);

3. increased inequality in the urban sector (due to the operation of

the Lewis Model);

4. a shift of the terms of trade against agriculture.

(iii) Econometric analysis

The Prussian tax statistics can also be used to examine some of the

causes of inequality in the rural Kreise. In 1916 an analysis of tax

returns in 1914 at the Kreis level was published in the Zeitschrift of the

Prussian Statistical Bureau. The assessments were based on income in

the previous year, so this is an analysis of the Prussian income distribu-

tion in the Wnal year before the war.55

The data gives the numbers of taxpayers, and their dependants in

four categories: those with annual incomes below 900 Marks, those with

between 900 and 3,000 Marks, those with between 3,000 and 9,500

Marks, and those with over 9,500 Marks. This made it possible to

produce estimates of inequality at the Kreis level, and these could then

be subjected to statistical analysis. The results of the overall analysis are

summarized in Table 9.5.

The analysis showed that diVerent-sized cities were associated with

diVerent levels of inequality in nearby regions. Larger cities tended to

55 There is a full report on the sources used and the results of the econometric analysisin Grant (2002b).


have high levels of inequality, and this appears to spill over into nearby

regions. But the eVect of smaller towns was more beneWcial: inequality

was lower in regions which were close to these. Kreise which were

remote from these smaller towns tended to have high inequality.

Inequality was lower in the more agricultural Kreise, conWrming the

sectoral diVerences shown in Table 9.2. There was also a pronounced

east–west division: inequality tended to fall as the distance from Rus-

sian Poland increased. However, in the rural areas the land distribution

was a major inXuence on overall inequality, so it may have been the

unequal land distribution which was causing inequality to be high in

the rural east. This can be demonstrated by considering the rural Kreise

separately; an analysis of this type is summarized in Table 9.6.

This conWrmed some of the points made in Table 9.5: inequality was

high in areas close to big cities, and in those which were remote from all

urban areas. But the inXuence of Russian Poland was considerably

weakened, and may well be insigniWcant in statistical terms. While the

earlier table might be interpreted as showing that inequality was

‘imported’—the availability of cheap labour from Russian Poland was

pushing up German inequality—it now appears that inequality in the

east was to a large extent ‘home-grown’. It was the unequal land distri-

bution in the east which caused inequality to be high in these areas.

Turning to the major cities, Table 9.7 reports the results of the analy-

sis of inequality in these urban areas. There was no tendency for in-

equality to be higher in cities further to the east, which conWrms that the

Table 9.5. Summary of results of statistical analysis of inequality in

583 Prussian Kreise, 1913/14


Factors which tend to increase inequality:

Proximity of the border with Russian Poland

Remoteness (as measured by the distance from towns and cities with at least

20,000 inhabitants)

Factors which tend to decrease inequality:

High rates of agricultural employment



Proximity of large cities (at least 100,000 inhabitants)

Source: Summary of the results given in table 13 of Grant (2002b).


east–west division was mainly due to diVerences in the land distribution.

However, inequality was higher in cities with high levels of migration,

measured by the total number of moves, in and out, relative to the

resident population. In these cities, with highly Xexible labour markets

and elastic labour supplies, there was a larger gap between rich and poor.

This provides support for the view that the migration of the rural labour

surplus into urban areas was a factor driving up inequality. However,

the main ‘labour surplus’ period was over by 1914. If an eVect can be

Table 9.6. Summary of results of statistical analysis of inequality

in 478 rural Prussian Kreise, 1913/14



Remoteness (as measured by the distance from towns and cities with at least

20,000 inhabitants)



High rates of agricultural employment

Proximity of large cities (at least 100,000 inhabitants)



Proximity of the border with Russian Poland

Presence of medium-sized farms (20 to 100 hectares)

Table 9.7. Summary of results of statistical analysis of inequality

in 44 Prussian cities, 1913/14


Factors which increase inequality:

High levels of migration (moves both in and out)



Presence of textile, mining and printing industries

Presence of utilities


Presence of leather industry


found for this date, it suggests that the impact would have been much

greater in the 1870–95 period, when the bulk of the transfer took place.

The industrial structure of the cities was found to have an eVect on

inequality, although the statistical analysis was not entirely satisfactory.

Industries which were associated with high inequality included textiles

and mining, which may reXect the larger scale of operation of these

industries, but also printing, which was not organized on a large scale.

The leather industry, which was relatively unmechanized and organized

in small workshops, was associated with low levels of inequality.

8. Could the R i se of Inequality have been Avoided?

The results produced so far are in line with the conclusion reached by

some modern cross-section studies: there is a conditional Kuznets

Curve, which can emerge when the historical circumstances are right

but which, under diVerent conditions, might be obscured by other

factors. The question then is, what were the forces that produced the

Wrst stage of the Kuznets Curve in nineteenth-century Prussia? Could

the increase of inequality have been prevented?

The Wrst point is that, even at the peak, inequality was not particularly

high in Prussia by modern standards. Although income inequality in the

urban sector was quite high, the overall Wgure was held down by a relatively

egalitarian rural sector. What created the Kuznets Curve in Prussia was the

shift from a rural society with a low Gini-coeYcient to an urban or indus-

trial sector which had a level of inequality which was high by comparison

with the rural sector though not so unusual when compared to other coun-

tries going through the process of industrialization.

The second point is that, from the 1870s onwards, there was a shift

in the terms of trade against the agricultural sector as a result of devel-

opments on world markets. This had the eVect of widening the gap

between rural and urban incomes, and thus worsening the overall

income distribution. Although this was not the only cause of the rise in

inequality in the period, it contributed to the rise in the Gini-

coeYcient. This was due to the changing terms of trade on world

markets. Only a massive rise in agricultural protection could have kept

this from aVecting German agriculture.

A third point is that, in the early part of the century, inequality was

rising in the rural sector, due to a range of factors: demographic pres-

sure, expansion into remote eastern areas, the eVects of the agrarian


reforms, and changing social relations. This was only indirectly con-

nected to industrialization.

These factors provided a considerable reinforcement to the mechan-

ism speciWed by Kuznets himself. It would be quite possible for a

society which industrialized under diVerent circumstances to escape the

Kuznets Curve. The Prussian example points to some factors which

might help to achieve this: such as a relatively low level of demographic

pressure, a prosperous agricultural sector in which incomes kept pace

with those in urban areas, agrarian policies which promoted an equal

distribution of income in this sector, and better credit provision for

small farmers. These are the ‘good’ ways to escape a Kuznets Curve.

But the comparison with Britain also reveals that there is a ‘bad’ way to

escape it, and that is to start oV with such a high level of inequality in

the rural sector that the transition to an industrial society does not push

up overall inequality even though incomes in the urban sector are also

unequally distributed.

In short, the Kuznets Curve, while not just a peculiarity of German

history, owes its origins in Prussia to some features of nineteenth-

century Prussia which may or may not be replicated in modern

developing countries. It is the product of circumstances rather than an

automatic consequence of industrialization. But there was not much

that government action in nineteenth-century circumstances could have

done about this. A period of rising inequality was probably something

that just had to be endured.

Appendix 9A. Est imates of income inequality from

Pruss ian tax statist ics

Table 9A.1. Estimates of income inequality in Prussia 1822–46

1822 1826 1831 1836 1841 1846

Gini-coeYcient

as indices (1846¼100)

0.260 0.266 0.274 0.278 0.286 0.297

Gini-coeYcient 87.6 89.4 92.3 93.5 96.4 100.0

Upper 95% CI of

Gini-coeYcient

96.5 98.5 101.7 103.1 106.2 110.2

Lower 95% CI of

Gini-coeYcient

78.7 80.3 82.9 84.0 86.6 89.8



1853 1855 1858 1861 1864 1867 1870 1873

Gini-coeYcient


0.345 0.348 0.352 0.353 0.354 0.355 0.357 0.363

Gini-coeYcient 100.0 100.9 102.1 102.3 102.6 102.8 103.4 105.2

Upper 95% CI of Gini-coeYcient 111.7 112.7 114.1 114.3 114.6 114.9 115.5 117.5

Lower 95% CI of Gini-coeYcient 88.3 89.0 90.1 90.3 90.5 90.7 91.2 92.8


1873 1874 1875

U10/L40 1.44 1.51 1.47

Gini-coeYcient as indices with 1853¼100 0.315 0.325 0.318

Gini-coeYcient 105.2 108.5 106.2


1875 1876 1878 1880 1882 1883 1885 1887 1889 1891

Gini-coeYcient


0.340 0.341 0.357 0.371 0.373 0.370 0.374 0.374 0.379 0.382

Gini-coeYcient 107.1 107.6 112.5 116.8 117.7 116.7 117.9 118.0 119.5 120.5

Upper 95% CI of Gini-coeYcient 112.1 112.7 117.8 122.3 123.3 122.2 123.5 123.6 125.1 126.1

Lower 95% CI of Gini-coeYcient 102.1 102.5 107.2 111.3 112.2 111.2 112.4 112.5 113.8 114.8

328Industrialization

,Migration,

Inequality


1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902

Gini-coeYcient


0.321 0.324 0.324 0.323 0.315 0.318 0.324 0.327 0.331 0.333 0.334

Gini-coeYcient 119.2 120.1 120.1 119.8 117.0 118.1 120.0 121.4 122.9 123.6 124.1

Upper 95% CI 123.9 124.8 124.8 124.6 121.6 122.8 124.8 126.2 127.8 128.6 129.0

Lower 95% CI 114.5 115.3 115.3 115.1 112.4 113.4 115.3 116.5 118.0 118.7 119.1

1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914

Gini-coeYcient


0.331 0.332 0.335 0.340 0.338 0.334 0.333 0.319 0.316 0.315 0.318 0.329

Gini-coeYcient 122.6 123.1 124.3 126.0 125.5 124.0 123.5 118.2 117.3 117.0 117.9 122.0

Upper 95% CI 127.5 128.0 129.3 131.1 130.4 128.9 128.4 122.9 122.0 121.6 122.6 126.8

Lower 95% CI 117.7 118.2 119.4 121.0 120.5 119.1 118.6 113.5 112.6 112.3 113.3 117.1

Table 9A.6. Analysis of the estimated changes in each period

1822–46 1853–73 1875–91 1892–1906 1906–14

Change in Gini-coeYcient (%) þ14.2 þ5.2 þ12.5 þ5.7 �3.2

Upper 95% CI of change þ16.2 þ6.0 þ13.3 þ6.0 �3.1

Lower 95% CI of change þ12.2 þ4.3 þ11.6 þ5.5 �3.4

Upper 95% CI of highest variant þ21.4 þ20.2 þ13.6 þ8.3 �5.4

Lower 95% CI of lowest variant þ4.3 þ3.5 þ5.7 �1.3 �2.7

Source: ***Take in JJ From fo. 386***

Industrialization,

Migration,

Inequality329

10

Challenging the Kehrite

View of Imperial Germany

1. Introduction

The view that the Kaiserreich suVered from internal Xaws is so deeply

entrenched in German historiography that one doubts if it will ever be

dispelled, however much evidence is produced against it.1 The list of

Xaws has changed over time: concepts such as the lack of a middle-class

identity, ‘organized capitalism’ as a distinctive economic system, ‘social

imperialism’, the importance of the ‘failed revolution’ of 1848 have

been given up or downgraded. The term Sonderweg is used less. But

new lists are produced, which stress diVerent aspects of the social order,

internal structure, political system, and cultural outlook of Imperial

Germany: the importance of organized economic pressure groups, the

impotence of parliamentary control, the appeal of Social Darwinism,

the drift to extremist nationalism, and so on.

This study has not attempted to confront these arguments directly;

instead it has put forward an alternative view, taking many of its cues

from the comparative study of developing countries in general, and

from development economics in particular. The argument in its essence

is that the Kaiserreich suVered from no more than the normal problems

of a rapidly industrializing society.

This is not to argue that there were no special features and character-

istics of German society in the period. There were indeed unusual and

unique attributes, as there are in all societies. But as Richard Evans has

1 Even historians specializing in foreign policy have accepted much of the ‘Xawednation’ thesis: see Hildebrand (1995), 5–6 and 9. Hildebrand also accepts the interlockingof domestic and foreign policy, Hildebrand (1989), 77. The view that a more realisticforeign policy, involving an alliance with either Britain or Russia, was blocked for internalreasons is set out in Geiss (1976), 116, and Kehr is cited in support of this. Calleo (1978)provides a more sympathetic view of the ‘German problem’, and argues that GermanuniWcation should not be seen as ‘a malevolent accident which befell an otherwise harmo-nious European state system’, ibid. 3, but then goes on to accept the now discreditedargument about the connection between ‘organized capitalism’ and German colonialexpansion, put forward in Wehler (1969).

pointed out, commenting on the evolution of the Sonderweg thesis in

the works of Hans-Ulrich Wehler, the list of Xaws which Wehler has

put forward in recent work as an explanation of the particular course of

German history contains nothing that cannot be found in some degree

in other European states in the late nineteenth and early twentieth

centuries.2

And, it might be asked, why is the comparison limited to Europe in

this period? Industrialization is no longer a purely European phenom-

enon. It is an experience through which a number of non-European

societies have passed or are passing. And in virtually every case it has

been a disruptive and dislocating process.

Evans also commented that the Sonderweg argument has become one

which is increasingly concentrated on the failure to make political pro-

gress commensurate with the pace of social and economic transform-

ation. This is a point which is widely accepted even by historians who

are not associated with the Sonderweg view. It is one of the common-

places of Wilhelmine historiography.3

But a broader comparison of societies, inside and outside Europe,

going through the initial stages of industrialization reveals that few if

any have made political progress during this period. Even in Britain,

the earliest phase of industrial growth, usually dated 1780 to 1830, was

not a period which saw much in the way of progressive democratic

politics. In many other countries, there have been declines into civil

war, authoritarianism, or military government.

When this broader context is considered, the insistence that Imperial

Germany should have made political progress commensurate with its

economic advance can appear rather bizarre.4

2. Social , Economic , and Political

Trajectories in Imperial Germany

In the history of the Kaiserreich the results of the 1912 election have a

particular signiWcance: the Social Democratic Party gained 110 seats,

2 In a review of Wehler’s Deutsche Gesellschaftsgeschichte, reprinted in Evans (1997).3 Theodor Schieder was one example of a historian of a diVerent persuasion who made

this point, Schieder (1960), 85.4 The view that industrialization and progress towards democracy are linked is a com-

ponent of ‘modernization’ theory, and this is often regarded as a typically Europeanphenomenon; see Eley (1996) for comments on this from a German perspective.

Challenging the Kehrite View 331

reversing the losses made in the ‘Hottentot’ election of 1907, and

exceeding the previous high-water mark of 81 seats achieved in 1903. It

was now the largest party in the Reichstag; if it could gain the support

of the left liberals and the parties of the national minorities (Poles,

Danes, and Alsatians) it would form a ‘disaVected’ block of 185 out of a

Reichstag of 397 deputies.

This had a number of important consequences. On the right it pro-

duced a sense of marginalization, which was intensiWed when the gov-

ernment made concessions to the SPD, including the introduction of

direct taxation at the Reich level, in order to secure the joint passage of

the 1913 Finance Bill and the Army Bill of that year. Kehrite historians

such as Dirk Stegmann have argued that the radicalization of the right,

and the consequent abandonment of Sammlungspolitik, were important

factors in the intensiWed foreign policy crisis which culminated in the

events of July 1914.5

More broadly, the results of the election called into question the

quasi-democratic system of constitutional checks and balances which

maintained the Kaiserreich’s uneasy compromise between diVerent

social and regional groups. A Reichstag which possessed legislative

powers but which did not control the executive had been a successful

solution to many of the problems faced in 1871: how to provide repre-

sentation for others in an empire dominated by Prussia; how to secure

local autonomy from encroachments by a central executive; how to

reconcile previously dominant groups (the landed aristocracy, the urban

Burgertum) to the experiment of universal male suVrage.6 But the con-

tinuation of this compromise required that all participants accepted the

implied limitations, and this was particularly diYcult for the Social

Democrats, because they were largely excluded from participation in

the political process at the level of local and regional government.

Any compromise which would have a reasonable prospect of securing

Social Democratic support over the long term, rather than as a short-

term tactical expedient (as in 1913), would have had to have included

reform at the local and regional level as well as the strengthening of

5 Stegmann (1970), 257–93; according to this, the election led to a renewed imperialistagitation, ibid. 277. The ‘shameless, chronic persecution’ of Bethmann-Hollweg by thesegroups had, it is argued, an eVect on his policy in July 1914, Chickering (1984), 289.

6 The Bismarckian constitution has attracted much criticism from historians: ‘mon-strous’ and ‘irregular’ are two adjectives that have been used, Sauer (1968), 413. But thisunderestimates the problems it had to resolve. For an alternative view, see Rauh (1973),47 V.

332 Challenging the Kehrite View

Reichstag authority.7 There are many historians who would argue that

such a compromise was not acceptable to powerful interest groups, and

that a military coup, or Staatsstreich, would have occurred had not the

outbreak of war intervened to make this unnecessary.8 The war was

welcomed by such groups as a means of halting the progress towards

democracy.

Thus, in the Kehrite scheme, the election has a special signiWcance: itis the point where sharpening internal divisions directly impinged on

Germany’s strategic position, making the need for a solution to the

problems of Germany’s diplomatic isolation into a pressing imperative.

It also made it clear that any internal solution would have to involve

major concessions to a party with a revolutionary tradition and regarded

by its conservative opponents as disloyal.9

Against this, there is the more favourable picture presented by the

economic trends discussed in the previous chapters. Combining these

trends it is possible to construct, as a mental picture, a curve of the

intensity of the socio-economic problems facing Imperial Germany,

meaning by this both purely economic problems, and social problems

having economic roots. The trajectory of this curve shows a rising trend

to 1895–1900 and then declines. But this was not the only set of prob-

lems which beset the Kaiserreich. There were diVerent sets involving

socio-cultural issues (dissatisfaction with or alienation from social and

cultural structures and trends) and socio-political questions, deWningthese to include purely political problems and social divisions which

manifest themselves within the operation of the political system.

Voting behaviour is one symptom of the latter, though the movement

of opinion within parties is another aspect which is also important

although less easy to measure. If the shares of the diVerent parties are

plotted (Figure 10.1) this shows the steady gains made by the Social

Democrats from the late 1880s onwards and the way that this squeezed

7 For a discussion of the tactical nature of the SPD’s support of the government in1913, see Groh (1973), 434–6. The background to the vote is analysed in Stargardt (1994),108–24.

8 The danger of a Staatsstreich was probably not that great, Pogge von Strandmann andGeist (1965).

9 The extent to which the SDP was ever a revolutionary party is debatable. Marxistideas certainly inXuenced the leadership, but had less appeal at lower levels, Lidtke (1985).Its ‘disloyalty’ would seem to have been at least partly disproved by its vote for the 1913Army Bill. Despite this, Epkenhans (1991) has argued that there was a fear that a politicalcompromise would lead to a ‘storm of disarmament’, ibid. 413, and that this, together withthe internal political situation, led to the Flucht nach Vorn of July 1914.


other parties. However, it would be wrong to conclude from this that

there was a radicalization of internal politics in the Wilhelmine period.

The internal politics of the Social Democratic Party have also to be

considered. These were moving in a more centrist direction, and the

party was making a determined eVort to broaden its support and move

outside its natural constituency of the urban working class. This was

rewarded by a gain in electoral support.

The area of German politics which was shifting in a more radical

direction was the conservative right. But this was losing ground. The

best result for the right-wing parties came in the 1893 election, when

the combined right-wing vote was 16.9%, but this had fallen to 9.6%

by 1912. There was a fall in seats gained from 88 to 46. The shift to

more radical policies was not beneWcial in electoral terms.

In short, the pattern of electoral politics in Imperial Germany was

not one of increasing polarization. Parties which adopted more moder-

ate policies gained votes; parties which shifted towards the extremes

lost votes. This was exactly as would be expected in a normal, stable,

Minoritiesand others

Conservativesand other right

Nationalliberals

Freeconservatives

Lift liberals

Centre

SPD18710

20

40

60

80

100

1874 1877 1878 1881 1884 1887 1890 1893 1898 1903 19121907

Fig. 10.1. Percentage shares of different parties in Reichstag elections,

1871–1912


democratic system. But the gains made by the SPD meant that the

party had to be admitted into the Wilhelmine political establishment:

the option of continuing to ignore it no longer existed.

3. The Dog that D idn ’t Bark :

Economic Issues and the 1912 Election

The SDP’s ability to break out from its core constituency depended on

its success in appealing to artisans, tradesmen, workers in small work-

shops in the urban areas, and in attracting votes in smaller towns which

were not major industrial centres.10 For these voters, the importance of

the economic issues stressed in earlier party policy, such as public

ownership of major enterprises, was at best peripheral.

It is worth considering quite how small the Social Democrats’ core

constituency was. In 1910 the urban population, those living in major

towns and cities (91 cities with at least 50,000 inhabitants), was 17million out of a total German population of 64 million. A survey of

employment in these cities showed a total of 1.6 million men (and 0.5

million women), aged 16 and above, employed in enterprises with at

least 10 employees. Bearing in mind that some of these would have

been too young to vote, and also that 29% of the urban population were

Catholics, this suggests that the Social Democratic core, urban Protest-

ant working-class men, was between 1.2 and 1.3 million voters. But the

SDP vote was well above this level. It totalled 4.25 million in 1912. Itwas 1.78 million in 1893 (when the size of the core was considerably

smaller).

Figure 10.2 contains a comparison of the SDP share of the vote with

the percentage of the German population living in urban areas. This

illustrates the way that structural change in German society was

favouring the party’s electoral advance. It also suggests that the party

broke out of its core constituency relatively early, in the 1890s. In the

early period the party’s share of the vote was roughly in line with

the share of the German population living in cities of at least 100,000:

the major industrial centres where large-scale enterprise was common.

But from 1890 onwards the party’s share moves with the percentage of

the population living in towns and cities with at least 20,000 inhabit-

ants, many of which were not industrial centres at all.

10 Sperber (1997), 64, estimates that the party had 45% of the Protestant middle-classvote, and 13% of the Catholic middle-class vote in 1912.


The graph shows the deviations from trend caused by the anti-socialist

laws of the 1880s, and it also shows the exceptional nature of the 1907

election: the 1912 election represented a recovery of the relationship

previously established. It is also worth noting that the proportion of the

German population living in towns with at least 20,000 inhabitants was

rising by 6 percentage points each decade. If the SDP could hold onto

the electoral share implied by this trend, its vote would rise proportion-

ately—by over a million votes every ten years.

The ability of the party to address broader social and political issues

made it attractive to a swath of opinion outside the larger cities. It was

also helped by the ineVectiveness of other opposition parties: voters

who wanted signiWcant change in the political structure of Imperial

Germany had no other party to turn to.

The statistical analysis of election results can elucidate some of these

points, though it is necessary to stress that statistical tests can only be

run where there are questions involving measurable issues. It is possible

to look at questions such as ‘Did high unemployment in particular

constituencies cause a rise in the socialist vote?’; others, such as ‘Did

the desire for constitutional change lead to an increase in the SDP

191119061901189618911886188118761871

40

30

20

10

0

SPD vote

% in cities >20,000

% in cities >100,000

Fig. 10.2. The share of the Social Democratic Party in Reichstag elections

compared with the proportions of the German population living in urban areas,

1871–1912


vote?’, are not so easily answered. Moreover, statistical analysis at the

aggregate level does not provide direct evidence on voter motivation;

this would require surveys of individual voter behaviour, attitudes, and

social background. It shows which factors are associated with higher or

lower probabilities of a vote for a given party. It tests the inXuence of

the general social and economic environment on voting behaviour.11

The data set prepared for this analysis was split in two. The Wrstincluded the major cities and the second looked at results in Prussia

outside these urban areas. This enabled statistical sources giving infor-

mation on economic conditions in the urban areas to be used in the Wrst

data set, and sources on agricultural structure to be included in the

second.

Table 10.1 gives a summary of the statistical analysis of voting be-

haviour in 70 large towns and cities (the full results are given in Table

10B.1 in Appendix 10B). The analysis Wrst demonstrated the import-

ance of city size: the SDP share was higher in the larger cities. The

second important factor was religious aYliation: Catholics were much

less likely to vote for the SDP (the proportion of non-Catholics voting

for the SDP was 2–21⁄2 times that for Catholics).12

Table 10.2 shows the implications of these results for cities of diVer-

ent sizes.13 The importance of city size is clearly demonstrated. The

growth of cities tended to favour the SPD. Migration patterns which

caused cities to expand were therefore a factor in the party’s advance.

Catholicism remained an obstacle even in the larger cities. However,

when Prussian cities are separated from cities outside Prussia an im-

portant diVerence is revealed: the eVect of Catholicism was considerably

stronger inside Prussia. In Prussian cities the Catholic population was

much less likely to vote SPD. In a city of 500,000, non-Catholics would

be three times as likely to vote SPD as Catholics. But outside Prussia

11 To give an example of this: only a survey of individual voters could show whetherunemployed workers were more likely to vote for the SPD. A statistical association be-tween high unemployment in a given constituency and a high SPD vote might show this,but it might also show that employed workers in a constituency with high unemploymentwere more likely to vote SPD. The two eVects cannot be distinguished. All that can beshown is a general result that unemployment inXuences voting behaviour.

12 The SPD was gaining ground with the Catholic working class, though its share wasonly 22% against 54% for Protestant workers, according to estimates by Sperber (1997),37–8. Contrary to the view put forward by Schmadeke (1995), 183–91, Catholicism wasnot an insuperable obstacle to the SPD. Schmadeke’s conclusion, that the SPD’s rise wascoming to an end, does not appear to be justiWed by the evidence.

13 The coeYcients used are those given in column a of Appendix Table 10B.1.


this was less important. In a similar-sized non-Prussian city the ratio

between non-Catholics and Catholics would only be 1.3–1.14

It might be thought that this was a reXection of the importance of the

Polish-speaking Catholic population in Prussian cities. But inspection

14 This conWrms a point made by historians who have looked at regional electionresults. The SPD was a strong contender for the Catholic working-class vote in non-Prussian areas. Blackbourn (1980), 193, comments on the strength of the party inworking-class Catholic areas in Baden and Wurttemberg. The SDP had 28.3% of the votein Baden in 1912 (Baden was 59.3% Catholic) but only 9.8% in Regierungsbezirk Koblenz(65.5% Catholic) and 4.6% in Regierungsbezirk Trier (79.3% Catholic). Figures fromRitter (1980).

Table 10.1. Summary of results of statistical analysis of the 1912

SPD vote in 70 major German cities


Factors which were associated with a higher SPD vote:

Size of the city (larger cities had higher SPD shares)

A large SDP vote in 1878

Factors which were associated with a lower SPD vote:

Catholicism

Catholicism within Prussia (additional eVect)

Factors which had little or no inXuence on voting behaviour:

Levels of inequality (measured by the U10/L40 ratio)

Local labour market conditions (the ratio of applicants to vacancies)


Factors which were associated with a higher SPD vote:

High wages

High mobility

Large-scale industry (large factory size)

Table 10.2. Predicted SPD share of the vote for cities of different

sizes and different religious affiliations

Number of inhabitants in the city

50,000 100,000 500,000 1,000,000

City population is:

80% Catholic 20.5 24.6 35.7 41.1

20% Catholic 37.8 43.4 56.7 62.2


of the data reveals that this was not the case. The socialist vote was also

unusually low in Rhineland cities like Aachen and Trier which had few

Poles. Rather it seems to reXect the strength of Rhineland Catholicism,

the social cohesion of a minority in a predominantly Protestant soci-

ety.15 The after-eVects of the Kulturkampf almost certainly played a

part as well.

There was a strong element of persistence in the socialist vote: the

cities which had a relatively large vote in 1877 tended also to have high

votes in 1912. Table 10.3 gives the Wgures for the two elections for the

ten cities with the highest votes in 1912. The SPD share was already

above 40% in six out of the ten cities. Much of the growth in the

party’s overall share came from the spread of socialist ideas from these

core cities, initially to other large cities, and then to smaller towns and

other areas.16

15 Another possible reason, suggested by Rohe (1982), 236, is that in Prussian areassuch as the Ruhr there was a confessional aspect to employment relations: typically,Catholic workers were employed by Protestant entrepreneurs. This meant that, in theseareas, Catholicism developed a ‘social’ dimension which was less evident in non-PrussianCatholic regions, where workers and employers were both likely to be Catholics. See alsoEvans (1997) and Sperber (1983) for the consequences of the Kulturkampf.

16 Hamburg and Saxony were strongly represented at the original meeting of theAllgemeiner Deutscher Arbeiterverein in 1863, Berger (2000), 49. See also the maps of SDPsupport in Schmadeke (1995) which show the strength of the party in areas aroundSaxony, particularly Thuringia and Franconia. There was a geographical element to thediVusion of SPD support.

Table 10.3. Cities with the highest SPD vote in 1912; comparison

with 1877 percentage share of votes cast (first-round votes)

SPD vote 1912 SPD vote 1877

Berlin 75.0 39.2

Chemnitz 64.3 54.9

Altona 63.4 51.0

Hamburg 61.2 40.0

Nuremberg 60.7 48.3

Zwickau 60.6 61.4

Charlottenburg 58.7 16.2

Mulhausen 56.1 �Brunswick 55.6 46.1

Solingen 55.6 35.8


The analysis also showed that there was only a limited eVect on the

socialist vote from economic conditions within the cities. There was

some evidence that the socialist vote was higher in the more prosperous

cities (where wages were higher), and in cities with large-scale industry,

measured by factory size, though the connections were not strong ones.

This in turn reXected the success of the party in widening its appeal

beyond the conWnes of its core constituency: workers in large-scale

industry.17 The presence of this core constituency is just discernible in

the results, but the eVect is only a slight one. Another weak Wnding (in

statistical terms) is the existence of a relationship between high mobil-

ity, the total number of moves in and out of each city in 1910–12

relative to population, and the size of the SPD vote. The socialist vote

rose in cities with more Xuid labour markets, but the result is not a

strong or a secure one.

Two factors which were shown to have virtually no eVect on voting

behaviour were labour market conditions and inequality. The Wrst was

measured by the ratio of applicants to vacancies at town labour

exchanges. Where economic conditions were adverse, the ratio of appli-

cants to vacancies would be high. This was not a factor which had any

discernible impact on the SPD vote. Inequality was measured by the

ratio between the estimated share of the top 10% of households in total

income and the estimated share of the bottom 40%. In the more

unequal towns, the top 10% had a share which was Wve or six times as

large as that of the bottom 40%; in the more egalitarian towns the

shares were about equal. But these diVerences had no signiWcant eVect

on votes cast in the 1912 election.18

The analysis of the vote in rural or small-town Prussian constituen-

cies is summarized in Table 10.4 (full results are in Table 10B.2). The

SPD share in these constituencies averaged 20.6%, which was rather

less than in the towns analysed earlier (42.2% on average). However,

the SPD advance in these small town and rural constituencies was an

important component of the party’s overall gains. The average share

had risen from 16.3% in 1907.

17 The SPD no longer represented a ‘homogeneous group of proletarians’, Fairbairn(1997), 216.

18 The methods used to produce these estimates of the income distributions in thePrussian Kreise are described in Grant (2002b). Eley (2002), 56–7, comments on the lackof a relationship between economic factors and the SPD vote in a comparison ofRemscheid and Hamborn: Hamborn workers were the ‘very epitome of the brutalised andexploited factory proletariat’ but the SPD was weaker there than in Remscheid.


The analysis shows that there were two factors which explained

much of the variation in the SPD vote: the percentage of the population

occupied outside agriculture and the Catholic percentage. The agricul-

tural population was less likely to vote for the SPD, and the relative

decline of agriculture was a factor working in the SPD’s favour in these

rural constituencies.19 Catholicism was an important factor working

against the SPD especially in these Prussian areas. In the Polish-speak-

ing regions the SPD vote was held down even further, although it was

higher in constituencies where Lithuanians were present.

The SDP vote was highest in areas near to the larger cities (with over

100,000 inhabitants) but it was also high in areas near to towns with a

population of at least 20,000. But the party’s share was reduced in the

19 This result is also found in the analysis by Fairbairn (1997), 228. Fairbairn alsoshows that the negative eVect from agricultural employment became stronger between1890 and 1912, ibid. 230. As pointed out by Steinbach, electoral research in Germanhistory is ‘relatively immature’ and the type of multi-variable statistical analysis of elect-oral results presented in this chapter is rare for pre-1914 Germany, Steinbach (1992), 121.Fairbairn’s analysis is an exception. Suval (1985), 122, produces correlations which areinteresting but does not use multivariate analysis. Nor does Schmadeke (1995). Winkler(1993), 251 does not give t-statistics, so there is no way of assessing the quality of theresults shown.

Table 10.4. Summary of results of statistical analysis of the 1912 SPD

vote in 204 constituencies outside the major cities (Prussian

constituencies only)


I. Factors which were associated with a higher SPD vote:

A high level of non-agricultural employment

Large numbers of Gutsbezirke relative to other Gemeinden

A high Lithuanian-speaking population

II. Factors which were associated with a lower SPD vote:

Catholicism

A high Polish-speaking population


Remoteness from major cities (over 100,000 inhabitants) and from towns

(over 20,000 inhabitants)

III. Factors which had little or no influence on voting behaviour:

Levels of inequality (measured by the U10/L40 ratio)

Wage rates


more remote constituencies, and this was an eVect which was independ-

ent of the occupational structure of these constituencies. Towns and

large cities were the centres of socialist activity, and this had an eVect

on nearby areas.

Economic variables had little eVect apart from the occupational struc-

ture. Neither wage levels nor inequality (as measured by income in-

equality) had any eVect on voting behaviour. This repeats the Wndingreported earlier from the analysis of votes in urban constituencies: eco-

nomic issues do not appear to have been important factors in the 1912

election. Inequality in land ownership, however, as measured by the

percentage of the agricultural area which was in large farms and estates

(over 100 hectares), was a factor working against the SPD. Traditions of

deference may have kept the socialist vote down, or fears of retaliation if

the secrecy of the ballot was not thought to be absolutely guaranteed.20

However, a factor which operated in the opposite direction was the

presence of Gutsbezirke. The SPD vote was higher in constituencies

where a large percentage of the local Gemeinden were classiWed as Guts-

bezirke: aristocratic estates which had in the past had an important

administrative function. The other categories were Landgemeinden and

Stadtgemeinden. Around 25% of the administrative units in these con-

stituencies were Gutsbezirke, though this exaggerates their importance

as they were usually much smaller than the Landgemeinden.

In the past the aristocratic owners of the Guter had had considerable

control over the local police and administration, and judicial functions

of their own. These were largely abolished by the Kreisordnung of

1872.21 What remained was an important position in local society, and

some Wscal privileges, notably an exemption from the supplement

charged on the Prussian income tax assessment to Wnance the expend-

iture of the Landgemeinden and Stadtgemeinden. Against this, the owners

of the Guter had to provide some services which would otherwise have

been paid out of local government funds.22

20 Anderson (2000), 157–9, argues that the secrecy of the ballot was not assured insome rural areas.

21 The judicial role of the Ritterguter was removed in 1849–51; the police functionswere abolished in 1872, von Below (1891), 675 V. However, the Ritterguter remainedseparate administrative districts until 1927, Puhle (1986), 85, and some historians haveargued that much of their authority and inXuence survived the 1872 Kreisordnung, Carsten(1989), 120.

22 The supplement could be a substantial one, an increase of 50% or more in 663Gemeinden according to Meisel (1911), 298. The aristocratic owners of the Guter alsobeneWted from favourable tax treatment from the Landrate (district administrators),according to Witt (1985). This would have been a further cause of resentment.


Not all large farms and estates were Guter: there were many estates

that had been sold by their original owners or that had never the func-

tions of a Gut. And not all Guter were large agricultural holdings: some

estates had been broken up or rented out while still retaining the title

and privileges of a Gut. In places the system was a relic with little

practical relevance. What these results indicate is that this system still

caused resentment.

In general, economic factors (income inequality, wage levels, un-

employment, and enterprise size) did not have much eVect on the SPD

vote. The party was appealing to a broader constituency, concerned

with other issues. But political inequality, the sense of being excluded

from decision-making, whether at the local or the national level, did

matter. And this is the main signiWcance of the result with the Gutsbe-

zirke variable.23

It is not the case that the Gutsbezirke were a particularly onerous part

of the political and administrative system of Prussia in 1912. There

were other more important concerns for electors willing to vote for

radical change. Their presence happens to be something which can be

measured, and which varied considerably between diVerent constituen-

cies. Other issues, such as the three-class electoral system, aVected all

constituencies, so there is no variation in the importance of the variable

which can be picked up by statistical analysis. These results indicate

that there was a desire for change in the political and administrative

system, and that this desire for change was an important factor in the

SDP vote. Voters who wanted progress towards democracy turned to

the SDP as the party most likely to achieve this.24

The SPD responded to this new role. Its policy statements and leaX-

ets emphasized constitutional issues, and it signalled an increased will-

ingness to play a constructive role in the political process.25 Moreover,

and more dramatically, there was a shift in the social make-up of the

23 The question of political ‘fairness’, the sense of being excluded by the suVragesystem, was one of the central issues of Wilhelmine politics, Fairbairn (1992).

24 Another way to protest against the three-class electoral system was to abstain in localelections. The turnout in Landtag elections could fall as low as 8% for electors in the thirdclass, Anderson (2000), 196. In some rural areas the left-liberals also beneWted fromdissatisfaction with the electoral system, Thompson (2000), 301. The electoral system alsokept many German city administrations in liberal hands, despite SPD majorities, and itmay be surmised that resentment of this was a factor in the strength of the SPD vote inurban areas, Pogge von Strandmann (1992).

25 Bertram (1964), 169; this trend towards an emphasis on political objectives was inpart a reaction to the SDP’s setback in the 1907 election: Crothers (1941), 213 and 229;Schorske (1955), 225–6.


SDP’s Reichstag candidates. In 1907 the party had Welded 132 working-

class candidates (Arbeiter, Handwerker, and Angestellte); by 1912 this

had fallen to 57. By contrast, the numbers of private sector managers

and oYcials (Privatbeamte) had risen from 50 to 175.26 Table 10.5 gives

the social breakdown of the SPD candidates. Private sector managers

and businessmen constituted a majority of the candidates oVered in

1912. The other striking feature of the table is the under-representation

of anyone in public sector employment: oYcials (OVentliche Beamte,

Justizbeamte), army oYcers, and teachers. These were just 1.5% of

26 The size of the shift raises the question whether it represented a genuine change, orwhether candidates previously identiWed as Angestellte or even Arbeiter were now classify-ing themselves as Privatbeamte. If so, it still represents a shift in the party’s approach:candidates were stressing their middle-class aYliations instead of seeking to appeal to anarrow working-class constituency.

Table 10.5. Social origins of the SDP candidates, 1907 and 1912

elections, compared to those of other parties (as percentages)

SPD

candidates

Candidates of

other parties

1907 1912 1907 1912

Entrepreneurs,

businessmen:

Agriculture

Commerce and industry

0.8

16.6

1.5

10.0

21.4

11.1

19.5

10.4

Managers, officials: Public sector 3.0 1.5 10.9 9.1

Legal 0.0 0.0 12.3 8.0

Private sector 12.6 43.9 1.6 4.9

Religious: Protestants 0.3 0.0 1.4 3.3

Catholics 0.0 0.0 4.7 3.6

Professional: Army officers 0.0 0.0 1.2 1.8

Lawyers 2.8 3.5 7.1 10.0

Doctors 0.5 0.5 2.3 2.7

Writers 23.1 23.3 9.5 9.5

Teachers 0.3 0.0 6.3 9.0

Employees: Workers 33.2 14.3 5.3 5.6

Others: 7.0 1.5 4.8 2.6

Total number of candidates 398 399 1285 1153

Source: Calculated from figures in Bertram (1964), 158.


SPD candidates in 1912, but 28.6% of those put forward by other

parties.27

The composition of the SPD’s candidates reXected the shift in its

policy orientations: it represented the politically excluded middle classes

as well as the working class. The statistical analysis presented in this

section conWrms that economic issues, which would have been major

concerns of working-class electors, played a relatively unimportant role

in the explanation of the size of the SDP vote. The party was appealing

to a broader constituency: those who wanted political change as well as

those who wanted radical alterations in the structure of society.

4. Economics and the T irp itz Plan :

The Forces Driv ing Germany towards War

The weakness of the political structures of the Bismarckian Reich had

left voters who wanted political change with little alternative but to vote

for a party which, in its original policy statements, had stood for a far-

reaching programme of revolutionary social and economic change. The

rapprochement between the aims of the SDP leadership and those of its

newer and wider base was a gradual one, as the party shifted its em-

phasis from economic issues to the pursuit of political reform, and

began to indicate its willingness to work within the system. However,

the growing strength of the party and the desire for change which it

represented presented a serious problem for the Bismarckian system.

Could the system evolve in a democratic direction without a violent

upheaval?

It is an open question whether this could have been accomplished if

the only issues facing pre-1914 Germany had been internal ones. Could

concessions have been made which would have strengthened parliamen-

tary control without conceding full majority rule? How would the

strengthened Reichstag authority have been regarded by the non-

Prussian states? How far were the conservative forces opposed to

further democratization prepared to go to obstruct such a settlement?

Would the radicals within the SPD have accepted a compromise?

But this is all speculation. For the issues of democratic control and

accountability which were raised by the trajectory of electoral results

27 It is likely that public sector employees would have been reluctant candidates,fearing discrimination against anyone who put themselves forward as candidates for an‘unpatriotic’ party.


meant that the Kaiserreich confronted the problem of constitutional

change at just the same time as the trajectory of Germany’s diplomatic

isolation and weakening strategic position caused a series of crises in

external politics, one of which ended in the outbreak of the First World

War. If there was ever a chance of achieving a normal transition to a

democratic settlement, this disappeared in July 1914.

To consider the background to the deteriorating external position it

is worth returning to one of the arguments put forward in Chapter 8,

when explaining why economic developments had created the condi-

tions for political stabilization post-1900: the favourable prospects for

German trade. Germany was able to use its comparative advantage in

the production of industrial goods, particularly in the ‘high technology’

sectors, to gain market share in world trade, and it could then use these

revenues to import an increasing volume of food and raw materials.28

This was a development which had profound strategic implications, and

it meant that relations with Britain, previously a peripheral issue in

German diplomacy, were now of central importance.

In the 1860s Bismarck was able to largely ignore British positions on

issues such as the annexation of Schleswig-Holstein and the progress

towards German uniWcation: little more than lip service was paid to

British concerns, British oVers of mediation were treated with barely

disguised condescension, and substantive British proposals were mostly

ignored. But this reXected the realities of the position: Britain was in no

position to intervene eVectively in a land war in continental Europe;

Prussia was a largely self-suYcient economy, whose ability to wage war

would not be aVected by a British blockade and which had few non-

European interests vulnerable to British sea power.29

To have maintained a position of self-suYciency and strategic inde-

pendence would have required policies which supported a higher level

of agricultural production than that achieved under the Kaiserreich.

TariVs would have had to be high enough to enable the grain sector to

match production costs in the American Mid-West. Resources would

have had to have been maintained in agricultural production, not

switched to industry, requiring a substantial reduction in the rate of

rural–urban migration. To achieve this, agricultural prices and incomes

28 The shift in the German position with regard to cereals was given in Table 7.1.There were substantial increases in the importation of other foodstuVs and raw materials.

29 Bismarck’s remark that Britain was a quantite negligeable in European politics wasquite justiWed from a Prussian point of view as long as Prussia remained solely a Contin-ental power, Kennedy (1980), 191.


would have had to be increased relative to urban levels. German

farmers would have beneWted at the expense of the urban consumer.

By choosing a moderate level of protection, which did not prevent a

substantial rise in imports, Imperial Germany adopted a policy which

provided some relief to agriculture but which did not prevent an in-

crease in living standards for the urban working class. Improved

incomes for workers contributed to the post-1896 fall in the urban

Gini-coeYcient. This in turn created economic conditions favourable to

the normalization of internal politics. It raised the likelihood that the

Social Democratic Party could be persuaded to turn away from revolu-

tionary politics and become an accepted part of the political system.

Germany faced a genuine socio-strategic dilemma. Economic self-

suYciency could be maintained only by the pursuit of policies which

were clearly hostile to the interests of the urban working class and

which would have perpetuated the alienation of that class. By contrast,

rising urban living standards could be secured only by recourse to

imported foodstuVs and raw materials. To a considerable extent these

came from British possessions, or from areas dominated by British

capital, and the overwhelming proportion crossed seas controlled by the

Royal Navy.30

Even in the event of a war with France and Russia, from which

Britain stood apart, the state of the German navy around 1900 was a

serious strategic deWciency. In 1903 Germany was outnumbered by

France alone. Germany had 9–12 ships of the line; France had 21. It

would only be possible to have a local superiority over the French

Atlantic Xeet if the French Mediterranean Xeet were prevented from

moving north and if the Russian Baltic Fleet did not join with the

French.31

To see Germany’s Flottenpolitik solely in terms of its eVect on in-

ternal politics, as historians in the Kehrite school tend to do, to see it as

a part of a Sammlungspolitik, a cause for the rallying of forces loyal to

the Empire, is to ignore or obscure this genuine dilemma.32 The eco-

nomic facts are clear: Germany in the late 1890s was not in the same

30 That there were particular problems associated with Germany’s geographical pos-ition is a point made in Sturmer (1990), but this study does not deal with the interactionof geography and the evolution of foreign trade.

31 Buchel memorandum, 27 Mar. 1903, Lambi (1984), 203.32 There is a way in which some Kehrite historians have linked the Flottenpolitik to

trade issues and that is to argue that German industry needed colonial expansion to gainnew markets, which in turn required a powerful Xeet, Wehler (1969), 112–26.


position as Prussia in the 1860s. Ignoring Britain was no longer an

option. Bismarckian condescension was not an appropriate policy. The

trade routes which brought imported food and colonial products to

Germany crossed those which brought the same goods to Britain. Only

one power could control these seas. Germany would have to confront or

conciliate.

In the Kehrite literature, the existence of a genuine strategic dilemma

is either downplayed or ignored altogether. Economic factors are not

considered as a part of the geo-political equation. Yet the German

government was certainly aware of the importance of foreign trade to

the German economy. Hohenlohe wrote in a letter in 1897: ‘Germany’s

trade has grown to such an extent that the government is obliged to

protect it. This can only be done by a Xeet which is more than just a

coastal protection force, but one capable of ensuring the safe passage of

our imports.’33 And the Kaiser wrote ‘richtig’ in the margin of a gov-

ernment report against the following passage: ‘without her overseas

trade Germany could not exist, and this trade would not prosper for

long without the protection of a Xeet capable of commanding respect’.34

Tirpitz himself was well aware of the economic implications of the

plan. He saw an intimate connection between the expansion of the Xeet

and the growth of German trade. In his public statements he tended to

play down the trade implications in order to avoid antagonizing the

agrarian lobby, who would have preferred Germany to remain self-

suYcient in food, rather than exporting industrial products to pay for

food imports. But there was an awareness of the strategic consequences

of the changing structure of German imports.35

Critics of the Tirpitz naval programme need to explain how

Germany could have resolved this dilemma. What was the alternative

to building a navy? A British alliance was one possibility, and this was

considered at various times by Germany’s leaders. Yet this had the clear

implication that Germany could be drawn into conXict with other Euro-

pean powers (most probably Russia) as a result of disputes arising from

Britain’s imperial interests and possessions. The British Empire would

33 Letter from Hohenlohe to SoldendorV, 7 Nov. 1897; quoted in Berghahn (1971),133.

34 The document was a report from the Prussian representative in Bavaria, Berghahn(1971a), 134. In 1899 Tirpitz predicted that, without a Xeet, Germany would sink back tothe status of a poor agricultural country, Herwig (1987), 35.

35 Deist (1976), 110–12. Trade issues were, however, used to gain the support ofinXuential lobby groups and industrial organizations, Pogge von Strandmann (1972).


be defended ‘on the banks of the Elbe’, at immense cost to Germany.36

To have eschewed naval expansion in the absence of a formal British

alliance would have meant that Germany would have little choice but to

accept British mediation in the event of a prolonged European war.

Britain would be the arbitrator of Europe, with unpredictable conse-

quences. In addition, German trade would have been unprotected from

attack by the French or Russian navies should such a war involve these

powers.37

Eventually, after 1945, West Germany was able to reap the rewards,

in terms of rising living standards and political stabilization, of a trade-

orientated economic policy. But this was done in the context of an

American alliance which did not just provide security to West German

trade routes but also oVered an eVective guarantee of land frontiers as

well. A pre-1914 British alliance oVered only the Wrst part of this: trade

security at the expense of a more dangerous position on land.

To recognize this dilemma is not to deny that internal weaknesses

had an adverse eVect on German foreign policy. A more far-sighted

foreign policy, more Xexible and less inXuenced by domestic political

considerations, might have kept potential opponents isolated, might

have prevented the formation of a power bloc hostile to Germany,

might have pursued a policy of naval expansion less obviously driven

by hostility to British interests, and might, in short, have avoided

the drift towards war on terms which were unfavourable to Germany.

The contrast with British policy is instructive. Once German policy had

shifted to one of evident confrontation, the British response was indeed

far-sighted and Xexible. Outstanding disputes with France, Russia,

and the United States were settled, even at the cost of giving up long-

held positions (conceding a Russian sphere of inXuence in Persia, recog-

nizing the Monroe doctrine). An empire built on an implicit sense of

racial superiority concluded an alliance with Japan. German policy, by

contrast, was hesitant and needlessly antagonistic. Internal weaknesses

had a part in this. But Germany was not the only country pre-1914

whose foreign policy was adversely aVected by internal weakness.

36 As the elder Pitt had claimed to have won Canada in 1759, by distracting France inGermany.

37 The Kehrite historians’ concentration on the Tirpitz plan has obscured the fact thatmuch of German naval planning was concerned with France and Russia—for example, the1899 plan, Lambi (1984), 193. Any navy large enough to match the combined strength ofFrance and Russia, and therefore give Germany the option of planning for a prolongedwar against these two countries, was also large enough to be a threat to Britain.


Russia and Austria-Hungary are two obvious examples. It was the qual-

ity of the British policy response to Germany’s challenge which was

truly exceptional.38

Kehrite historians have ignored the economic dimension of the

Tirpitz plan, and concentrated on the internal political aspects. The

result has been to underestimate the genuine dilemma faced by German

policy-makers, and to present the plan as a cynical attempt by an out-

moded elite to retain power and obstruct political reform. Even if such

motives did play a part, the success of the naval programme in at-

tracting widespread support owed a lot to the economic imperatives

which lay behind it.

5. Growth in Mature and in Industrializ ing Societies :

The Economic Preconditions for Polit ical Progress

Economic growth in mature industrial societies diVers from that in

industrializing societies in that the main engine of growth is improve-

ments in productivity within each sector not the transfer of labour and

other resources from sectors where productivity is low into more

modern sectors. In mature economies the proportion of the labour force

in agriculture and other lagging sectors (if agriculture is still lagging) is

typically a low one, below 10%. Further transfers from this sector can

make only a modest contribution to overall growth.

In the early stages of industrialization, structural shifts are an import-

ant component of overall growth. Indeed, productivity growth within

each sector is often quite modest, so that the contribution of structural

change is a vital one. Thus a high rate of migration out of agriculture

into the urban or industrial sector is intrinsic to the growth process, as

well as a major force of social transformation.

The Lewis Model is valuable in that it provides an explanation for

some features of the operation of urban labour markets in industrial-

izing countries, and may help to explain the high level of urban inequal-

ity found in Germany and other countries. But this is only part of the

broader process of social change brought about by rural–urban

38 The contempt with which German foreign policy is treated by German historianscan come as a surprise: was German policy really that of a ‘deluded upstart’, Hildebrand(1995), 10.


migration: a rural society bound by traditional ties of obligation, defer-

ence, and reciprocity is cast aside and a new urban structure, built on

very diVerent principles, has to be constructed in its place. Inevitably,

this takes time. Initially, the sense of dislocation produced by migration

is intensiWed by the contrast between the stability of rural institutions

and values compared to those of an urban society which is in a state of

Xux.As the growth process continues, the contribution made by structural

change will tend to fall, unless exceptional factors intrude (such as the

check to structural change experienced by Germany and many other

European countries in 1913–50, which left room for a rapid catch-up in

1950–73). As the rural labour surplus is used up, the opportunities for

rapid growth as a result of the redeployment of labour which was

previously underemployed will be reduced. Migration in and out of

urban labour markets will become a stabilizing rather than a destabil-

izing process.

The analysis of the previous chapters has concentrated on the social

and economic consequences of the ‘labour surplus’ phase of industrial-

ization. It has looked at the eVects of migration out of agriculture into

urban labour markets, it has considered the movement of inequality in

the rural and urban sectors, and it has examined the eVects of change

and transformation in agriculture. Standing back from these trends, it is

possible to discern the broad movements: a pattern of sharpening ten-

sions and divisions up to the 1890s followed by a period of stabilization

and improvement. The elements of this better position post-1895 are

the following:

1. An end to the ‘labour surplus’ pattern of migration dominated by

the transfer of surplus population out of agriculture. The sharp

decline in emigration to the United States after 1895 is a further

indication of the changing nature of migrant behaviour.

2. The stabilization and slight decline in the overall Prussian Gini-

coeYcient after 1900, combined with the fact that inequality within

the rural and urban sectors was declining from 1896 onwards.

3. The improved position of German agriculture after 1895, as indi-

cated by rising land prices, the reduction in forced sales, and

increases in rural wage levels. This in turn reduced the pressure

on urban labour markets.

4. The recovery of the German capital markets from the crisis of the

1870s was complete by 1900; the robustness of the new system


was demonstrated by the fact that investment levels remained

high despite short-term Wnancial crises in the post-1900 period.

The urban sector was better able to Wnance the creation of jobs for

rural migrants.

5. The strong growth of German exports, particularly in ‘high tech’

sectors, promised well for the prospects of the German economy in

the twentieth century. So long as world trade remained open, and

volumes increased, Germany could expect a steady improvement in

its share of world exports, making it possible to Wnance an increased

bill for imported raw materials and foodstuVs, and to deliver con-

tinued improvements in living standards for German citizens.

From the mid-1890s onwards the German economy developed in

directions which created the preconditions for social and political stabil-

ization.39 These favourable preconditions were not enough, in them-

selves, to guarantee the successful transformation of the Kaiserreich into

a modern democracy. Even if there had been no outbreak of war in

1914 there was still much that could have gone wrong. But circum-

stances were more favourable in the early twentieth century than they

had been in the 1870s and 1880s.

In the end, though, the weakness of the Kehrite argument lies in the

normality of the pre-1914 German experience. Germany went through

a process of rapid industrialization, and this brought with it economic

and social problems common to many countries going through this

phase. The tools needed to understand this are available in the com-

parative study of industrializing societies. There was no Sonderweg

before 1914. The outbreak of war in 1914 had other roots, other causes.

6. July 1914: A Derailment or the End of a Sackgasse?

The idea of this study was conceived during a visit to China in 1995.

There was an evident labour surplus in the countryside; the migration

of this surplus to already crowded cities was clearly going to have

serious social, economic, and political consequences. This observation

39 This view contrasts with the one put forward by some historians that industrializa-tion inevitably led to new social conXicts, Puhle (1976), 340. For GeoV Eley, the ‘centralcontradiction of Wilhelmine politics’ was failure of the state to adapt to the new task ofsocial reconciliation caused by the basic ‘antagonism of capital and labour’, Eley (1980),349. The eVect of economic and social change has also been blamed for the readiness ofthe middle classes to embrace radical nationalist ideas, Chickering (1984), 302.


prompted a general reXection. If there is a labour surplus period in

economic development—a ‘hump’ which many countries have to sur-

mount—does this aVect the stability of countries going through this

process? Are they potentially dangerous factors in world politics? Is this

an explanation of the political problems of pre-1914 Germany? Did

these in turn aVect German foreign policy? Will a similar process make

China a disruptive force in world politics in the twenty-Wrst century?The conclusions are in many ways more optimistic than the fears

which were prompted by this visit. The autonomy of the foreign policy

problems and the internal problems of the German empire is the main

conclusion to emerge. While this is not a conclusion which necessarily

applies to every state in every era, it is at least more comforting than the

implication of the Kehrite view: that a state which lacks the ability to

tackle its internal problems will turn to an aggressive foreign policy in

an attempt to defuse domestic tensions.

The picture given in this study is one of a society which, by 1914,

was having considerable success in dealing with the problems of rapid

industrialization and the ‘labour surplus’ period. These were problems

which were unprecedented in their scale and intensity. Germany 1870–

1913 was the Wrst ‘developing economy’, in the sense that it was the

Wrst to go through a process of telescoped industrialization and fast

economic growth characteristic of countries ‘catching up’ with a tech-

nological leader. The speed of German industrialization and urbaniza-

tion was rapid, perhaps three times as fast as Britain at a similar stage.

If there is a connection between the economic and political problems

of Imperial Germany, it probably lies in an earlier period, before 1895.

Evidence presented in this study shows that the time of maximum

rural–urban migration was also one of intense urban problems: high

inequality and low life expectancy. It was during this period that Marx-

ist ideas became inXuential in working-class districts. This develop-

ment, and the violence of the reaction by the Bismarckian state which it

provoked, had a profound eVect on the political system of Imperial

Germany. It is in this period that the ‘Weber–Lewis–Kuznets’ hypoth-

esis has most value for those seeking to explain the course of German

history.

But by 1914 the chances of democratic transformation were much

improved. There remained massive obstacles. For many historians,

these were so great as to be insuperable.40 Consequently, it can be

40 It was a situation of ‘intolerable complexity’, according to Eley (1980), 350.


argued that the rising demand for change inevitably created tensions

which, even if these were not directly linked to German policy during

July 1914, produced a background which aVected the decision-makers’

ability to resolve the crisis. The atmosphere of pessimistic fatalism, the

general weariness with the political process and the compromises it

required to keep operating, a willingness to take risks and consider

radical solutions—these were dangerous sentiments in a context which

required clear thought and an ability to appreciate the consequences of

a mistaken policy.41

The view that the Kaiserreich was in a political cul-de-sac, a Sack-

gasse, where the options open to the executive were narrowing, is one

which is held by many historians outside the Kehrite school.42 The

arguments of Manfred Rauh, that a process of ‘quiet Parliamentariza-

tion’ was already under way, have been widely derided.43 However, if

the option of peaceful progress was ruled out, then the solution would

have had to have been a violent one of some sort.

But, the implications of the arguments about economic preconditions

put forward in this study are that the path to democracy was getting

easier, not more diYcult. In which case, even if Rauh overstated the

extent of the progress made by 1914, the prospects for further advances

were improving. The events of 1914 represented a derailment, not the

hitting of the buVers at the end of a Sackgasse.


and Sources

(i) Appendix Table 10B.1

SDP1912


Social Democrat share of the vote in urban

constituencies, from Die Reichstagswahlen von

1912, SDR vol. 250 (1913).

41 The view that the policy which led to the outbreak of war was based on an inabilityto appreciate the consequences of German policy, leading to the mistaken view that theconXict could be limited, was put forward in Jarausch (1969). But the consensus now isthat Germany pushed for war, Pogge von Strandmann (1988).

42 The term Sackgasse is taken from the title of a paper by Volker Berghahn, Berghahn(1971).

43 The phrase is found in Rauh (1973), 347. The point is made more forcefully in Rauh(1977).


POPULATION(log) Log of total population in 1910, from SJDS

(1911), 67–8.

CATHOLIC% The percentage of the population who were

Catholic, 1910 census, SJDS (1911), 678–9.

PRUSSIA-CATH% As above, Prussian cities only.

NONPR-CATH% As above, non-Prussian cities only.

MOBILITY Average moves in and out, divided by total

population. Data from SJDS (1911), 67–8, and

subsequent issues.

TOWNWAGES1912 Average wages (Ortsubliche Tagelohne) for male

workers, aged 16 and over, from SJDS (1913),

826.

FACTORYSIZE Total number of employed workers divided by

the number of enterprises, 1912, from SJDS

(1916), 386–7.SDP1878 Social Democrat share of the vote in urban

constituencies, from SDR a.f. vol. 37 (1879).

INEQUALITY Estimated levels of inequality, as ratios of the

income shares of the top 10% to the bottom

40% (U10/L40 ratios). Data from Kuhnert

(1916); estimation procedure is given in

Appendix B of Grant (2002b).

(ii) Appendix Table 10B.2

SDP1912


Social Democrat share of the vote in non-

urban constituencies, from Die Reichstagswah-

len von 1912, SDR vol. 250 (1913).

%AGOCCUP07 The percentage of the population occupied in

agriculture, from the 1907 occupational census,

SDR vol. 209 (1909).%CATHOLIC The percentage of the population who were

Catholic, 1890 census, PS vol. 121, pp. 152–98.

%SPEAKPOLISH As above, the percentage of the population

who spoke Polish.

%SPEAKLITHUAN As above, for Lithuanian speakers.

CITYDISTANCE(a) Distance in kilometres from the Kreis mid-


inhabitants in 1900 (measured from contem-

porary maps).


CITYDISTANCE(c) As above, for cities with at least 50,000 inhab-

itants.

LAND>100HA The percentage of the total agricultural area in

holdings of over 100 hectares, from 1895 Agri-

cultural Census, SDR n.f. 212.

LAND20-100HA As above, for holdings of between 20 and 100

hectares.

%GUTSBEZIRKE The numbers of Gutsbezirke as percentages of

all administrative units, from PS vol. 81

(1884), part 1, pp. 20–65.

INEQUALITY As Table 10.1 (averages of Kreis results for

each constituency).

RURALWAGES Average wages (Ortsubliche Tagelohne) for rural

areas; the Wgures are averages of male and

female daily rates, 1892 and 1901, ZdKPSB(1904), 520–8.

Appendix 10B. Econometric Analysis

The dependent variable in these regressions was the share of the Social

Democrats in the 1912 election. The analysis showed that the SDP vote

was highest in the larger cities (the log of the resident population has a

strong eVect on voting behaviour), and also that the presence of a large

Catholic population held it down (column a of Table 10B.1). In the

second column a distinction is drawn between non-Prussian cities and

Prussian cities. The inXuence of Catholicism is shown to have been

signiWcantly greater inside Prussia.

In the other columns a variety of economic variables are introduced.

These turned out to have, at the most, a slight eVect on voting behav-

iour. Two labour market indicators were used. MOBILITY is a meas-

ure of migration (total moves per head of population for 1910–12); this

was weakly linked with an increase in the SPD vote, but the result was

neither robust nor statistically signiWcant. A second indicator, the ratio

of applicants to jobs reported by the labour exchanges, was completely

insigniWcant (a t-statistic of 0.16 when added to the equation estimated

in column c).44 Short-term movements in job market conditions did not

44 The variable gave the ratio of applicants to vacancies for 1910–12, derived fromtown labour exchange statistics (SJDS 1911–13).


Table 10B.1. Regression results for linear log-odds voting model, 1912 election: dependent variable is logistic

transformation of the SDP vote in major cities; estimated by OLS

a b c d e f

Constant �1.51 �1.55 �1.94 �2.30 �1.22 �2.91

POPULATION(log) þ0.332 (3.79) þ0.341 (4.28) þ0.350 (4.10) þ0.314 (3.48) þ0.167 (2.26) þ0.271 (2.35)

CATHOLIC% �0.0143 (5.72) �1.985 (6.90)

PRUSSIA-CATH% �0.0183 (7.34) �0.0183 (6.98) �0.0188 (7.72) �0.0086 (2.87)

NONPR-CATH% �0.0068 (2.29) �0.0026 (0.67) �0.0079 (2.65) �0.0039 (1.46)

MOBILITY þ0.0098 (1.22)

TOWNWAGES1912 þ0.263 (1.51) þ0.228 (1.75) þ0.617 (2.75)

FACTORYSIZE þ0.0022 (1.55) þ0.0016 (1.27) þ0.0010 (0.57)

SDP1878 þ0.232 (4.61)

INEQUALITY �0.063 (0.72)

N 70 70 65 70 55 46

S 0.567 0.516 0.512 0.503 0.332 0.520

R2 0.435 0.540 0.539 0.576 0.706 0.637

adj R2 0.418 0.519 0.519 0.543 0.669 0.592

F-test of regression 25.82 25.84 19.70 17.40 19.19 14.03

RSS 21.563 17.560 15.699 16.185 5.302 10.822

appear to matter at all. Wage levels were positively linked with the

socialist vote (column c), as was factory size (the average number of

employees per enterprise with at least 10 workers), but both results are

below the 10% signiWcance level.

In the fourth column the socialist share of the vote in 1878 was

included (SPD1878). As might be expected, this has an eVect on the

other coeYcients. It shows the persistence of the socialist vote and the

importance of the early establishment of a separate working-class

culture and organization.

Finally, measures of income inequality derived from Prussian tax

statistics were included: the ratio of the share of the top 10% of the

distribution to that of the bottom 40% (INEQUALITY). This meant

that the analysis could only be run for the 46 Prussian cities. The

results showed no connection between inequality and the socialist vote.

Inequality does not appear to have been a factor in the electoral per-

formance of the SPD in urban areas.

Table 10B.2 gives the results of a similar analysis for the remaining

Prussian constituencies. This is a data set covering 204 rural or small

town constituencies. Two variables have a major eVect on voting behav-

iour: the percentage occupied in agriculture and the Catholic propor-

tion. Both coeYcients are strongly negative, and the two variables

together explain much of the variance in the dependent variable

(column a). Again, the slow relative decline of the agricultural popula-

tion was a factor working in favour of the SDP. Adding the percentage

of the population who spoke Polish (column b) conWrms the point made

earlier, that there remains a strong negative eVect from Catholicism

even when allowance is made for this factor. However, Lithuanian-

speakers showed a preference in favour of the SPD, so the presence of

this group tends to raise the SPD vote (these were found in Memel and

a few other constituencies in the Gumbinnen region of East Prussia).

Cities also had an eVect on the surrounding countryside. The two

city distance variables, measuring the distance from major cities, both

have coeYcients that are negative and signiWcant: the SPD vote tended

to fall as the distance from major cities rose. The eVect of the larger

cities (CITYDISTANCE(a) over 200,000 inhabitants) is larger than

that for smaller cities (CITYDISTANCE(c) over 50,000 inhabitants),

as would be expected given the strength of the SPD vote in the major

cities.

The eVect of large farms and estates was examined using two vari-

ables. The Wrst was the percentage of agricultural land in holdings of


Table 10B.2. Analysis of the SDP vote in Prussian constituencies (excluding urban constituencies), 1912.Dependent variable is logistic transformation of vote

a b c d e

Constant 1.417 1.262 1.588 0.816 1.307

%AGOCCUP07 �0.0473 (14.9) �0.0371 (12.1) �0.0383 (11.9) �0.0360 (10.8) �0.0370 (12.0)

%CATHOLIC �0.0240 (15.4) �0.0181 (11.2) �0.0182 (11.2) �0.0179 (10.9) �0.0183 (10.8)

CITYDISTANCE(a) �0.0127 (4.37) �0.0128 (4.42) �0.0126 (4.30) �0.0128 (4.37)

CITYDISTANCE(c) �0.0110 (2.37) �0.0110 (2.37) �0.0102 (2.15) �0.0107 (2.27)

%SPEAKPOLISH �0.0122 (4.25) �0.0127 (4.41) �0.0122 (4.27) �0.0121 (4.24)

%SPEAKLITHUAN þ0.0281 (2.37) þ0.0294 (2.48) þ0.0298 (2.47) þ0.0281 (2.36)

LAND>100HA �0.0199 (3.80) �0.0182 (3.37) �0.0206 (3.88) �0.0200 (3.80)

%GUTSBEZIRKE þ0.0296 (5.67) þ0.0295 (5.67) þ0.0321 (5.31) þ0.0294 (5.61)

INEQUALITY �0.225 (1.28)

RURALWAGES þ0.150 (0.83)

LAND20-100HA �0.00124 (0.36)

N 204 204 204 204 204

S 0.810 0.661 0.660 0.662 0.663

R2 0.680 0.793 0.795 0.794 0.793

adj R2 0.677 0.785 0.785 0.784 0.784

F-test 214.00 93.44 83.52 83.00 82.70

RSS 131.75 85.30 84.59 85.00 85.25

over 100 hectares (LAND>100HA). This variable had a negative eVect

on the SDP vote: workers on large farms and estates tended to be less

likely to vote for the SPD, perhaps for reasons of deference or loyalty to

the estate-owner, or possibly out of fear that such a vote would be

discovered and lead to retribution. However, the second variable,

%GUTSBEZIRKE, has a positive sign. This had a persistent eVect on

voting behaviour, and this result is robust to the inclusion or exclusion

of other variables. The estimated coeYcients for this variable are unex-

pectedly large given the relative unimportance of the Guter in this per-

iod. They are rather higher than the coeYcients on LAND>100HA,

indicating that resentment of the Guter was apparently a rather stronger

eVect on voting behaviour than deference on larger holdings.

The remaining columns introduce some other variables: the results in

column d show that the level of wages did not aVect voting behaviour,

in contrast to the results for the larger cites. Those in column e show

that the presence of medium-sized holdings did not aVect the results

either (these holdings typically employed some labour). But perhaps the

most interesting result from this table is in column c, which shows that

inequality at the Kreis level, measured by income inequality, had no

signiWcant eVect on the SPD vote. This conWrms the result given in

Table 10.B1, column f, that the level of inequality was not a factor in

the SDP vote in the cities. Political inequality, as represented by the

%GUTSBEZIRKE variable, had much more eVect.


REFERENCES AND SOURCES

a. abbreviations used for main statist ical sources

1. Publications of the Imperial Statistical OYce

ErgVSDR Erganzungsheft zur Vierteljahrsheft zur Statistik des Deutschen Reichs

MSDR Monatsheft zur Statistik des Deutschen Reichs

SDR Statistik des Deutschen Reichs

SJDR Statistisches Jahrbuch fur das Deutsche Reich

VSDR Vierteljahrsheft zur Statistik des Deutschen Reichs

2. Publications of the Prussian Statistical OYce

PS Preußische Statistik

SJPS Statistisches Jahrbuch fur das Preußischen Staat

ZdKPSB Zeitschrift des Koniglich Preußischen Statistischen Bureaus

3. Other publications

SJB Statistisches Jahrbuch der Stadt Berlin

SJDS Statistisches Jahrbuch deutscher Stadte

SJK Statistisches Jahrbuch der Stadt Koln

b. references

Abel , W. (1962), Geschichte der deutschen Landwirtschaft vom fruhen Mittelalter

bis zum 19. Jahrhundert (Stuttgart: E. Ulmer).

Abelshauser , W. (1983), Wirtschaftsgeschichte der Bundesrepublik Deutschland

1945–1980 (Frankfurt am Main: Suhrkamp).

Albrecht , K. (1927), ‘Die Preußischen Gutsbezirke’, Zeitschrift des Preuß-

ischen Statistischen Landesamts, 3: 344–477.

Amsden , A. (1989), Asia’s Next Giant: South Korea and Late Industrialization

(Oxford: Oxford University Press).

——(2001), The Rise of ‘the Rest’: Challenges to the West from Late-Industrial-

izing Economies (Oxford: Oxford University Press).

Anand , S. and Kanbur , S. M. (1993), ‘The Kuznets process and the

inequality–development relationship’, Journal of Development Economics, 40/

1: 25–52.

Anderson , M. L. (2000), Practicing Democracy: Elections and Political Culture

in Imperial Germany (Princeton: Princeton University Press).

Asian Productivity Organisation (1994), Agrarian Structure and Reform Measures

(Tokyo: APO).

Atkinson , A.B. (1996), ‘Income distribution in Europe and the United

States’, Oxford Review of Economic Policy, 12: 15–27.

——(1996a), ‘Seeking to explain the distribution of income’, in J Hills (ed.),

The New Inequalities: The Changing Distribution of Income and Wealth in the

United Kingdom (Cambridge: Cambridge University Press).

——and Brandolini , A. (1999), ‘Promises and Pitfalls in the Use of ‘‘Sec-

ondary’’ Data-Sets: Income Inequality in OECD Countries’, manuscript,

NuYeld College, Oxford and Banca D’Italia, Research Department.

——Rainwater , L. and Smeeding , T. M. (1995), Income Distribution in

OECD Countries, OECD Social Policy Studies no. 18.

Aug E-Larib E, M. (1950), La Politique agricole de la France de 1800 a 1940

(Paris: Presse Universitaire).

Baack , L. J. (1977), Agrarian Reform in Eighteenth Century Denmark, Univer-

sity of Nebraska Studies 56.

Bade , K. J. (1980), ‘German emigration to the United States and continental

emigration to Germany in the late nineteenth and early twentieth centuries’,

Central European History, 12/4: 348–77.

——(1984), ‘Vom Auswanderungsland zum ‘‘Arbeitseinfuhrland’’: kontinentale

Zuwanderung und Auslanderbeschaftigung in Deutschland im spaten 19. und

fruhen 20. Jahrhundert’, in K. J. Bade (ed.), Auswanderer, Wanderarbeiter,

Gastarbeiter (OstWldern: Scripta Mercaturae Verlag).

——ed. (1987), Population, Labour and Migration in 19th and 20th Century

Germany (Leamington Spa: Berg).

——(1991), ‘German transatlantic emigration in the nineteenth and twentieth

centuries’, in P. C. Emmer and M. Morner (eds.), European Expansion and

Migration (New York: Berg).

Baines , D. (1994), ‘Population, migration and regional development,

1870–1939’, in R. Floud and D. McCloskey (eds.), The Economic History of

Britain since 1700, vol. 2 (Cambridge: Cambridge University Press).

Bairoch , P. (1989), ‘Les trois revolutions agricoles’, Annales, 44/2: 317–54.

Ballod , C. (1914), ‘Die Bevolkerungsbewegung der letzten Jahrzehnte in

Preußen und in einigen anderen wichtigen Staaten Europas’, Zeitschrift des

Koniglich Preußischen Statistischen Bureaus, 3: 237–90.

Bardhan , P. (1989), ‘Alternative approaches to the theory of institutions in

economic development’, in P. Bardhan (ed.), The Economic Theory of Agrar-

ian Institutions (Oxford: Oxford University Press).

362 References and Sources

Barfuss , K. M. (1985), Gastarbeiter in Nordwestdeutschland 1884–1918

(Bremen: Staatsarchiv).

Barkin , K. D. (1970), The Controversy over German Industrialisation

1890–1902 (Chicago: University of Chicago).

Basu , K. (1997), Analytical Development Economics: The Less Developed Econ-

omy Revisited (Cambridge, Mass.: MIT Press).

Bateman , J. (1879), The Great Landowners of Great Britain and Ireland

(London: Harrison).

Baten , J. (1999), Ernahrung und wirtschaftliche Entwicklung in Bayern

(1730–1880), Beitrage zur Wirtschafts- und Sozialgeschichte 82 (Stuttgart:

F. Steiner).

Bauerlicher Zustand in Deutschland (1883), Schriften des Verein fur Sozialpolitik,

vols. 21–4.

Below , G. von (1891), ‘Rittergut’, in Handworterbuch der Staatswissenschaft,

suppl. vol. I: 672–5.

Benabou , R. (1996), ‘Inequality and growth’, NBER Macroeconomics Annual,

xi: 11–74.

Berger , S. (2000), Social Democracy and the Working Class in Nineteenth and

Twentieth Century Germany (Harlow: Longman).

Berghahn , V. (1971), ‘Das Kaiserreich in der Sackgasse’, Neue Politische

Literatur, 16: 494–506.

——(1971a), Der Tirpitzplan: Genesis und Verfall einer innenpolitischen Krisen-

strategie unter Wilhelm II (Dusseldorf: Droste Verlag).

——(1994), Imperial Germany,1871–1914: Economy, Society, Culture and Polit-

ics (Providence and Oxford: Berghahn).

Berkner , L. K. and Mendels , F. (1980), ‘Inheritance systems, family struc-

ture and demographic patterns in Western Europe, 1700–1900’, in C. Tilly

(ed.), Historical Studies of Changing Fertility (Princeton: Princeton University

Press).

Berlin, Statistisches Amt (1887), Die Berliner Volks-Zahlung vom 2. Dezember

1885 (Berlin).

——(1893), Die Bevolkerungs- und Wohnungs-Aufnahme vom 2. Dezember 1890

in der Stadt Berlin (Berlin).

——(1913–16), Die Grundstucks-Aufnahme vom 15. Oktober 1910, sowie die

Wohnungs- und Bevolkerungs-Aufnahme vom 1. Dezember 1910 in der Stadt

Berlin und 44 Nachbar-Gemeinden (Berlin).

Berthold , R. (1978), ‘Die Veranderungen im Bodeneigentum und in der

Zahl der Bauernstellen, der Kleinstellen und der Ritterguter in den

preußischen Provinzen Sachsen, Brandenburg und Pommern wahrend der

Durchfuhrung der Agrarreformen des 19. Jahrhundert’, Jahrbuch fur

Wirtschaftsgeschichte, Sonderband, 10–116.


Berthold , R. et al., eds. (1990), Geschichte der Produktivkrafte in Deutschland

von 1800 bis 1945. Band 1: Produktivkrafte in Deutschland 1800 bis 1870

(Berlin: Akademie Verlag).

Bertram , J. (1964), Die Wahlen zum Deutschen Reichstag vom Jahre 1912.

Parteien und Verbande in der Innenpolitik des Wilhelminischen Reichs (Dussel-

dorf: Droste).

B i edermann , E. (1918), ‘Die Einkommens- und Vermogensverhaltnisse der

preußischen Bevolkerung im Zeitraum 1895–1914, Zeitschrift des Koniglich

Preußischen Statistischen Bureaus, 60–86.

B irdsell , N. (1988), ‘Economic approaches to population growth’, in H.

Chenery and T. Srinivasan (eds.), Handbook of Development Economics, vol.

I (Amsterdam: North-Holland).

B jorn , C. (1977), ‘The peasantry and agrarian reform in Denmark’, Scandi-

navian Journal of Economic History, 25/1: 117–37.

Blackbourn , D. (1980), Class, Religion and Local Politics in Wilhelmine Ger-

many: The Centre Party in Wurttemberg before 1914 (New Haven: Yale Uni-

versity Press).

——(1997), The Long Nineteenth Century: The Fontana History of Germany

1780–1918 (London: HarperCollins).

——and Eley , G. (1984), The Peculiarities of German History (Oxford).

Board of Trade (1908), ‘Report of an Enquiry by the Board of Trade into

Working Class Rents, Housing and Retail Prices Together with Rates of

Wages in Certain Occupations in the Principal Industrial Towns of the

German Empire’, Parliamentary Papers, Accounts and Papers (47), cd 4032.

B ohme , H. (1966), Deutschlands Weg zur Gro�macht: Studien zum Verhaltnis

von Wirtschaft und Staat wahrend der Reichgrundungszeit 1848–1881 (Berlin

and Cologne: Oxford University Press).

Bolsker-Schlicht , F. (1997), ‘Deutsche Saisonarbeiter in den Niederlan-

den bis zum 19. Jahrhundert’, in H.-H. Nolte (ed.), Deutsche Migrationen

(Munster: LIT Verlag).

Borchardt , K. (1982), ‘Was there a capital shortage in the Wrst half of the

nineteenth century in Germany?’, in K. Borchardt, Perspectives on Modern

German Economic History and Society (Cambridge: Cambridge University

Press).

Borjas , G. J. (1994), ‘The economics of immigration’, Journal of Economic

Literature, 23 (Dec.), 1667–1717.

Bourgignon , F. (1990), ‘Growth and inequality in the dual model of devel-

opment: the role of demand factors’, Review of Economic Studies, 57: 215–28.

——and Morrison , C. (1998), ‘Inequality and development’, Journal of De-

velopment Economics, 57: 233–57.

Bowley , A. L. (1946), Elements of Statistics, 6th edn. (New York: Staples).

Boyer , G. R. (1990), An Economic History of the English Poor Law, 1750–1850

(Cambridge: Cambridge University Press).


Bray , F. (1986), The Rice Economies: Technology and Development in Asian

Societies (Oxford: Blackwells).

Bresciani . C. (1907), ‘Uber die Methoden der Einkommenverteilungstatistik’,

Jahrbucher fur Nationalokonomie und Statistik, 33: 577–600.

Brezis , E., Krugman , P. and Ts iddon , D. (1993), ‘Leapfrogging in inter-

national competition: a theory of cycles in national technological leadership’,

American Economic Review, 83/5: 1211–19.

Brinkmann , T. (1935), Theodore Brinkmann’s Economics of the Farm Business

(Berkeley: Social Science Research Council).

Broadberry , S. N. (1997), ‘Anglo-German Productivity DiVerences: A Sectoral

Analysis’, unpubl.

——(1997a), The Productivity Race: British Manufacturing in International Per-

spective, 1850–1990 (Cambridge: Cambridge University Press).

Brodrick , G. (1881), English Land and Landlords (London: Cassell, Petter,

Galpin).

Broes ike , M. (1907), ‘Die Binnenwanderung im Preußischen Staat’, Zeits-

chrift des Koniglich Preußischen Statistischen Bureaus, 1: 1–62.

Brouwer , W. (1957), ‘A. Schultz-Lupitz zum 125. Geburtstag’, Zeitschrift fur

Agrargeschichte und Agrarsoziologie, 5/1: 69–75.

Bruno , M., Ravaillon , M. and Squire , L. (1996), Equity and Growth in

Developing Countries: Old and New Perspectives on the Policy Issues, Policy

Research Working Paper 1563 (Washington: World Bank).

Bry , G. (1960), Wages in Germany 1871–1945 (Princeton: Princeton University

Press).

Buchsteiner , I. (1994), ‘Besitzkontinuitat, Besitzwechsel und Besitzverlust in

der Gutswirtschaft Pommerns 1879–1910’, in H. Reif (ed.), Ostelbische Agrar-

gesellschaft im Kaiserreich und in der Weimarer Republik (Berlin: Akademie

Verlag).

Burgd orfer , F. (1931), ‘Migration across the frontiers of Germany’, in I

Ferenczi and W. F. Willcox, International Migrations, vol. 2 (New York:

NBER).

Calleo , D. (1978), The German Problem Reconsidered: Germany and the World

Order, 1870 to the Present (Cambridge: Cambridge University Press).

Carsten , F. (1989), A History of the Prussian Junkers (Aldershot: Scolar

Press).

Catt , C. (1986), ‘Farmer and factory worker: rural society in Imperial Ger-

many, the example of Maudach’, in R. J. Evans and W. R. Lee (eds.), The

German Peasantry (London: Croom Helm).

Caumanns , U. (1994), Technischer Fortschritt und Sozialer Wandel in deutschen

Ostprovinzen (Bonn: Kulturstiftung der deutschen Vertriebenen).

Chickering , R. (1984), ‘We men who feel most German’: a cultural study of

the Pan-German League (London: George Allen & Unwin).


Clapham , J. H. (1921, repr. 1966), The Economic Development of France and

Germany 1815–1914 (Cambridge: Cambridge University Press).

Clarke , G. (1995), ‘More evidence on income distribution and growth’, Jour-

nal of Development Economics, 47: 403–27.

Coale , A. J. and Watkins , S. C., eds. (1986), The Decline of Fertility in

Europe (Princeton: Princeton University Press).

Conrad , J. (1891), ‘Bodenzersplitterung’, in Handworterbuch der Staatwis-

senschaft, vol. 3 (Jena: G. Fischer).

Conze , W. (1969), ‘The eVects of nineteenth century liberal agrarian reform

on social structure in central Europe’, in F. Crouzet (ed.), Essays in European

Economic History (London: E. Arnold).

Craigie , P. (1887), ‘The size and distribution of agricultural holdings in

England and abroad’, Journal of the Royal Statistical Society, 100: 86–142.

Crew , D. (1979), Town in the Ruhr: A Social History of Bochum 1860–1914

(New York: Columbia University Press).

Crothers , G. D. (1941), The German Elections of 1907 (New York: AMS).

Dade , H. (1913), Die deutsche Landwirtsaft unter Kaiser Wilhelm II (Halle:

C. Marhold).

Dasgupta , P. (1993), An Inquiry into Well-Being and Destitution (Oxford:

Oxford University Press).

Deininger , K. and Squire , L. (1996), ‘A new data set measuring income

inequality’, World Bank Economic Review, 10/3: 565–91.

——and——(1997), ‘Economic growth and income inequality: re-examining

the links’, Finance and Development (Mar.), 38–41.

——and ——(1998), ‘New ways of looking at old issues: inequality and

growth’, Journal of Development Economics, 57: 259–87.

Deist , W. (1976), Flottenpolitik und Flottenpropaganda: das Nachrichtenbureau

des Reichsmarineamtes 1897–1914 (Stuttgart: Deutsche Verlagsanstalt).

Del Fabbro , R. (1996), Transalpini: Italienische Arbeitwanderung nach Sud-

deutschland im Kaiserreich 1870–1918, Studien zur historischen Migrations-

forschung 2 (Osnabruck: Rausch).

Desai , A. V. (1968), Real Wages in Germany, 1871–1914 (Oxford: Oxford

University Press).

D ickler , R. A. (1975), ‘Organisation and change in productivity in Eastern

Prussia’, in W. N. Parker and E. L. Jones (eds.), European Peasants and their

Markets (Princeton: Princeton University Press).

D illwitz , S. (1973), ‘Die Struktur der Bauernwirtschaft von 1871 bis 1914’,

Jahrbuch fur Geschichte, 9: 47–127.

Dodd , D. (1993), Historical Statistics of the United States: Two Centuries of the

Census, 1790–1990 (Westport: Greenwood).

Dollar , D. and Kraay , A. (2000), ‘Growth is Good for the Poor’, manu-

script, www.worldbank.org/research


www.worldbank.org/research

Dovring , F. (1956), Land and Labour in Europe, 1900–1950 (The Hague:

Martinus NijhoV).

Dumke , R. (1991), ‘Income inequality and industrialisation: the Kuznets curve

re-examined’, in Y. Brenner, H. Kaelble, and M. Thomas (eds.), Income

Distribution in Historical Perspective (Cambridge: Cambridge University

Press).

Dyer , B. (1896), ‘Green manuring’, Journal of the Royal Agricultural Society of

England and Wales, 773–9.

Edwardes , J. and Ogilvie , S. (1995), Universal Banks and German Industri-

alisation: A Reappraisal, CEPR Discussion Paper 1171.

Eggert , U. (1883), ‘Die Bewegung der Holzpreise und Tagelohn-Satze in

den preußischen Staatsforsten von 1800 bis 1879’, Zeitschrift des Koniglich

Preußischen Statistischen Bureaus, 23/1: 1–44.

Ehmer , J. (1991), Heiratsverhalten, Sozialstruktur, Okonomisches Wandel, Kri-

tische Studien zur Geschichtswissenschaft 92 (Gottingen: Vandenhoeck &

Rupprecht).

Eley , G. (1980), Reshaping the German Right (New Haven: Yale University

Press).

——(1986), From UniWcation to Nazism: Re-Interpreting the German Past

(Boston and London: Allen & Unwin).

—— , ‘German history and the contradictions of modernity: the bourgeoisie,

the state and the mastery of reform’, in G. Eley (ed.), Society, Culture, and the

State in Germany, 1870–1930 (Ann Arbor: University of Michigan Press),

43–103.

——(2002), Forging Democracy: The History of the Left in Europe 1850–2000


Engel , E. (1868), ‘Die Ergebnisse der Classensteuer, der classiWcierten Ein-

kommensteuer und der Mahl- und Sachlachtsteuer im preußischen Staate’,

Zeitschrift des Koniglich Preußischen Statistischen Bureaus, 1: 25–84.

——(1875), ‘Die Ergebnisse der Classensteuer und der classiWcierten Einkom-

mensteuer im preußischen Staate’, Zeitschrift des Koniglich Preußischen Statis-

tischen Bureaus, i. 108–48.

Engerman , S. L. and Sokoloff , K. L. (1997), ‘Factor endowments, insti-

tutions, and diVerential paths of growth among new world economies: a view

from economic historians of the United States’, in S. Haber (ed.), How Latin

America Fell Behind: Essays on the Economic Histories of Brazil and Mexico,

1800–1914 (Stanford: Stanford University Press).

Epkenhans , M. (1991), Die wilhelminischen Flottenrustung 1908–1914

(Munich: Oldenbourg).

Erickson , C. (1957), American Industry and the European Immigrant (Cam-

bridge, Mass.: Harvard University Press).

Evans , E. L. (1997), The German Centre Party 1870–1933: A Study in Political

Catholicism (Carbondale: Southern Illinois University Press).


Evans , R. (1997), Rereading German History: From UniWcation to ReuniWcation

(London: Routledge).

Evershed , H. (1884), ‘Improvements of the plants of the Weld’, Journal of the

Royal Agricultural Society of England and Wales, 77–113.

Fairbairn , B. (1992), ‘Interpreting Wilhelmine elections: national issues, fair-

ness issues and electoral mobilisation’, in E. L. Jones and J. Rettalack (eds.),

Elections, Mass Politics and Social Change in Modern Germany: New Perspec-

tives (Cambridge: Cambridge University Press).

——(1997), Democracy in the Undemocratic State: The German Reichstag Elec-

tions of 1898 and 1903 (Toronto: University of Toronto Press).

Fe i , J. E., Ranis , G. and Kuo , S. Y. (1979), Growth with Equity: The Taiwan

Case (World Bank and Oxford University Press).

Fe instein , C. H. (1972), National Income, Expenditure and Output of the

United Kingdom 1855–1965 (Cambridge: Cambridge University Press).

——(1988), ‘The rise and fall of the Williamson Curve’, Journal of Economic

History, 47/3: 699–729.

——(1998), ‘Pessimism perpetuated: real wages and the standard of living in

Britain during and after the Industrial Revolution’, Journal of Economic His-

tory, 58/3: 625–58.

——and Thomas , M. (2000), ‘A Plea for Errors’, manuscript, All Souls Col-

lege, Oxford.

Feldenkirchen , W. (1982), ‘Zur Kapitalbeschaftigung und Kapitalverwen-

dung bei Aktiengesellschaften des deutschen Maschinenbaus im 19. und

beginnenden 20. Jahrhundert’, Vierteljahresschrift fur Wirtschaftsgeschicht, 49:

38–74.

Ferenczi , I. and Willcox , W. F. (1929–31), International Migrations, 2 vols.

(New York: NBER).

Ferguson , N. (1994), ‘Public Wnance and national security: the domestic

origins of the First World War revisited’, Past and Present, 142: 141–68.

——(1998), The Pity of War (London: Allen Lane).

F inck von Finckenstein , H. W. Graf (1960), Die Entwicklung der Land-

wirtschaft in Preußen und Deutschland 1800–1930 (Wurzburg: Holzner).

F ircks , A. von (1879), ‘Ruckblick auf die Bewegung der Bevolkerung im

Preußischen Staat 1816–74’, Preu�ische Statistik, 48/1.

Flora , P., Kraus , F. and Pfennig , W. (1983), State, Economy and Society in

Western Europe 1815–1973: A Data Handbook, 2 vols. (Frankfurt am Main:

Campus).

Fohlin , C. (2002), ‘Corporate capital structure and the inXuence of universal

banks in pre-World War I Germany’, Jahrbuch fur Wirtschaftsgeschichte,

113–34.

Frankenstein , K. (1893), Die Arbeiterfrage in der deutschen Landwirtschaft

(Berlin: R. Oppenheim).


Frauendorfer , S. (1956), ‘Naturwissenschaften und Technik als dyna-

misches Moment’, Zeitschrift fur Agrargeschichte und Agrarsoziologie, 4/2:

113–27.

——(1957), Ideengeschichte der Agrarwirtschaft und Agrarpolitik im deutschen

Sprachgebiet, Band 1 (Munich: Bayerischer Landwirtschaftsverlag).

Fremdling , R. (1991), ‘Productivity comparisons between Great Britain and

Germany, 1855–1913’, Scandinavian Journal of Economic History, 39/1:

28–42.

Fussell , G. E. (1962), ‘The early days of chemical fertilisers’, Nature, 195:

750–4.

Galor , O. and Ts iddon , D. (1994), Human Capital Distribution, Techno-

logical Progress and Economic Growth, CEPR Discussion Paper no. 971

(London: CEPR).

Galos , A., Gentzen , F. and Jak okczyck , W. (1966), Die Hakatisten: Der

Deutsche Ostmarkverein 1894–1934 (Berlin: VEB).

Ge isenberger , S. and MUller , J. H. (1972), Die Einkommensstruktur in

verschiedenen deutschen Landern 1874–1914, Schriften zu Regional- und Ver-

kehrsproblemen 10 (Berlin).

Ge iss , I. (1990), ‘The German version of imperialism, 1898–1914: Weltpoli-

tik’, in G. Schollgen (ed.), Escape into War? The Foreign Policy of Imperial

Germany (Oxford: Berg).

Gerschenkron , A. (1943), Bread and Democracy in Germany (Berkeley: Uni-

versity of California Press).

——(1962), Economic Backwardness in Historical Perspective (Cambridge, Mass:

Harvard University Press).

Ghimure , K., ed. (2001), Land Reform and Peasant Livelihoods: The Social

Dynamics of Rural Poverty and Agrarian Reform in Developing Countries, UN

Institute for Social Development (New York: UN).

Goldin , C. (1994), ‘The political economy of immigration restriction in the

United States, 1890–1921’, in C. Goldin and G. Libecap (eds.), The Regu-

lated Economy: A Historical Approach (Chicago: University of Chicago Press).

Golding , F. (1930), ‘Die Wanderbewegung in Ostpreu�en seit der Jahrhun-

dertwende mit besonderer Berucksichtigung der Abwanderung vom Lande’,

Zeitschrift des Preußischen Statistischen Landesamts, 69/2: 200–34.

Goltz , T. Frhr . von der (1875), Die Lage der landlichen Arbeiter im

deutschen Reich (Berlin: P. Parey).

——(1893), Die landliche Arbeiterklasse und der preu�ische Staat (Jena: G.

Fischer).

Gommel , R. (1979), Realeinkommen in Deutschland: ein internationaler Vergleich

(Nurnberg: Im Selbstverlag).

Grant , O. (2000), ‘Internal Migration in Germany, 1870–1913’, Oxford

D.Phil. thesis.


Grant , D. (2002a), Productivity in German Agriculture: Estimates of Agricul-

tural Productivity from Regional Accounts for 21 German Regions: 1880/4,

1893/7 and 1905/9, Oxford University Discussion Paper in Economic and

Social History, available at http://www.nuV.ox.ac.uk/ Economics/History/

——(2002b), Does Industrialisation Push Up Inequality? New Evidence on the

Kuznets Curve from Nineteenth Century Prussian Tax Statistics, Oxford

University Discussion Paper in Economic and Social History, available at

http://www.nuV.ox.ac.uk/Economics/History/

——(2002c), ‘Max Weber and Die Lage der Landarbeiter im ostelbischen Deutsch-

land: a statistical examination’, forthcoming in the Jahrbuch fur Wirtschafts-

geschichte, 2: 61–84.

——(2002d ), ‘Explaining the Decline of German Emigration to the United

States after 1895’, manuscript, NuYeld College, Oxford.

——(2002e), ‘ ‘‘Few Better Farmers in Europe?’’ Productivity, Change and

Modernisation in East-Elbian Agriculture 1870–1913’, manuscript, NuYeld

College, Oxford.

Greene , W. H. (2000), Econometric Analysis, 4th edn. (New Jersey: Prentice-

Hall).

Gregory , P. R. (1997), ‘Searching for consistency in historical data: alterna-

tive estimates of Russia’s industrial production, 1887–1913’, Journal of Eco-

nomic History, 57/1: 196–202.

Grimmer-Solem , E. (1999), ‘The Science of Progress: The Rise of Historical

Economics and Social Reform in Germany, 1864–1894’, Oxford D.Phil.

thesis.

Groh , D. (1973), Negative Integration und revolutionarer Attentismus: Die

deutsche Sozialdemokratie am Vorabend des Ersten Weltkrieges (Frankfurt am

Main: Propylaen).

Grumbach , F. and Konig , H. (1957), ‘Beschaftigung und Lohne der

deutschen Industriewirtschaft 1888–1954’, Weltwirtschaftsarchiv, 79: 125–54.

Guttsman , W. (1981), The German Social Democratic Party 1875–1933

(London: Allen & Unwin).

Habakkuk , J. (1958), ‘Population growth and economic development’, in

J. Habakkuk et al., Lectures on Economic Development (Istanbul: Sermet Mat-

baash).

Haberle , R. (1938), ‘German approaches to internal migration’, in D. Thomas

(ed.), Research Memorandum on Migration DiVerentials (New York: Social

Science Research Council).

Hagen , W. (1980), Germans, Poles and Jews: The Nationality ConXict in the

Prussian East, 1772–1914 (Chicago: University of Chicago Press).

Haines , M. (1977), Economic–Demographic Interrelations in Developing Agricul-

tural Regions: A Case Study of Prussian Upper Silesia, 1840–1914 (New York:

Arno).


http://www.nuff.ox.ac.uk/Economics/History/

http://www.nuff.ox.ac.uk/Economics/History/

——(1979), Fertility and Occupation: Population Patterns in Industrialisation

(New York: Academic).

Hajnal , J. (1965), ‘European marriage patterns in perspective’, in D. Glass

and D. Eversley (eds.), Population in History (Chicago: Aldine).

Hanley , S. B. and Yamamura , K. (1977), Economic and Demographic Change

in Pre-Industrial Japan, 1600–1868 (Princeton: Princeton University Press).

Hardach , K. W. (1967), Die Bedeutung wirtschaftlicher Faktoren bei der Wie-

dereinfuhrung der Eisen- und Getreidezolle in Deutschland 1879 (Berlin:

Duncker & Humblot).

Harnisch , H. (1977), ‘Die Bedeutung der kapitalistischen Agrarreform fur

die Herausbildung des Inneren Markte und die industrielle Revolution in der

ostlichen Provinzen Preußens in der ersten Halfte des 19. Jahrhunderts’,

Jahrbuch fur Wirtschaftgeschichte, 63–82.

——(1984), Kapitalistische Agrarreform und Industrielle Revolution (Weimar: H.

Bohlhaus).

——(1986), ‘The background to the agricultural reforms in feudal Prussia east

of the Elbe 1760–1807’, in R. J. Evans and W. R. Lee (eds.), The German

Peasantry (London: Croom Helm).

Hatton , T. J. (1993), A Model of UK Emigration 1870–1913, CEPR Discus-

sion Paper no. 771.

——(1997), ‘The immigrant assimilation puzzle in late nineteenth century

America’, Journal of Economic History 57: 34–62.

——and Will iamson , J. G., eds. (1994), Migration and the International

Labour Market, 1850–1939 (London: Routledge).

——and ——(1996), ‘What drove the mass migrations from Europe in the

nineteenth century?’ Population and Development Review, 20: 533–59.

——and ——(1998), The Age of Mass Migration: Causes and Economic Impact


Hatzius , J. (1994), Regional Migration, Unemployment and Vacancies: Evidence

from West German Microdata, Oxford, Institute of Economics and Statistics,

Applied Economics Discussion Paper no. 164.

Hauser , A. (1965), ‘Schweizer Bauern als Kolonisten in Preußen und

Litauen’, Zeitschrift fur Agrargeschichte und Agrarsoziologie, 13: 212–19.

Haushofer , H. (1957), Die deutsche Landwirtschaft im technische Zeitalter,

vol. 5 of G. Franz (ed.), Deutsche Agrargeschichte (Stuttgart: Ulmer).

Hayami , Y. (1997) Development Economics: From the Poverty to the Wealth of

Nations (Oxford: Oxford University Press).

Helferrich , K. (1915), Deutschlands Volkswohlstand (Berlin: G. Stilbe).

Helling , G. (1966), ‘Zur Entwicklung der Produktivitat in der deutschen

Landwirtschaft im 19. Jahrhundert’, Jahrbuch fur Wirtschaftsgeschichte,

129–41.

Hempel , L. (1957), ‘Heuerlingswesen und Croftersystem’, Zeitschrift fur

Agrargeschichte und Agrarsoziologie, 5/2: 169–79.


Henning , F. W. (1977), ‘Produktionskosten und Preisbildung fur Getreide in

den letzten Jahrzehnten des 19. Jahrhunderts’, Zeitschrift fur Agrargeschichte

und Agrarsoziologie, 25: 214–36.

Hennock , E. P. (2000), ‘The urban sanitary movement in England and Ger-

many, 1838–1914: a comparison’, Continuity and Change, 15/2: 269–96.

Hentschel , V. (1978), Wirtschaft und Wirtschaftspolitik im wilhelminischen

Deutschland (Stuttgart: Klett-Cotta).

——(1979), ‘Erwerbs- und Einkommensverhaltnisse in Sachsen, Baden und

Wurttemberg vor dem Ersten Weltkrieg’, Vierteljahrschrift fur Sozial- und

Wirtschaftsgeschichte, 66/1: 26–73.

Herbert , U. (1990), A History of Foreign Labour in Germany 1880–1980 (Ann

Arbor: University of Michigan Press).

Herwig , H. (1987), ‘Luxury Fleet’: The Imperial German Navy 1888–1918

(London: AshWeld).

Hess , K. (1994), ‘Zur wirtschaftlichen Lage der Gro�agrarier im ostelbischen

Preu�en 1867/71 bis 1914’, in H. Reif (ed.), Ostelbische Agrargesellschaft im

Kaiserreich und in der Weimarer Republik (Berlin: Akademie Verlag).

H iggins , M. and Will iamson , J. G. (1999), ‘Explaining Inequality the

World Round: Cohort Size, Kuznets Curves and Openness’, manuscript,

Federal Reserve Bank of New York and Harvard University.

H iggs , R. (1971), ‘Race, skill and earnings: American immigrants in 1909’,

Journal of Economic History, 31: 420–9.

H ildebrand , K. (1989), Deutsche Au�enpolitik, 1871–1918 (Munich: Olden-

bourg).

——(1995), Reich–Nation-State–Great Power: ReXections on German Foreign

Policy 1871–1945 (London: German Historical Institute).

Hobson , J. M. (1997) The Wealth of States: A Comparative Sociology of Inter-

national Economic and Political Change (Cambridge: Cambridge University

Press).

Hochstadt , S. (1996), ‘The socio-economic determinants of increasing mo-

bility in nineteenth century Germany’, in D. Hoerder and L. P. Moch (eds.),

European Migrants (Boston: Northeastern Press).

——(1999), Mobility and Modernity: Migration in Germany 1820–1989 (Ann

Arbor: University of Michigan Press).

Hoerder , D. and Nagler , J., eds. (1995), People in Transit (Cambridge:

Cambridge University Press).

Hoffmann , W. (1963), ‘The take-oV in Germany’, in W. Rostow (ed.), The

Economics of Take-OV into Sustained Growth (London: Macmillan).

——(1965), Das Wachstum der Deutschen Wirtschaft seit der Mittel des 19. Jahr-

hunderts (Berlin: Springer Verlag).

Hohls , R. (1995), ‘Germany’, in P. Scholliers and V. Zamagni (eds.), Labour’s

Reward (Aldershot: E. Elgar).


Hohorst , G. (1977), Wirtschaftswachstum und Bevolkerungsentwicklung in

Preußen 1816 bis 1914 (Munster: University of Munster).

Honigsheim , P. (1949), ‘Max Weber as historian of agriculture and rural life’,

Agricultural History, 23/3: 179–212.

Hourwich , I. (1922), Immigration and Labour: The Economic Aspects of Euro-

pean Immigration to the United States (New York: Huebsch).

Hunt , J. C. (1974), ‘Peasants, grain tariVs and meat quotas’, Central European

History, 7: 311–31.

Hutchinson , E. P. (1956), Immigrants and their Children 1850–1950 (New

York: Wiley).

Institut International d’Agriculture (1915), Annuaire International de Statistique

Agricole 1913/14.

Jackman , R. and Savouri , S. (1992), ‘Regional migration in Britain: an

analysis of gross Xows using NHS central register data’, Economic Journal,

102: 1433–50.

Jackson , J. J. (1997), Migration and Urbanisation in the Ruhr Valley 1821–1914

(New Jersey: Humanities Press).

Jarausch , K. (1969), ‘The illusion of limited war: Chancellor Bethmann Holl-

weg’s calculated risk’, Central European History, 48–76.

Jazairy , I., Alamgir , M. and Panuccio , T. (1992), The State of World

Rural Poverty: An Enquiry into its Causes and Consequences, International

Fund for Agricultural Development.

Jeck , A. (1970), Wachstum und Verteilung des Volkseinkommens (Tubingen:

Mohr).

Jeidels , O. (1905), Das Verhaltnis der deutschen Gro�banken zur Industrie,

Staats- und Sozialwissenschaftliche Forschungen, 24.

Jenks , J. W. and Lauck , W. J. (1926), The Immigration Problem: A Study of

American Immigration Conditions and Needs (New York: Huebsch).

Kaelble , H. (1976), ‘Social stratiWcation in Germany in the 19th and 20th

centuries: a survey of research since 1945’, Journal of Social History, 10:

144–65.

——(1986), Industrialisation and Social Inequality in Nineteenth Century Europe

(Leamington Spa: Berg).

——and Federspiel , R. (1990), Soziale Mobilitat Berlin 1825–1957, Histor-

ische Statistik von Deutschland, vol. X.

——and Hohls , R. (1989), Die Regionale Erwerbsstruktur im Deutschen Reich

und in der Bundesrepublik 1895–1970, Quellen und Forschungen zur histor-

ischen Statistik von Deutschland.

Kaerger , K. (1890), ‘Die Sachsengangerei’, Landwirtschaftliche Jahrbucher, 19:

237–522.

Kaiserlich Statistisches Amt (1916), Die Deutschen im Ausland und die Auslander

im Deutschen Reich, Erganzungsheft zur Vierteljahrsheft zur Statistik des

Deutschen Reichs 4.


Kaldor , N. (1956), ‘Alternative theories of distribution’, Review of Economic

Studies, 23/2: 94–100.

Kamphoefner , W. D. (1986), ‘At the crossroads of economic development:

background factors aVecting emigration from nineteenth-century Germany’,

in I. A. Glazier and L. De Rosa (eds.), Migration across Time and Nations

(New York: Holmes and Meier).

——(1987), The Westfalians: From Germany to Missouri (Princeton: Princeton

University Press).

——(1999), ‘Deutschen-Amerikaner: Musterknaben der Einwanderung?’,

IMIS-Beitrage, 12.

Kehr , E. (1975), Battleship Building and Party Politics in Germany (Chicago:

University of Chicago Press).

——(1977), Economic Interest, Militarism and Foreign Policy (Berkeley: Univer-

sity of California Press).

Kennedy , P. (1980), The Rise of the Anglo-German Antagonism, 1860–1914


Keyder , C. and O’Brien , P. (1978), Economic growth in Britain and France,

1780–1914: Two Paths to the Twentieth Century (London: Allen & Unwin).

Keyser , E. (1959), ‘Die Ansiedlung von Schwaben in Westpreu�en

1776–1805’, Zeitschrift fur Agrargeschichte und Agrarsoziologie, 7/2: 169–76.

K i esewetter , H. (1988), Industrialisierung und Landwirtschaft: Sachsens Stel-

lung im regionalen Industrialisierungsprozess Deutschlands im 19. Jahrhundert

(Cologne: Bohlau).

K indleberger , C. (1967), Europe’s Postwar Growth: The Role of Labour

Supply (Cambridge, Mass.: Harvard University Press).

Klemm , V. (1994), ‘Die Agrarwissenschaften und die Modernisierung der

Gutsbriebe in Ost-und Mitteldeutschland’, in H. Reif (ed.), Ostelbische Agrar-

gesellschaft im Kaiserreich und in der Weimarer Republik (Berlin: Akademie

Verlag).

Knapp , G. F. (1887), Die Bauernbefreiung und der Ursprung der Landarbeiter in

den alteren Theilen Preussens (Leipzig: Duncker & Humblot).

——(1891), Die Landarbeiter bei der Stein-Hardenbergischen Gesetzgebung

(Strassburg: Heitz and Mundel).

Knodel , J. (1974), The Decline of Fertility in Germany 1871–1939 (Princeton:

Princeton University Press).

——(1988), Demographic Behaviour in the Past (Cambridge: Cambridge Univer-

sity Press).

Knoke , A. (1911), Auslandische Wanderarbeiter in Deutschland, Diss., Leipzig.

Kocka , J., ed. (1975), ‘Soziale Schichtung und Mobilitat in Deutschland im

19. und 20. Jahrhundert’, Geschichte und Gesellschaft 1/1 (Gottingen: Van-

denhoeck und Ruprecht).

——(1990), Arbeitsverhaltnisse und Arbeiterexistenzen: Grundlagen der Klassenbil-

dung im 19. Jahrhundert (Bonn: Dietz).


Kolakowski , L. (1978), Main Currents of Marxism, 3 vols. (Oxford: Oxford

University Press).

Kollmann , W. (1959), ‘Industrialisierung, Binnenwanderung und ‘‘Soziale

Frage’’ ’, Vierteljahrschrift fur Sozial- und Wirtschaftsgeschichte.

——(1960), Sozialgeschichte der Stadt Barmen im 19. Jahrhundert (Tubingen:

J. C. B. Mohr).

——(1964), ‘The population of Germany in the age of industrialisation’, in

H. Moller (ed.), Population Movements in Modern European History (London:

Collier-Macmillan).

——(1974), ‘Demographische ‘‘Konsequenzen’’ der Industrialisierung in

Preu�en’, in W. Kollmann (ed.), Bevolkerung in der Industriellen Revolution

(Gottingen: Vandenhoeck und Ruprecht).

——(1976), ‘Bevolkerungsgeschichte 1800–1870’, in H. Aubin and W. Zorn

(eds.), Handbuch der deutschen Wirtschaft- und Sozialgeschichte, vol. 2 (Stutt-

gart: Klett-Cotta).

——(1980), Quellen zur Bevolkerungsgeschichte Deutschlands, vol. 1 of Quellen

zur Bevolkerungs-, Sozial- und Wirtschaftsstatistik Deutschlands (Boppard:

Boldt).

——and Marschalck , P. (1973), ‘German emigration’, Perspectives in Ameri-

can History, 7: 499–554.

Komlos , J. (1995), The Biological Standard of Living in Europe and America

1700–1900: Studies in Anthropometric History (Aldershot: Variorum).

Konitz , C. von (1895), ‘Uber das Verhaltnisse zwischen Rohertrag und Rein-

ertrag in landwirtschaftlichen Betrieben verschiedener Wirtschaftssysteme’,

Landwirtschaftliche Jahrbucher 24: 309–444.

Koselleck , R. (1967), Preu�en zwischen Reform und Revolution (Stuttgart:

Klett-Cotta).

Kosters-Kraft , M. (1998), Wanderarbeit und Gro�baustelle: Niederlander

beim Bau des Dortmund-Ems-Kanals 1892–1900, Studien zur historischen

Migrationsforschung 8 (Osnabruck: Rausch).

Kuczynski , J. (1945), A Short History of Labour Conditions in Germany, 1800

to the present day (London: F. Muller).

Kuczynski , R. (1897), Der Zug nach der Stadt (Stuttgart: J. G. Cotta).

——(1908), Einkommen und Miete (Berlin: Berlin Statistische Amt).

——(1913), Arbeitslohn und Arbeitszeit in Europa und Amerika (Berlin:

J. Springer).

Kuhnert , F. (1911), ‘Die Ergebnisse der preu�ischen Einkommens- und

Erganzungssteuer: Veranlagung nach Stadtegruppen fur das Jahr 1908’, Zeits-

chrift des Koniglich Preu�ischen Statistischen Bureaus, 1–54.

——(1916), ‘Einkommensgliederung der preu�ischen Bevolkerung im Zei-

traum 1902–1911’, Zeitschrift des Koniglich Preu�ischen Statistischen Bureaus,

269–309.


Kuznets , S. (1955), ‘Economic growth and income inequality’, American Eco-

nomic Review, 45: 1–28.

Laffont , J.-J. and Matouss i , M. (1995), ‘Moral hazard, Wnancial con-

straints and share-cropping in El Oulja’, Review of Economic Studies, 62/3:

381–99.

Lambi , I. N. (1984), The Navy and German Power Politics 1862–1914 (Boston:

Allen & Unwin).

——(1963), ‘Free trade and protection in Germany 1868–1879’, Vierteljahrs-

chrift fur Sozial- und Wirtschaftsgeschichte, Part 44.

Laux , H.-D. (1989), ‘The components of population growth in Prussian cities,

1875–1905, and their inXuence on urban population structure’, in R. Lawton

and W. R. Lee (eds.), Urban Population Development in Western Europe from

the Late Eighteenth to the Early Twentieth Century (Liverpool: Liverpool Uni-

versity Press).

Lebovics , H. (1967), ‘ ‘‘Agrarians’’ versus ‘‘industrializers’’: Social Conserva-

tive resistance to industrialism and capitalism in late nineteenth century Ger-

many’, International Review of Social History, 12: 31–65.

Lee , J. Z. and Wang , F. (1999), One Quarter of Humanity: Malthusian Myth-

ology and Chinese Realities, 1700–2000 (Cambridge, Mass.: Harvard Univer-

sity Press).

Lee , W. and Marschalck , P. (2000), ‘Demographic change and industrial-

isation in Germany, 1815–1914: Bremen in comparative perspective’, History

of the Family, 5/4: 373–90.

Lee , W. R. (1977), Population Growth, Economic Development and Social Change

in Bavaria 1750–1850 (New York: Arno).

——ed. (1979), European Demography and Economic Growth (London: Croom

Helm).

——(1979a), ‘Regionale DiVerenzierung im Bevolkerungswachstum Deutsch-

lands im fruhen 19. Jahrhundert’, in R. Fremdling and R. H. Tilly (eds.),

Industrialisierung und Raum (Stuttgart: E. Klett).

——(1980), ‘The mechanism of mortality change in Germany 1750–1850’,

Medizinhistorisches Journal, 15.

——(1981), ‘Family and ‘‘modernisation’’: the peasant family and social change

in nineteenth century bavaria’, in R. Evans and W. R. Lee (eds.), The German

Family: Essays on the Social History of the Family in Nineteenth and Twentieth

Century Germany (London: Croom Helm).

Lees , A. (1979), ‘Critics of urban society in Germany, 1854–1914’, Journal of

the History of Ideas, 40: 61–83.

Leps ius , R. (1966), ‘Parteinsystem und Sozialstruktur. Zum Problem der

Demokratisierung der deutschen Gesellschaft’, in W. Abel (ed.), Wirtschaft,

Geschichte und Wirtschaftsgeschichte: Festschrift zum 65. Geburtstag von Fried-

rich Lutge (Stuttgart: G. Fischer).


Leunig , T. and Voth , H.-J. (1996), ‘Did smallpox reduce height? Stature

and the standard of living in London, 1770–1873’, Economic History Review,

49/3: 541–60.

Levy , H. (1911), Large and Small Holdings: A Study in English Agricultural

Economics (Cambridge: Cambridge University Press).

Lewis , W. A. (1954), ‘Development with unlimited supplies of labour’, Man-

chester School of Economics and Social Studies, 20: 139–92.

L idtke , V. (1985), The Alternative Culture: Socialist Labour in Imperial Ger-

many (New York: Oxford University Press).

L indenlaub , J. and Kohne-Lindenlaub , R. (1991), ‘Business Wnancing by

Krupp 1811–1848’, German Yearbook on Business History, 9–86.

L indert , P. (1994) ‘Unequal living standards’, in R. Floud and D. McCloskey

(eds.), The Economic History of Britain: Volume One: 1700–1860 (Cambridge:

Cambridge University Press).

——and Will iamson , J. G. (1980) American Inequality (New York: Aca-

demic Press).

L ipton , M. (1980), ‘Migration from rural areas of poor countries: their impact

on rural productivity and income distribution’, World Development, 8: 1–24.

Lourens , P. and Lucassen , J. (1999), Arbeitswanderung und beruXiche Spe-

zialisierung: die lippischen Ziegler im 18. und 19. Jahrhunderts, Studien zur

historischen Migrationsforschung 6 (Osnabruck: Rausch).

Low , D. (1996), The Egalitarian Moment: Asia and Africa 1950–1980 (Cam-

bridge: Cambridge University Press).

Lubinski , A. (1995), Entlassen aus dem Untertanenverband: Die Amerikaaus-

wanderung aus Mecklenburg-Strelitz (Osnabruck: Rausch).

LU tge , F. (1963), Geschichte der deutschen Agrarfassung von fruhen Mittelalter

bis zum 19. Jahrhundert (Stuttgart: E. Ulmer).

McGouldrick , P. F. and Tannen , M. B. (1977), ‘Did American manufac-

turers discriminate against immigrants before 1914?’, Journal of Economic

History, 37: 723–46.

Maddison , A. (1991), Dynamic Forces in Capitalist Development (Oxford:


——(1994), ‘Explaining the economic performance of nations, 1820–1989’, in

W. Baumol et al. (eds.), Convergence of Productivity (Oxford: Oxford Univer-

sity Press).

——(1995), Monitoring the World Economy, 1820–1992 (Paris: OECD).

Mantl , E. (1997), Heirat als Privileg: obrigkeitliche Heiratsbeschrankungen in

Tirol und Vorarlberg 1820 bis 1920 (Vienna: Verlag fur Geschichte und Poli-

tik).

Marschalck , P. (1973), Deutsche Uberseewanderung im 19. Jahrhundert (Stutt-

gart: Klett).


Marschalck , P. (1987), ‘The age of demographic transition: mortality and

fertility’, in K. Bade (ed.), Population, Labour and Migration in 19th and 20th

Century Germany (Leamington Spa: Berg).

Massey , D. S. (1988), ‘Economic development and international migration in

comparative perspective’, Population and Development Review, 14: 383–413.

Matthews , R., Fe instein , C. and Odling-Smee , J. (1982), British Eco-

nomic Growth 1856–1973 (Oxford: Oxford University Press).

Matz , K.-J. (1980), Pauperismus und Bevolkerung: Die gesetzlichen Ehebeschran-

kerungen in der suddeutschen Staaten wahrend des 19. Jahrhunderts (Stuttgart:

Klett-Cotta).

Matzerath , H. (1985), Urbanisierung in Preu�en 1815–1914 (Stuttgart: W.

Kohlhammer).

Me isel , F. (1911), ‘Moral und Technik bei der Veranlagung der preußischen

Einkommensteuer’, Jahrbuch fur Gesetzgebung, Verwaltung und Volkswirtschaft

im Deutschen Reich, 35/1: 285–374.

Me itzen , A. (1868), Der Boden und der landwirtschaftlichen Verhaltnisse des

Preußischen Staates (Berlin: Finanzministerium).

——(1885), ‘Ermittelungen uber die durchschnittliche Hohe der Grund-

buchschulden der bauerlichen Besitzungen in 52 Amtsgerichtsbezirken der

Preußischen Staat nach dem Stand des Jahres 1883’, Landwirtschaftliche Jahr-

bucher, 14/2: 1–32.

Meyer , F. (1938), ‘German internal migration statistics: methods, sources and

data’, in D. Thomas (ed.), Research Memorandum on Migration DiVerentials

(New York: Social Science Research Council).

M iaskowski , A. von (1882), ‘Das Erbrecht und die Grundeigentumvertei-

lung im deutschen Reich’, Schriften des Vereins fur Sozialpolitik, 20.

M ilward , A. and Saul , S. B. (1977), The Development of the Economies of

Continental Europe, 1850–1914 (London: Allen & Unwin).

Ministry of Agricul ture , Fisher ies and Food (1968 ) , A Century of

Agricultura l Stat is t ics : Great Bri ta in 1866–1966 (London:

HMSO).

M itchell , B. R. (1981), European Historical Statistics 1750–1975, 2nd edn.

(London: Macmillan).

M ittler , E. S. (1929), Untersuchungen uber Landarbeiterverhaltnisse, Berichte

uber Landwirtschaft (Berlin: P. Parey).

Mokyr , J. (1990), The Lever of Riches: Technological Creativity and Economic

Progress (Oxford: Oxford University Press).

Mommsen , W. (1995), Imperial Germany: Politics, Culture and Society in an

Authoritarian State (London: Arnold).

Nadobnik , M. (1908), ‘Die Abnahme des durchschnittlichen Heiratsalters in

Deutschland’, Zeitschrift des Koniglichen Preussischen Landesamts, 48: 69–84.

Neuburger , H. and Stokes , H. H. (1974), ‘German banks and German

growth, 1880–1913’, Journal of Economic History, 34: 710–31.


Neuhaus , G. (1904), ‘Die ortsublichen Tagelohne gewohnlicher Tagearbeiter

in Preussen 1892 und 1901’, Zeitschrift des Koniglich Preu�ischen Statistischen

Bureaus, 310–46.

Neumann , R. P. (1978), ‘Working class birth control in Wilhelmine Ger-

many’, Comparative Studies in Society and History, 20: 408–28.

Newbery , D. (1977), ‘Risk sharing, share-cropping and uncertain labour

markets’, Review of Economic Studies, 23: 187–207.

——(1989), ‘Agricultural institutions for insurance and stabilization’, in

P. Bardhan (ed.), The Economic Theory of Agrarian Institutions (Oxford:


N ichtweiss , J. (1959), Die auslandischen Saisonarbeiter in der Landwirtschaft

der ostlichen und mittleren Gebiete des Deutschen Reichs: Ein Beitrag zur

Geschichte des preußisch-deutsch Politik von 1890 bis 1914 (Berlin: Rutten und

Loening).

N ickell , S. (1996), ‘The low-skill low-pay problem: lessons from Germany

for Britain and the US’, Policy Studies, 17: 17–21.

Nove , A. (1969), An Economic History of the U.S.S.R. (London: Allen Lane).

Nugent , W. (1992), Crossings: The Great Transatlantic Migrations, 1870–1914

(Bloomington: Indiana University Press).

Nurkse , R. (1953), Problems of Capital Formation in Underdeveloped Countries

(Oxford: Basil Blackwell).

O’Brien , P. and Prados de la Escosura , L. (1992), ‘Agricultural prod-

uctivity and European industrialisation’, Economic History Review, 45:

514–36.

O’Rourke , K. (1997) ‘The European grain invasions, 1870–1913’, Journal of

Economic History, 57/4: 775–801.

——Taylor , A. M. and Will iamson , J. G. (1996), ‘Factor price conver-

gence in the late nineteenth century’, International Economic Review, 37/3:

499–530.

——Will iamson , J. G. and Hatton , T. (1994), ‘Mass migration, commod-

ity market integration and real wage convergence’, in T. J. Hatton and J. G.

Williamson (eds.), Migration and the International Labour Market, 1850–1939

(London: Routledge).

Oberpennig , H. (1998), Migration und Fernhandel im ‘Todden-System’: Wan-

derhandler aus dem nordlichen Munsterland im mittleren und nordlichen Europa

des 18. und 19. Jahrhunderts (Osnabruck: Rausch).

Offer , A. (1989), The First World War: An Agrarian Interpretation (Oxford:


Olson , M. (1982), The Rise and Decline of Nations: Economic Growth, StagXa-

tion and Social Rigidities (New Haven: Yale University Press).

Oppenheimer , F. (1898), Gro�grundeigentum und soziale Frage (Berlin:

VITA).


Orsagh , T. (1969), ‘Lohne in Deutschland 1871–1913: neuere Literatur’,

Zeitschrift fur die gesamte Staatswissenschaft, 125: 476–83.

Paasche , H. (1883), ‘Die rechtliche und wirtschaftliche Lage des Bauern-

standes in Mecklenburg-Schwerin’, in Bauerliche Zustand in Deutschland,

vol. 3, Schriften des Verein fur Sozialpolitik, 24.

Pas inetti , L. (1962), ‘Rate of proWt and income distribution in relation to the

rate of economic growth’, Review of Economic Studies, 29: 267–79.

Perkins , W. A. (1981), ‘The agricultural revolution in Germany’, Journal of

Economic History, 10/1: 71–118.

——(1984), ‘The German agricultural worker, 1815–1914’, Journal of Peasant

Studies, 11/3: 3–27.

——(1986), ‘Dualism in German agrarian historiography’, Comparative Studies

in Society and History, 28/2: 287–306.

Perrson , T. and Tabellini , G. (1994), ‘Is inequality harmful for growth?’,

American Economic Review, 84/3: 600–21.

Peterson , C. (1995), Bread and the British Economy: c.1770–1870 (Aldershot:

Scholar).

Phelps Brown , E. H. and Browne , M. (1968), A Century of Pay (London:

Macmillan).

Plaul , R. (1986), ‘The rural proletariat: the everyday life of rural labourers in

the Magdeburg region, 1830–80’, in R. J. Evans and W. R. Lee (eds.), The

German Peasantry (London: Croom Helm).

Pogge von Strandmann , H. (1972), ‘Nationale Verbande zwischen Welt-

politik und Kontinentalpoltik’, in H. Schottelius and W. Deist (eds.), Marine

und Marinepolitik im kaiserlichen Deutschland 1871–1914 (Dusseldorf: Droste).

——(1988), ‘Germany and the coming of war’, in R. Evans and H. Pogge von

Strandmann (eds.), The Coming of the First World War (Oxford: Oxford

University Press).

——ed. (1988a), Walter Rathenau: Industrialist, Banker, Intellectual and Polit-

ician: Notes and Diaries 1907–1922 (Oxford: Clarendon).

——(1992), ‘The Liberal power monopoly in the cities of Imperial Germany’,

in E. L. Jones and J. Rettalack (eds.), Elections, Mass Politics and Social

Change in Modern Germany: New Perspectives (Cambridge: Cambridge Uni-

versity Press).

——(1999), ‘Revolutionsfurcht, Sozialreformen und wirtschaftliches Gleichge-

wicht: Anmerkungen zur Thunenrezeption’, in O. N. Haberl and T. Korenke

(eds.), Politische Deutungskulturen: Festschrift fur Karl Rohe (Baden: Nomos).

——and Geiss , I. (1965), Die Erforderlichkeit des Unmoglichen: Deutschland am

Vorabend des Ersten Weltkrieges (Frankfurt am Main: Europaische Verlagsan-

stalt).

Pohl , E. (1908), Die Lohn und Wirtschaftsverhaltnisse der Landarbeiter in Masu-

ren in den letzten Jahrzehnten, Diss., Konigsberg.


——(1984), ‘Forms and phases of industrial Wnance up to the Second World

War’, German Yearbook on Business History, 75–94.

Pohl , M. (1976), Einfuhrung in die deutsche Bankengeschichte (Frankfurt am

Main: Knapp).

——(1986), Entstehung und Entwicklung des Universalbankensystems (Frankfurt

am Main: Knapp).

Pollard , S., ed. (1980), Region und Industrialisierung (Gottingen: Vanden-

hoeck und Ruprecht).

——(1981), Peaceful Conquest: The Industrialization of Europe 1760–1970


Prados de la Escosura , L. (2000), ‘International comparisons of real prod-

uct: an alternative data set’, Explorations in Economic History, 37/1: 1–41.

Prest , A. R. (1948), War Economies of Primary Producing Countries (Cam-

bridge: Cambridge University Press).

Prokopovitch , S. (1926), ‘The distribution of national income’, Economic

Journal, 36: 69–82.

Prosterman , R. and Ri edinger , J. (1987), Land Reform and Democratic

Development (Baltimore: Johns Hopkins University Press).

Puhle , H.-J. (1972), Von der Agrarkrise zum Prafaschismus (Wiesbaden:

F. Steiner).

——(1976), ‘Parlament, Parteien und Interessenverbande 1890–1914’, in M.

Sturmer (ed.), Das Kaiserliche Deutschland (Dusseldorf: Droste Verlag).

——(1986) ‘Peasants and Lords in the Kaiserreich’, in R. Moeller (ed.), Peas-

ants and Land in Modern Germany: Recent Studies in Agricultural History


Quante , P. (1933), ‘Die Flucht aus der Landwirtschaft’, Zeitschrift des Koni-

glich Preu�ischen Statistischen Bureaus, 3: 277–380.

——(1958), Die Abwanderung aus der Landwirtschaft, Kieler Studien 48.

——(1959), ‘Die Bevolkerungsentwicklung der preu�ischen Ostprovinzen im

19. und 20. Jahrhundert’, Zeitschrift fur Ostforschung, 8/4: 481–99.

Radetzki , W. (1930), Die inlandischen landwirtschaftlichen Wanderarbeiter

Deutschlands, Arbeiten aus dem Institut fur Wirtschaftslehre der Universitat

Breslau.

Ranis , G. and Schultz , T. P. (1988), The State of Development Economics:

Progress and Perspectives (Oxford: Basil Blackwell).

Rauchberg , H. (1900), ‘Die Landwirtschaft im deutschen Reich’, Archiv fur

soziale Gesetzgebung und Statistik, 15: 554–97.

Rauh , M. (1973), Foderalismus und Parlamentarismus im Wilhelminischen Reich

(Dusseldorf: Droste).

——(1977), Die Parlamentarisierung des Deutschen Reiches (Dusseldorf: Droste

Verlag).


Ravallion , M. and Chen , S. (1997), ‘What can new survey data tell us

about recent changes in distribution and poverty?’, World Bank Economic

Review, 11: 357–82.

Rehbein , F. (1911), Das Leben eines Landarbeiters (Jena: E. Diederichs).

Re ich , U. (1998), Aus Cottbus und Arnswalde in die Neue Welt: Amerikaauswan-

derung aus Ostelbien im 19. Jahrhundert (Osnabruck: Rausch).

Re if , H., ed. (1994), Ostelbische Agrargesellschaft im Kaiserreich und in der Wei-

marer Republik (Berlin: Akademie Verlag).

Retallack , J. (1988), Notables of the Right (Boston: Unwin Hyman).

Reynolds , L. G., ed. (1975) Agriculture in Development Theory (New Haven:

Yale University Press).

R ichardson , H. W. (1965), ‘Over-commitment in Britain before 1930’,

Oxford Economic Papers, 17: 237–62.

R iehl , W. H. (1854), Die Naturgeschichte des Volkes als Grundlage einer

deutschen Sozial-Politik, vol. 1: Land und Leute (Stuttgart: J. G. Cotta).

R i esser , J. (1911), The German Great Banks (Washington: National Monetary

Commission).

R itter , G. (1980), Wahlgeschichtliches Arbeitsbuch: Materialien zur Statistik des

Kaiserreichs 1871–1918 (Munich: C. A. Beck).

——and Tenfelde , K. (1992), Arbeiter im Deutschen Kaiserreich 1871 bis 1914

(Bonn: Dietz).

Rohe , K. (1982), ‘Die ‘‘verspatete’’ Region’ in P. Steinbach (ed.), Probleme der

politischen Partizipation im Modernisierungsprozess (Stuttgart: Klett-Cotta).

Rohl , J. (1969), Germany without Bismarck: the Crisis of Government in the

Second Reich, 1890–1900 (Tubingen: Rainer Wunderlich Verlag).

Rolfes , M. (1976), ‘Landwirtschaft 1850–1914’, in H. Aubin and W. Zorn

(eds.), Handbuch der deutschen Wirtschaft- und Sozialgeschichte, vol. 2 (Stutt-

gart: Klett-Cotta).

Romer , C. (1986), ‘Spurious volatility in historical unemployment data’, Jour-

nal of Political Economy, 94: 1–37.

——(1989), ‘The pre-war business cycle reconsidered: new estimates of gross

national product, 1869–1908’, Journal of Political Economy, 97: 1–37.

Rosenberg , H. (1985), ‘The pseudo-democratisation of the Junker class’, in

G. Iggers (ed.), The Social History of Politics (Leamington Spa: Berg).

Roßler , H., ed. (1995), Es zieht eben einer den Anderen nach: Wanderungen und

ihre Wirkungen auf ausgewahlte Gebiete Schleswig-Holsteins und Ostelbiens

(Scripta Mercaturae Verlag).

Rostas , L. (1948), Comparative Productivity in British and American Industry,

NIESR Occasional Paper no. 13.

Saalfeld , D. (1963), ‘Zur Frage des bauerlichen Landverlustes im Zusam-

menhang mit den preu�ischen Agrarreformen’, Zeitschrift fur Agrargeschichte

und Agrarsoziologie, 11: 163–71.


Sauer , W. (1968), ‘Das Problem des Nationalstaats’, in H. Bohme (ed.), Pro-

bleme der Reichsgrundungszeit 1848–1879 (Cologne: Kiepenheuer und Witsch).

Schama , S. (1995), Landscape and Memory (London: HarperCollins).

Schieder , T. (1961), Das Deutsche Kaiserreich von 1871 als Nationalstaat (Col-

ogne: Westdeutscher Verlag).

Schissler , H. (1986), ‘The Junkers: notes on the social and historical sign-

iWcance of the agrarian elite in Prussia’, in R. Moeller (ed.), Peasants and

Land in Modern Germany: Recent Studies in Agricultural History (London:

Allen & Unwin).

Schm Adeke , J. (1995), Wahlerbewegung im Wilhelminischen Deutschland. Vol. 1:

Die Reichstagswahlen von 1890 bis 1912: eine historisch-statistische Untersu-

chung. Vol. 2: Die Reichstagswahlen von 1890 bis 1912: Wahlergebnisse und

Strukturen im Kartenbild (Berlin: Akademie Verlag).

Schmoller , G. (1895), ‘Die Einkommensverteilung in alter und neuer Zeit’,

Jahrbuch fur Gesetzgebung, Verwaltung und Volkswirtschaft im deutschen Reich,

19: 1067–94.

——(1895a), ‘Uber die Einkommensverteilung’, Jahrbuch fur Gesetzgebung und

Statistik.

——(1900), Grundriss der allgemeinen Volkswirtschaftlehre (Munich: Duncker

und Humblot).

Schneppe , F. (1960), ‘Die Ertragssteigerung auf den leichten Boden der

Luneburger Heide’, Zeitschrift fur Agrargeschichte und Agrarsoziologie, 8/1:

14–22.

Schniedewind , K. (1994), Begrenzter Aufenthalt im Land der unbegrenzten

Moglichkeiten: Bremer Ruckwanderer aus Amerika 1850–1914 (Stuttgart: Franz

Steiner).

Schofield , R. S., Reher , D. and Bideau , A., eds. (1991), The Decline of

Mortality in Europe (Oxford: Oxford University Press).

Sch ollgen , G., ed. (1990), Escape into War? The Foreign Policy of Imperial


Schorske , C. (1955), German Social Democracy 1905–1917: The Development

of the Great Schism (Cambridge, Mass.: Harvard University Press).

Schott , S. (1903), ‘Das Wachstum der deutschen Grosstadte seit 1871’, Sta-

tistisches Jahrbuch deutscher Stadte, 11: 129–49.

Schremmer , E. (1988), ‘Faktoren, die den Fortschritt in der deutschen Land-

wirtschaft im 19. Jahrhundert bestimmen’, Zeitschrift fur Agrargeschichte und

Agrarsoziologie, 36/1: 37–77.

Schultz , T. W. (1964), Transforming Traditional Agriculture (New Haven:

Yale University Press).

Sering , M. (1893), ‘Die Innere Kolonisation im ostlichen Deutschland’,

Schriften des Verein fur Sozialpolitik, 56.


Sering , M. ed. (1899–1910), Die Vererbung des landlichen Grundbesitzes im

Konigreich Preu�en, 4 vols., Landwirtschaftliche Jahrbucher, vol. 28, supp. vol.

1; vol. 29, supp. vol. 3; vol. 37, supp. vol. 5; vol. 34, supp. vol. 2.

——ed. (1932), Die Deutsche Landwirtschaft, Berichte uber Landwirtschaft

Sonderheft 50 (Berlin: P. Parey).

S i emens , W. von (1893, trans. 1966), Inventor and Entrepreneur: Recollections

of Werner von Siemens (London: Lund Humphries).

S i evers , K., ed. (1981), Die deutsche und skandinavische Amerikaauswanderung

im 19. und 20. Jahrhunderts (Neumunster: Karl Wachholtz).

Skrubbeltrang , F. (1954), Agricultural Development and Rural Reform in

Denmark, FAO Agricultural Studies no. 22.

Soetbeer , A. (1882), Umfang und Verteilung des Volkseinkommens im Preu�-

ischen Staate, 1872–1878 (Leipzig: Duncker & Humblot).

Sperber , J. (1983), ‘The shaping of political Catholicism in the Ruhr Basin,

1848–1880’, Central European History, 16/4: 347–67.

——(1997), The Kaiser’s Voters: Electors and Elections in Imperial Germany

(Cambridge: Cambridge University Press).

Spree , R. (1988), Health and Social Class in Imperial Germany (Oxford: Berg).

Squire , L. (1981), Employment Policy in Developing Countries. (Washington:

World Bank).

Stargardt , N. (1994), The German Idea of Militarism: Radical and Socialist

Critics 1866–1914 (Cambridge: Cambridge University Press).

Stark , O. (1991), The Migration of Labour (Oxford: Basic Blackwell).

Statistisches Reichsamt (1932), Das deutsche Volkseinkommen vor und nach dem

Krieg, Einzelschriften zur Statistik des Deutschen Reichs 24 (Berlin: R. Hob-

bing).

Stegmann , D. (1970), Die Erben Bismarcks: Parteien und Verbande in der

Spatphase des wilhelminischen Deutschlands—Sammlungspolitik, 1897–1918

(Cologne: Kiepenheuer & Witsch).

Steinbach , P. (1992), ‘Reichstag elections in the Kaiserreich: the prospects

for electoral research in the interdisciplinary context’, in E. L. Jones and

J. Rettalack (eds.), Elections, Mass Politics and Social Change in Modern

Germany: New Perspectives (Cambridge: Cambridge University Press).

Steller , P. (1903), ‘Die Maschinenindustrie Deutschlands’, Schriften des Ver-

eins fur Sozialpolitik, 106/3.

Stern , F. (1961), The Politics of Cultural Despair (Berkeley: University of

California Press).

Stiglitz , J. (1974), ‘Incentives and risk sharing in share-cropping’, Review of

Economic Studies, 41: 219–55.

St Ubel , H. (1935), Staat und Banken im Preu�ischen Anleihewesen von

1871–1913 (Berlin: C. Heymann).

St Urmer , M., ed. (1976), Das Kaiserliche Deutschland (Dusseldorf: Droste

Verlag).


——(1990), ‘A nation state against history and geography: the German di-

lemma’, in G. Schollgen (ed.), Escape into War? The Foreign Policy of Imperial


Suval , S. (1985), Electoral Politics in Wilhelmine Germany (Chapel Hill: Uni-

versity of North Carolina Press).

Taylor , L. and Anda , P. (1988), ‘Long run income distribution and growth’,

in H. Chenery and T. Srinivasan (eds.), Handbook of Development Economics,

vol. 1 (Amsterdam: North-Holland).

Taylor , A. M. and Will iamson , J. G. (1997), ‘Convergence in the age of

mass migration’, European Review of Economic History, 1: 27–63.

Taylor , L. (1979), Macro Models for Developing Countries (New York:

McGraw-Hill).

Tcherkinsky , M. (1922), The Landschaften and their Mortgage Credit Oper-

ations in Germany 1770–1920 (Rome: International Institute of Agriculture).

Temple , J. (1999), ‘Income Distribution in the Harris–Todaro Model’, manu-

script, NuYeld College, Oxford.

Tenfelde , K. (1977), Sozialgeschichte der Bergarbeiterschaft an der Ruhr im 19.

Jahrhundert (Bonn: Verlag Neue Gesellschaft).

Thiel , H. (1887), ‘Der Verbrauch von Kunstdunger im Preussischen Staate’,

Landwirtschaftliche Jahrbucher, 16: 447–79.

Thomas , B. (1973), Migration and Economic Growth: A Study of Great Britain

and the Atlantic Economy, rev. edn. (Cambridge: Cambridge University

Press).

Thompson , A. P. (2000), Left Liberals, the State and Popular Culture in Wil-

helmine Germany (Oxford: Oxford University Press).

Th Unen , J. H. von (1826), Der Isolierte Staat in Beziehung auf Landwirtschaft

und Nationalokonomie (Berlin: Wiegandt, Hempel und Parey).

T illy , R. (1986), ‘German banking 1850–1914: development assistance for the

strong’, Journal of European Economic History, 15: 113–52.

T ipton , F. R. (1976), Regional Variations in the Economic Development of

Germany in the Nineteenth Century (Middletown, Conn.: Wesleyan University

Press).

Todaro , M. P. (1969), ‘A model of labour migration and urban unemploy-

ment in less developed countries’, American Economic Review, 59: 138–48.

——(1997), Economic Development (New York: Longman).

Tracey , M. (1982), Agriculture in Western Europe: Challenge and Response,

1880–1980 (London: Granada).

Twose , A. (2001), Statistics and the German State, 1900–1945: The Making of

Modern Economic Knowledge (Cambridge: Cambridge University Press).

US Department of Commerce (1975), Historical Statistics of the United States:

Colonial Times to 1970.

US Immigration Commission (1909–10), Immigrants in Industries, 61st Con-

gress, 2nd Session, Senate Documents, no. 633, vols. 68–85.


US Immigration Commission (1910–11), Statistical Review of Immigration

1820–1910, 61st Congress, 3rd Session, Senate Documents, no. 756, vol. 20.

Van Zanden , T. (1991), ‘The Wrst green revolution’, Economic History

Review, 44: 643–64.

Verein fur Sozialpolitik (1893), ‘Verhandlungen des 20. und 21. Marz 1893’,

Schriften des Vereins fur Sozialpolitik.

Voelcker , A. (1883), ‘Manure’, Encyclopaedia Britannica, 15: 505–12.

Vogele , J. (1998), Urban Mortality Change in England and Germany,

1870–1913 (Liverpool: Liverpool University Press).

——and Woelk , W., eds. (2000), Stadt, Krankheit und Tod: Geschichte der

stadtischen Gesundheitsverhaltnisse wahrend der Epidemiologischen Transition

(Berlin: Duncker & Humblot).

Wade , R. (1990), Governing the Market: Economic Theory and the Role of

Government in East Asian Industrialisation (Princeton: Princeton University

Press).

Wagner , A. (1871), Rede uber die Soziale Frage. Gehalten am 12 Oktober 1871

(Berlin: Wiegandt und Grieben).

——(1902), Agrar- und Industriestaat (Jena: G. Fischer).

——(1904), ‘Zur Methodik der Statistik des Volkeinkommens und Volksver-

mogens’, Zeitschrift des Koniglich Preu�ischen Statistischen Bureaus, 2: 41–122.

——(1904a), ‘Weitere statistische Untersuchungen uber die Verteilung des

Volkseinkommen in Preu�en’, Zeitschrift des Koniglich Preu�ischen Statis-

tischen Bureaus, 4: 229–67.

Wagner , C., trans. G. Skelton (1978–80), Cosima Wagner’s Diaries (London:

Collins).

Wagner , P. (1896), Die Steigerung der Rohertrage der Landwirtschaft im Laufe

des 19. Jahrhunderts, Diss., Jena.

Walker , M. (1964), Germany and the Emigration, 1816–1885 (Cambridge,

Mass.: Harvard University Press).

Webb , S. (1982), ‘Agricultural protection in Wilhelmian Germany: forging an

empire with pork and rye’, Journal of Economic History, 42: 309–26.

Weber , A. (1908, trans. 1976), ‘From ‘‘Die Grossstadt und ihre soziale Pro-

bleme’’ ’, in A. Lees and L. Lees (eds.), The Urbanisation of European Society

in the Nineteenth Century (Lexington: Heath).

Weber , A. F. (1899), The Growth of Cities in the Nineteenth Century: A Study

in Statistics (New York: Macmillan).

Weber , M. (1893), ‘Die Lage der Landarbeiter im ostelbischen Deutschland’,

Schriften des Verein fur Sozialpolitik, 55.

——(1894, trans. 1979), ‘Developmental tendencies in the situation of East

Elbian rural workers’, Archiv fur soziale Gesetzgebung und Statistik, 7: 1–41,

translated in Economy and Society (Berkeley: University of California Press).


——(1948), ‘Capitalism and rural society in Germany’, in H. H. Gerth and

C. Wright Mills (eds.), From Max Weber: Essays in Sociology (London:

Kegan Paul).

——(1993), Landarbeiterfrage, Nationalstaat und Volkswirtschaftspolitik, ed.

W. Mommsen, Vol. 4 of M. Weber, Gesamtausgabe Abteilung I: Schriften und

Reden, 1892–1899.

Wegge , S. (1997), Migration Decisions in Mid-Nineteenth Century Germany,

Diss., Northwestern University, Evanston, Illinois.

——(1998), ‘Chain migration and information networks: evidence from nine-

teenth century Hesse-Cassel’, Journal of Economic History, 58/4: 957–86.

Wehler , H.-U. (1969), Bismarck und der Imperialismus (Cologne: Kiepenheuer

& Witsch).

——(1985), The German Empire 1871–1918 (Leamington Spa: Berg).

——(1995), Deutsche Gesellschaftsgeschichte. Vol. 3: 1849–1913 (Munich:

C. Beck).

Wellh oner , V. (1989), Gro�banken und Gro�industrie im Kaiserreich (Gottin-

gen:Vandenhoeck und Ruprecht).

Wenckstern , H. von (1911), Existenz-Bedingungen se�hafter Landarbeiter,

Diss., Rostock.

Wenneman , A. (1997), Arbeit im Norden: die Italiener im Rheinland und West-

falen des spaten 19. und fruhen Jahrhunderts, Schriften des IMIS 2 (Osnabruck:

Rausch).

W ill iamson , J. (1988), ‘Migration and urbanisation’, in H. Chenery and

T. Srinivasan (eds.), Handbook of Development Economics, vol. 1 (Amsterdam:

North-Holland).

W ill iamson , J. G. (1985), Did British Capitalism Breed Inequality? (Boston:

Allen & Unwin).

——(1991), Inequality, Poverty and History (Cambridge, Mass.: Basil Black-

well).

——(1995), ‘The evolution of global labour markets since 1830: background

evidence and hypotheses’, Explorations in Economic History, 32: 141–95.

——(1997), Growth, Distribution and Demography: Some Lessons from History,

NBER Working Paper no. 6244.

W inkler , J. (1993), Sozialstruktur, politische Traditionen und Liberalismus: eine

empirische Langsschnittstudie zur Wahlentwicklung in Deutschland, 1871–1933

(Opladen: Westdeutscher Verlag).

W itt , P.-C. (1970), Die Finanzpolitik des Deutschen Reiches von 1903 bis 1913

(Hamburg: Matthiesen).

——(1973), ‘Der preußische Landrat als Steuerbeamter 1891–1918: Bemerkun-

gen zur politischen und sozialen Funktion des deutschen Beamtentums’, in

I. Geiss and B. Wendt (eds.), Deutschland in der Weltpolitik des 19. und 20.

Jahrhunderts (Dusseldorf: Bertelsmann Universitatsverlag).


Witt , P.-C. (1985), ‘The Prussian Landrat as tax oYcial, 1891–1918’, in G.

Iggers (ed.), The Social History of Politics (Leamington Spa: Berg).

World Bank (1994), The East Asian Miracle: Economic Growth and Public Policy,

World Bank Policy Research Report.

Wrigley , E. A. (1961), Industrial Growth and Population Change: A Regional

Study of the CoalWeld Areas of North-West Europe in the Later Nineteenth

Century (Cambridge: Cambridge University Press).

Wygodzinski , W. (1917), Die Landarbeiterfrage in Deutschland (Tubingen:

J. C. B. Mohr).


INDEX

absent revolution thesis 23

agrarian reform 45

and aristocracy 34, 36

and banking system 32–3

in central and north-west

Germany 31–3

and child care 141

and compensation 24, 41–2

and consolidation of holdings 44

and credit 25, 32–3, 35

and demographic pressures 26

and Denmark 26, 28–31

and difficulty of 23

east of the Elbe 34–8

aristocracy’s role 36

distance from industrial areas 36

limited credit 35

peasants’ weak position 36

poor communications 34–5

success of 36–8

supposed failure of 34

and effects of 191

egalitarianism 53

and farm prices 26–8

and infant mortality 141

and labour service 31–2, 39, 41

and land ownership

distribution 320

and landlords’ land gains 32, 33

and market access 25–6

and objectives of 24, 28

and population growth 141

and productivity 24–5, 29

in Prussia 31

cultivated land area 40

peasant land losses 38–40

regional variation in impact

of 40–3

and regional variations in 43

and social costs of 43

and Sonderweg school 23

in southern Germany 44–5

and standards of comparison 23

and statistical analysis of 23

as test of political process 23–4

agriculture:

and contemporary views of 181

and cultivated land area 40

and debt levels 245–6

and demesne-farming 37–8

and economic development 15

and the economy:

decline of importance in 218–25

domestic demand 219, 220, 221,

223

international trade 216–23

protection 217–18, 221,

222, 224

and education 226

and emigration 81–2, 85–8

and enclosure 32, 34, 41–2

and farm size 48–9, 51–2

and immigration 92, 93

and impact of capitalist

values 12

and industrialization 17, 185–6

consequences of 187

impact of 215–18

and institutional problems

with 181

and Kuznets Curve 53–4

and labour contract system 183,

185, 208

and labour force 218

employed 46–7

productivity increase

246–7

self-employed 47

agriculture: (Continued)

and labour surplus 193, 209

and land ownership

distribution 49–53

tenancies 52–3

and Lewis Model 207–9

and migration 5, 6, 12–13, 102–4,

106, 109

concerns over 190–2

consequences of 187

costs of 186–7

large estates 192, 193, 194

market conditions 104–5, 107–8,

113

population density 249–50,

251–2

seasonal labour 74–5

as symptom of social

problems 191

and missing market hypothesis 184

and mutual obligations 186–7

and owner-occupation 46, 47

and payments-in-kind 182–3

and population changes 199–201

and prices 104–5, 220–1, 222

fall in 197

and productivity 16–17, 24–5, 29,

215–16

artificial fertilizers 234–5,

239–40, 242, 244

breeding improvements 245

compared with France 46

impact of increase 216

improvements in 218, 223–4

international comparisons 230–4

labour force 246–7

labour-saving technology 247–8

large estates 241–2

livestock intensity 229–30, 242,

245

migration 113, 123, 237, 246

regional patterns of 235–45

soil quality 242–4

sugar beet cultivation 239–41

yield improvement 227–9, 239,

244

and quit rate from 148–51, 171–3,

197–9

farm size 149–50, 171, 197

geographic variations 150–1

population growth 173

and structure of:

fertility decisions 142–3, 169

migration 55, 107, 113

and technological

improvements 201, 216,

225–6, 234–5, 242–5

labour-saving 247–8

and transport costs 25–6

and underemployment 12, 16,

185–8, 204–7, 214,

247–9, 273

partible inheritance 205–6

and wages:

influences on 201–4, 207–8,

211–13

rising level of 196–7

and Weber (Max) on 193–7

see also agrarian reform

Argentina, and emigration to 81

aristocracy, and agrarian reform 34,

36

Army Bill (1913) 332

artificial fertilizers, and

productivity 234–5, 239–40,

242, 244

‘Asian values’, and industrialization 4

Atkinson, A B 299

Australia 216

Austria-Hungary, and migration to

Germany 91

authoritarianism, and

industrialization 4–5

Baden, and agrarian reform 44

Bairoch, P 46, 230, 232

390 Index

banking system:

and agrarian reform 32–3

Denmark 30

and consolidation of 257

and expansion of 185

and investment 15, 254

financing of 257–9

and rural credit 25

and underdevelopment of 320

Bauernschutz 31, 36

Bavaria, and agrarian reform 44

Belgium, and immigration 80

Berlin, migration to 63–5

and further moves 71–3

and migrant occupations

66–7

and sources of migrants

68–70, 75

Berlin Statistical Bureau 98

Berliner Handelgesellschaft 257

birth rate 127

and cities 153–4

and decline in 317

and economic conditions 142

and less-developed

countries 317

and migration 106–7

and rural-urban differences 153

Bismarck, Otto von 346

Blackbourn, D 3 n4

Brandenburg, and agrarian

reform 41, 42, 43

Brazil 216

and industrialization 5

and land ownership

distribution 315

and soya-bean production 314

Broadberry, S N 233, 234

Broesike, Dr M 98

Browne, M 263

bureaucracy, and unrepresentativeness

of 2

Canada, and emigration to 81

capital:

and mobility of, and Lewis

Model 318–19

and ownership distribution 312,

313

Catholicism:

and demographic behaviour 154,

168–9

and population growth 146, 155–6,

174

and Social Democratic Party

support 337–9

child care, and agrarian reform 141

Chile, and industrialization 4

China, and labour surplus 352, 353

cities:

and birth rates 153–4

and contemporary views of 188–9

and death rates 154–5, 157–8,

174–5

and employment 268

and fertility rates 145–6,

152, 169

and impact of agricultural

decline 217–18

and inequality 323–4, 325–6

and life expectancy 157–8, 283

and migration to 63–6, 111–12

congestion effect 275

further moves 71–3

migrant occupations

66–7

sources of 68–70

and opportunities in 160

and population growth 151–6,

173–7

and rate of growth of 59, 61–2,

151, 293

and skill levels 312

see also urbanization

Clapham, J H 31, 225–6, 239

Index 391

Cologne:

and migrant occupations 66

and migration to 65–6

communications:


and improvements in 157

and migration 113

compensation, and agrarian

reform 24, 41–2

Conrad, J 44

consolidation, and agrarian reform 44

constitution, and election

of 1912 332

corporations, and investment 254

Craigie, P G 48–9

credit:

and agrarian reform 25, 32–3, 35

and Danish agrarian reform 30

and expansion of 185

Darmstadter Bank 257

Dasgupta, Partha 184 n5, 186, 187

death rates 127

and cities 154–5, 157–8, 174–5


and economic development 156–7

and migration 158

and migration analysis 71, 106

and wages 154–5, 170

debt, and agriculture 40 n44, 245–6

demography:

and birth rates 127

cities 153–4

decline in 317

economic conditions 142

less-developed countries 317

rural-urban differences 153

and comparison of east and

south 147

fertility rates 137

marriage and migration 137–9

marriage rates 134–7

migration movements 139–41

non-agricultural

employment 132–4

and death rates 127, 131

cities 154–5, 157–8, 174–5

economic development 156–7

wages 154–5, 170

and demographic transition 124

regional variations 127–8

and economic factors:

influence of 130–1

non-agricultural

employment 132–4, 172–3

and fertility rates 127, 131, 137

agricultural institutions 142–3

decline in 159

factors affecting 144, 146, 147,

152–3

large estates 146, 169

partible inheritance 143, 144–5,

169, 318

proximity to cities 145–6, 152,

169

in rural Prussian Kreise 142–5,

167–70

and infant mortality 141

and life expectancy 283

international comparisons 158


and marriage rates 131

economic conditions 131, 137,

177–80

migration 137–9

proximity to cities 152


and migration 112, 121–2, 124–5,

160

birth rate 106–7

death rates 106

population growth 129–30


and population growth 126–7

Catholicism 146, 155–6, 168–9,

174

392 Index

in cities 151–6, 173–7

delayed marriage 177–8

industrial structure 156, 174,

175

industrialization 141

inequality 147–8

Lewis Model 156–60

regional variations 128–30,

132–4

rural migration 148–51, 170–3

rural regulation of 142, 159, 317

wages 175

and population movements 125–7

Denmark, and agrarian reform 26,

28–31

and agricultural productivity 29

and context of 31

and financing of 30

and nobility 28–9

and non-legislative aspects 29

and peasants 30–1

and state initiative 28

and unequal benefits 30

Deutsche Bank 257

development economics:

and relative backwardness 3–4

see also Lewis Model

Disconto-Gesellschaft 257

Dresdner bank 257

economic development:

and agriculture 15

as alteration of production

function 310–11

and death rates 156–7

and financial sector 254

and inequality 14, 295

and investment 253–4

and land ownership

distribution 315

and ‘leapfrogging’ 264–5

and migration 6, 112, 122–3, 350

and model of 294

economy:

and agriculture:

decline in importance of 218–25

domestic demand 219, 220, 221,

223

industrialization 17, 185–6, 187,

215–18

international trade 218–21,

222–3

prices 220–1, 222

protection 217–18, 221, 222, 224

and consequences of migration:

skill endowments 279–81

wage convergence 274–9

and economic stabilization 351–2

and migration and economic

cycles 271–4, 291–2

and structural changes in 255–66

Anglo-German rivalry 263–4

export growth 255–6, 260–1,

265–6

financial sector 257

growth potential 265

investment 256–7

investment finance 257–9

new industries 264–5

raw material costs 262–3

recovery from recession 259–60

terms of trade 261–2, 264

wages 263–4

education:

and agriculture 226

and income distribution 311–12

and industrialization 215–16

Eggers, Count von 30–1

election of 1912:

and foreign policy 333

and political consequences

of 332–3


support:

agricultural population 341

aristocratic estates 342–3

Index 393

election of 1912: (Continued)

candidate composition 343–5

city size 337, 339

desire for change 343

econometric analysis 356–60


343

feelings of political

inequality 343

increase in 331–2

inequality 340, 342

population mobility 340

proximity to towns/cities 341–2

religious affiliation 337–9, 341

rural/small town areas 340–3

and statistical analysis, limitations

of 336–7

electoral politics:

and radical right 334

and urbanization 335–6

and voting trends 333–4

see also election of 1912

Elektroboom 256, 259

Eley, G 1 n1

elites, and unwillingness to

modernize 2

emancipation, and population

growth 142

emigration:

and agricultural character of 85–8

and decline in 89

and destination countries 81

and Germans living abroad 80–1

and labour surplus 56–7, 89

and regional variations in 85–6

and structural change in 80

to United States 81, 83–4, 321

agricultural employment 88–9

decline in 85

influence of agriculture 81–2

previous occupations 86–8

wage ratio 85

see also immigration; migration

employment:

and agriculture 46–7, 218, 246–7

and export-oriented

industries 266–7

and household 206, 212, 214, 249

and migrant participation rates 273

and non-agricultural 132–4, 172–3,

199

industrial concentration 268–9,

286–8

regional pattern of 267–8, 271

vertical integration 268, 269–71,

288

and self-employment 47, 206, 214

and sugar beet cultivation 201

see also underemployment

enclosure 32, 41–2

and aristocracy 34

and Danish agrarian reform 29, 30

Engel, Ernst 98, 190, 297–8, 305

Engels, Friedrich 297

Evans, Richard 330–1

exports:

and finished goods 261

and growth in 255–6

and terms of trade 10, 254

expropriation, and agrarian reform 24

famines 157

farm size:

and agriculture 48–9, 51–2

and productivity 241–2

and quit rate from

agriculture 149–50, 171, 197

and wages 201–2, 203, 213

see also large estates

Feinstein, Charles 296, 307

fertility rates 127

and agricultural institutions 142–3

and decline in 159

and delayed marriage 131, 159,

177–8


394 Index

and factors affecting 144, 146, 147,

152–3

and large estates 146, 169

and partible inheritance 143,

144–5, 169, 318

and Polish-speaking areas 145, 146,

147, 167–8

and proximity to cities 145–6, 152,

169

and regional variations in 137

in rural Prussian Kreise 142–5,

167–70

and wages 146, 147

see also farm size

Finance Bill (1913) 332

financial sector:

and development of 254, 257

and financing of investment 257–9

and underdevelopment of 320

Finckenstein, Finck von 37 n35

Fircks, A von 131

foreign direct investment 320

foreign policy:

and election of 1912 333


and internal politics 1

and naval expansion:

economic motives for 347–8

lack of alternatives 348–9

and social context of 2–3

France:

and agriculture:

farm size 48–9

owner-occupation 46

productivity 46

and urbanization 59

Fremdling, R 233

Geisenberger, S 299, 300

gender, and migration 76–9, 137–8

see also women

Gerschenkron, Alexander 3

Gilbert and Lawes 244

Goltz, Theodor von der 57 n2,

191

and Prussian land reform 34

Gutsherrschaft administrative system,

and agrarian reform east of the

Elbe 36

Hajnal, J 178

Handdienst, and agrarian reform 41

Hayami, Y 318

Helfferich, K 298, 299

Hesse, and agrarian reform 44

historical school, and income

distribution 297–8

historiography:

and economic models 18–19

and German exceptionalism

19–20

of Imperial Germany 1–2, 18,

331–2

and Tirpitz naval plan 350

see also Kehrite school

Hoffmann, Walter 9, 37 n36, 104,

218, 235, 257, 300

household behaviour, and fertility

decisions 318

household employment:

and underemployment 206, 214,

249

and wage levels 212

illegitimacy 139

and death rates 142

immigration 90–3

and agriculture 92, 93

from Austria-Hungary 91

and countries of origin 91

and increase in 90

and industrial workers 92, 93

and labour surplus 57

and net gains from 80


from Russia 90, 91

Index 395

Imperial Germany:

and Anglo-German

relations 263–4, 346

and democratic transition 345–6

economic stabilization 351–2

potential for 353–4

and electoral politics of 331–5

growth of Social Democratic

Party 335–45

and German exceptionalism 19–20

and industrialization 3, 331–2

standards of comparison 20–1,

332

success in coping with 353

and internal weakness of 1–3, 330,

349

and naval expansion:

economic motives for 347–8

lack of alternatives 348–9

naval weakness 347

and problems faced by 333

and socio-strategic dilemma

of 346–7

Imperial Statistical Office 98

income distribution, see inequality

incomplete revolution thesis 23

industrial structure:

and birth rates 155


and inequality 326


export-oriented industries 266–7


286–8

patterns of non-agricultural



288

and population growth 156, 174,

175

industrialization:

and adoption of capitalist values 12

and agriculture 17

eastern 185–7

impact on 215–18

structure of 55

and authoritarianism 4–5

and contemporary concerns

with 189–90

and education 215–16


and German debate on 295

and inequality 6, 13–14, 297–8,

305, 307–8, 316, 320

and internal weakness of Imperial

Germany 3, 331–2

and Kuznets Curve 13–14, 53, 295

and Lewis Model 6–11

and migration 5–6, 7, 56, 94

periods of 71

sources of 75–6

and nature of German 3, 20–1

in pioneer countries 20


and productivity growth 350

and relative backwardness 3–4

and rural society 185–6

consequences for 187

and speed of 59, 61, 62, 151, 293,

353

and standards of comparison 20–1,

332

and trade unions 7

and urbanization 15

inequality:

and avoidance of 326–7

and cities 323–6

and contemporary concerns

with 190

and debate over 297–9

and economic development 14, 295

and electoral politics 340, 342, 343

and historical school 297–8

and industrial structure 326

and industrialization 6, 13–14,

297–8, 305, 307–8, 316, 320

396 Index

and international

comparisons 306–8

and Kuznets Curve 13–14, 53, 295


and land ownership

distribution 49–53, 313–16,

324

and Lewis Model 6–11, 317

and migration 53, 56, 309–10, 316,

322, 325


and productive knowledge 311–13

skills 311, 316, 321

vocational training 311–12, 321

and proximity to Russian

Poland 324

in Prussia, nineteenth

century 303–6

and rural society:

divergences in 22

rural-urban gap 13–14, 54–5,

108

and savings’ concentration 317

and skill endowments 311–12,

315–16, 321

and sources of 319–21

analysis by simulation 322–3


and tax data:

opportunity provided by 303

problems with use of 299–302

and terms of trade 320, 323, 326

and urbanization 316, 323, 326

and worsening of 302–3

infant mortality:

and agrarian reform 141

and illegitimate children 142

inheritance, see partible inheritance

insurance 289–90

internal politics:


and industrialization 3, 331

and primacy of 1

international trade:

and agriculture 216–17, 218–21,

222–3

and Anglo-German rivalry 263–4,

346

and export growth 255–6, 260–1,

265–6

employment 266–7

and German naval policy 347–8

and impact of 254–5

and new industries 264–5

and raw material costs 262–3

and terms of trade 10, 254, 261–2,

264

investment:

and banking system 254

and capital mobility 319

and corporations 254

and economic development

253–4

and financial sector 254

and financing of 257–9

and improved opportunities

for 313

and inequality 321

and institutional constraints

on 320

and pattern of 256–7

and savings 253–4

Ipsen-Kollmann thesis 141

Ireland, and land ownership

distribution 315

Italy, and migration to Germany 90

Japan:


and agricultural prices 27

and land ownership

distribution 315

joint-stock companies, and

investment 254

Junkers, and opinions of

226 n14

Index 397

Kehrite school:

and election of 1912 333

and internal weakness of Imperial

Germany 1–2

and key concept of 1

and radicalization of the right 332

and Tirpitz naval plan 350

and weakness of argument 353

Kindleberger, C 56 n1

Knapp, Gustav 189

and migration from rural

east 191–2

and peasant land losses 38


Knodel, J 137, 142

knowledge, productive 311–13

and skills 311, 316, 321

and vocational training 311–12,

321

Kollmann, W 135, 139, 152

and Ipsen-Kollmann thesis 141

Koselleck, R 34

Kuczynski, Robert 98

Kuznets, Simon 13, 295, 312,

316–17

Kuznets Curve 4

and agriculture 53–4

and assumptions of 22

and conditionality of 326, 327

as contemporary fiction 296

and disputing of 296

and escaping from 327

as German peculiarity 296, 297

as historical fact 296, 297

and inequality and

industrialization 13–14, 53,

295

and Lewis Model 294, 296

and migration 295, 322

in Prussia, nineteenth

century 303–6


labour market:

and ‘Continental’ nature of 80, 94

and impact of capitalist values 12

and industrial development 6–8

and integration of 276

and non-agricultural employment,

and migration 132–4, 172–3,

199

and population increase 127

see also employment

labour service:

and abolition in Denmark 29


abolition of 31–2, 39

and fertility decisions 142–3

and inefficiency of 38 n38

labour surplus:

and criticism of concept of 16, 215

and eastern agriculture 193, 209

and emigration 56–7, 89

and immigration 57


and implications of 294

and inequality 310

and Kuznets Curve 13–14

and Lewis Model 6–9, 294

and migration 56, 93–4, 97, 100,

102, 103–4, 141

and political stability 353

and sources of 293

and underemployment 204–7,

249–50

land ownership:


and Danish agrarian reform 29

and distribution of 49–53

and income distribution 313–16,

324


and political power 314–15

and re-allocation of 315–16

and skill acquisition 315

398 Index

and the state 314

and turnover of 39–40

land prices 104–5, 106


land reform, see agrarian reform

land tax assessment 244

Landesrentenbank 32, 33

Landflucht 34

land-owners, and adoption of

capitalist values 12

Landschaften 25, 33

large estates:

and economies of scale 184

and fertility rates 146, 169

and migration 192, 193, 194


and quit rate from

agriculture 149–50, 171

and wages 201–2, 203, 213

Laux, H-D 151 n39

‘leapfrogging’, and economic

development 264–5

Lee, W 141

less-developed countries, and birth

rates 317

Levy, H 36 n30

Lewis, Arthur 156–7, 317

and employment 12

and the Lewis Model 6

Lewis Model 4

and capital mobility 318–19

and criticisms of 15–16

and eastern agriculture 207–9

and explanatory value of 350

and inequality 6–11, 317

and Kuznets Curve 294, 296

and labour supply 6–8

and labour surplus 6–9, 294

and migration 79–80, 97, 100

and modification to 9


and propositions of 294

and shortcomings of 11

and ‘turning point’ 7, 8, 294

and wages 9–11

Liebig, Justus von 244, 245

life expectancy:

and international comparisons 158

and migration 283


Lindert, Peter 296, 307

living standards:

in eastern Germany 139–41

and regulation of population

growth 159

and wage levels 212–13

local government, and Social

Democratic Party 332

Lupitz system 244, 245

Magdeburg, and peasant farmers

33

Malthus, Thomas 178

market access, and agrarian

reform 25–6

marriage:

and decline in fertility rates 159

and delayed 131, 159, 177–8

and economic conditions 131, 137,

177–80

and land ownership 143


and partible inheritance 145

and proximity to cities 152


and wages 178–80

Marschalck, P 141

Marx, Karl 6

Mecklenburg, and agrarian

reform 45

medicine 157

Meitzen, Adolph:

and peasant debt 40 n44


Meldepflicht system 57

middle class, and cities 151, 160

Index 399

migration:

and agriculture 5, 6, 12–13, 102–4,

106, 109



market conditions 104–5, 107–8,

113

productivity 113, 123, 237, 246

quit rate from 148–51, 171–3,

197–9

structure of 55, 107, 113

and demographic factors 112,

121–2, 124–5, 160

birth rate 106–7

death rates 106, 158

population growth 129–30


and economic cycles 271–4, 291–2


and economic factors 112, 122–3

non-agricultural

employment 132–4, 172–3,

199

and factors affecting attitude

towards 188

and gross migration 57

and historical research 97

and impact of 19

and industrial structure 266–71

export-oriented industries 266–7


286–8

patterns of non-agricultural



288

and industrialization 5–6, 7, 56, 94

and inequality 53, 56, 309–10, 316,

322, 325

rural-urban gap 13–14, 54–5, 108

and international 56–7

and international comparisons 58

and intra-urban 56

and Kuznets Curve 295, 322

and labour surplus 56, 93–4, 97,

100, 102, 103–4, 141

and Lewis Model 79–80, 97, 100

and marriage patterns 137–9

and measurement of 62–8

and net migration 57

and periods of 71, 94, 139

and population density 249–50,

251–2

and population mobility 63–6

migrant occupations 66–7


and rural population

growth 148–51, 170–3

and rural-urban 56, 102–4, 110–13


consequences of 187

costs of 186–7


problems 191

and seasonal 56, 64–5, 74–5

by sex 76–9, 137–8

and skill endowments 279–81

vocational training 311–12

and social change 350–1

and sources of migrants 68–76

agriculture-related 74–5

further moves 71–3

industrial-related 75–6

and statistical analysis of:

changing nature of

(1871–1913) 100–6

net migration by region

(1871–1910) 106–10, 115–16

panel data analysis

(1871–1910) 116–19

regression variables and

sources 114–15

rural-urban (1890

census) 110–13, 119–23

and statistical sources for 98

and underemployment 214, 249

400 Index

and unemployment 271–4

and urbanization 59–62, 111–12

and wages 56, 94, 111


wage dispersion 276

and women 76–9, 138–9

life expectancy 283

wages 281–3

see also emigration; immigration;

Lewis Model

missing market:


and share-cropping 184

Mittelstand, and preservation of 189

mobility 63

and cities 63–6

and migrant occupations 66–7

and sex of migrants 76–9

and sources of migrants 68–76

modernization, and incompleteness

of 2

Mommsen, Wolfgang:

and German exceptionalism 19

and ‘skirted decisions’ 1 n2

monitoring costs, and share-

cropping 183–4

Muller, J H 299, 300

Newbery, D 184 n4

newly industrializing countries

(NICs) 4

nobility, and Danish agrarian

reform 28–9

nutrition, and indicators of 139–40

O’Brien, P 233

occupations, and migration 66–7

Ostmarkverein 193

Pakistan, and land reform 46 n53

partible inheritance:

and fertility decisions 143, 144–5,

169, 318


and underemployment 205–6,

214

patents, and productive

knowledge 311, 312

peasants:

and agrarian reform:

bias in favour of 32, 320

credit availability 32–3

Denmark 28, 30–1

east of the Elbe 36

labour service 31–2, 39

land losses 38–40

and demesne-farming 37–8

and efficiency of 16

and non-agricultural earnings

44–5

and preservation of 189

Phelps-Brown, E H 263

Philippines, and agrarian reform 25

n5

Poland, see Russian Poland

Polish-speakers:

and fertility rates 145, 146, 147,

167–8

see also Russian Poland

Pomerania:

and agrarian reform 41,

43, 45

and large estate failures 40

and peasant landholdings 38

Posen, and agrarian reform

41, 42

Prados de la Escosura, L 233

prices:


and agriculture 104–5, 197, 220–1,

222

and death rates 131

and marriage rates 178–9

and wages 212

production, and economic

development 310–11

Index 401

productivity:

and agrarian reform 24–5, 29

and agriculture 16–17, 24–5, 29,

215–16

artificial fertilizers 234–5,

239–40, 242, 244

breeding improvements 245

compared with France 46

impact of increase 216

improvements in 218, 223–4


labour force 246–7

labour-saving technology 247–8

large estates 241–2

livestock intensity 229–30, 242,

245

migration 113, 123, 237, 246

regional patterns of 235–45

soil quality 242–4

sugar beet cultivation 239–41

yield improvement 227–9, 239,

244


and wages 7, 9–11

see also knowledge, productive

profits, and industrial development 6

Prokopovitch, S 13, 295

property rights:


and evolution of 315–16

and exceptional re-allocation

of 315

and productive knowledge 312–13

protection, and agriculture 217–18,

221, 222, 224

Prussia:

and agrarian reform 31, 42

consolidation 44

peasant land losses 38–40

regional variation in impact

of 40–3

and agrarian reform east of the

Elbe 34–8

aristocracy’s role 36

distance from industrial areas 36

limited credit 35

peasants’ weak position 36

poor communications 34–5

success of 36–8

supposed failure of 34

and agricultural prices 27–8

and rural credit 25

see also taxation data, Prussian

Prussian Statistical Bureau 98

public health 157

Quante, Peter 148

railway construction, and

migration 75–6

Rauh, Manfred 354

recruits, and height of 139–40

regional government, and Social

Democratic Party 332

regions 94–6

Reichstag, and constitutional position

of 332–3

relative backwardness:

and development economics 3–4


religious affiliation, and voting

behaviour 337–9, 341

Rentenbanken 33

‘revolutions from above’ 23

Rhineland, and agrarian reform

41

Royal Credit Bank (Denmark) 30

Ruhr, and sources of migrants 75

rural society:

and efficiency wages and

unemployment 187

and fertility rates 142–5,

167–70



consequences of 187

402 Index

and inequality 22

rural-urban gap 13–14, 54–5,

108

and migration:


consequences of 187

costs of 186–7


population density 249–50,

251–2


problems 191

and mutual obligations 186–7

and population changes 199–201

and population growth:

migration 148–51

regulation of 142, 159, 317

and wages, influences on 201–4,

207–8, 211–13

see also agrarian reform

Russia:


and migration to Germany 90, 91

Russian Poland 201

and inequality and proximity

to 324

and influence on wage

levels 202–3, 204, 208, 213,

279, 320–1

as source of labour 194, 199, 208

Saalfeld, D 38

Sammlungspolitik 2

savings:


and improved opportunities

for 313

and inequality 317

and investment 253–4

Saxony, Kingdom of:

and agrarian reform 32, 41, 42


and rural credit 25

Saxony Statistical Office 98

Schaaf’hausen Bankverein 257

Schissler, H 36 n31

Schmoller, Gustav 189, 190, 297

Schremmer, E 234, 235

Schrott, S 62 n6

Schultz, Theodore 16–17, 215,

246

Schultz-Lupitz, A 244, 245

science, and agriculture 225–6,

244–5

seasonal migration 56, 64–5

self-employment:

and agriculture 47

and underemployment 206, 214

serfdom, and abolition in

Denmark 28

Sering, Max 190, 191

share-cropping:

and criticisms of 181–2

as efficient solution 184

and missing market 184

and monitoring costs 183–4

and positive views of 183

Silesia:

and agrarian reform 41, 42


skill endowments:

and income distribution 311, 316,

321

and land ownership

distribution 315


and vocational training 311–12

‘skirted decisions’, and internal

weakness 1–2

smallholdings, and migrants 281–2

Smith, Adam 181

social change, and rural-urban

migration 350–1

Social Darwinism 2, 330

Social Democratic party 310

and broadening appeal of 335

Index 403

Social Democratic party (Continued)

and electoral support for:

agricultural population 341

aristocratic estates 342–3

candidate composition 343–5

city size 337, 339

desire for change 343



343

feelings of political

inequality 343

gains in 1912 election 331–2

growth of 333–6

inequality 340, 342

population growth 154, 156,

174, 175

population mobility 340

proximity to towns/cities 341–2

religious affiliation 337–9, 341

rural/small town areas 340–3

and internal politics of 334

and local/regional government 332

Soetbeer, A 298, 299

Sonderweg school 23, 330, 331

South Korea:


and land ownership

distribution 315

and land reform 46 n53

soya-bean production 314

Spain, and industrialization 4

Spanndienst, and agrarian reform 41

Spannfahige, and agrarian reform 32,

33

state, the, and land ownership

distribution 314

statistics, and contemporary use

of 98–9

steam-threshing 247–8

Stegmann, Dirk 332

Stein-Hardenburg laws 192

stillbirths 126, 135

sugar beet cultivation:

and artificial fertilizers 239–40

and employment 201


and wages 204, 212

Switzerland, and emigration to 81

Tagelohner 15

Taiwan:


and land ownership

distribution 315

tariffs, see protection

taxation data, Prussian:

and income distribution 303–6


measurement of 297–9

worsening of 302–3

and opportunity provided by 303

and problems with use of 299–302

technology:

and agriculture 216, 225–6, 234–5,

242–5

impact on employment 201

labour-saving 247–8

tenancies, and agricultural land

52–3

tenant rights, and Danish agrarian

reform 29, 30–1

terms of trade:

and inequality 320, 323, 326

and movement in 10–11, 254

Thunen, J H von 36 n30

Tirpitz, Alfred von 348

Tirpitz naval plan 348, 349

and Kehrite school 350

Todaro, M P 317

trade unions:


and wage levels 321

transport costs, and agriculture

25–6

‘turning point’, and Lewis Model 7,

8, 294

404 Index

underemployment 11 n17

and agriculture 12, 16, 185–8,

204–7, 214, 247–9, 273

partible inheritance 205–6

and definition problems 250

and informal urban sector 15

and labour surplus 204–7, 249–50

and migration 214, 249

unemployment:

and efficiency wage rates 187


United Kingdom:

and Anglo-German rivalry 263–4,

346

response to German foreign

policy 348, 349

and farm size 48

and immigration 80

and land ownership

distribution 49, 51

and political development of 46,

331

and urbanization 59–62

United States:

and agriculture 216

exports 217, 219–20

and city growth rates 62

and emigration to 81, 83–4, 321

agricultural employment 88–9

decline in 85

influence of agriculture 81–2

previous occupations 86–8

wage ratio 85

and income distribution

313–14

urbanization:

and contemporary views of 188–9


and electoral politics 335–6

and impact of agricultural

decline 217–18


and inequality 316, 323, 326

and international comparison of

rates of 59–62

and life expectancy 157–8, 283

and migration 59–62, 111–12

and population growth 151–6,

173–7


vote 335–6

and social dislocation 59, 351

and speed of 59, 61, 62, 151, 293,

353

Van Zanden, T 233–4

vocational training:

and income distribution 311, 321


Voelcker, Augustus 234–5

voting behaviour, see electoral politics

wages:

and advantages of cash

payments 182

and agricultural:

agricultural/industrial

ratio 105–6

influences on 201–4, 207–8,

211–13

rising level of 196–7

and Anglo-German

comparison 263–4

and death rates 154–5, 170

and decline in rural 302

and efficiency wage rates 187

and fertility rates 146, 147

and German/American ratio 85

and labour supply 6, 7

and marriage rates 178–80

and migration 56, 94, 111

impact of 190


wage dispersion 276

women 281–3

and payments-in-kind 182–3

Index 405

wages: (Continued)


and productivity 7, 9–11

and proximity of Russian

Poland 202–3, 204, 208, 212,

279, 320–1

and regional variations 275–6,

278–9

and rural-urban gap 108

and trade unions 321

Wagner, Adolf 189, 190, 298

Wagner, Richard 297

Weber, Adna 58

Weber, Adolf 188–9

Weber, Max 11–12, 17, 107, 187,

191, 192, 207

and eastern agriculture 193–7

capitalist organization 195

children of labourers 195

impact of migration 196

labour relationships 194–5

large estates 193, 194

political concerns 193–4


and labour-saving technology 247


Wehler, Hans-Ulrich 2 n3, 331

and German exceptionalism 19–20

Westphalia:



Williamson, J G 264, 296, 313

women:

and illegitimate children 142

and migration 76–9, 138–9

life expectancy 283

wages 281–3

working class 154, 156, 174

World Bank 4, 307

Wurttemberg, and agrarian reform 44

Zollverein, and creation of 20

406 Index