Methodologies for Measuring Agricultural PriceIntervention Effects

SWP394

World Bank Staff Working Paper No. 394

June 1980

by: P.squale L. Scandizzo and Colin BruceAjriculture and Rural Development Department

Copyright (r) 1910The Worla' Ban"1818 H Streee, N.W.Washington, D.C. 20433, U.S.A.

The views and interpretations in this document are those of theauthors and should not be attributed to the World Bank, to itsaffiliated organizations, or to any individual acting in their behalf.

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

The views and interpretations in this document are those of the author and should

not be attributed to the World Bank, to its affiliated organizations, or to any

individual acting in their behalf.

THE WORLD BANK

INTERNATION.'z- rMO!iZ'W- FUND

Staff Working Paper No. 394 JmL1si

, U 1 u 1985

June 1980lIMMNlATIONAL BlJe FO1

FMCOCNSTRUCTiOil iilD DLVELOPt-LFNT

METHODOLOGIES FOR MEASURING AGRICULTURAL PRICE INTERVENTION EFFECTS

This paper describes the use of micro-economic, informal methodologies

in six country case studies (Argentina, Egypt, Kenya, Pakistan, Thailand and

Yugoslavia) and in subsequent additional analysis. The research has confirmed

that there is extensive intervention with market forces by governments inLDCs, which has turned the domestic terms of trade against agriculture and has

significant efficiency and income distribution effects. The methodologies --

which include the use of nominal and effective protection coefficients, domestic

resource cost, and net economic benefit coefficients, producer and consumer

subsidy equivalents and producer and consumer surpluses -- are evaluated in

terms of their capacity: (a) to provide Bank staff with operationally useful

instruments of analysis; (b) to offer appropriate advice to client country

governments and their planning institutions concerned with pricing policies; and

(c) to help bridge the methodological gap between project cost benefit analysis

on the one hand and sector and country economic work on the other hand. While

the micro-economic methodologies have their limitations, they do provideuseful tools of analysis, especially for countries which have sets of national

parameters (accounting prices and conversion factors), which can also be used in

project cost benefit analysis. Application of the methodologies clearly involves

additional country economic and sector work, but the extent of the discriminationagainst agriculture in general and food production in particular justifies

some substitution or the use of additional resources on this type of policy

analysis on a systematic basis. The paper recommends a three stage work program,

and some short-cuts are proposed. This paper should be read in conjunction with

"Agricultural Prices, Taxes and Subsidies, a Review of Experience", a forthcoming

Staff Working Paper by Gilbert Brown and Graham Donaldson; and four case studies

for: Argentina (Reca), Egypt (Cuddihy), Pakistan (Brown and Gotsch) and Thailand

(Bertrand).

Prepared by: Pasquale L. Scandizzo and Colin BruceAgriculture and Rural Development Department

Copyright 0 1980The World Bank1818 H Street, N.W.Washington, D.C. 20433 U.S.A.

METHODOLOGIES FOR MEASURING AGRICULTURAL PRICE

INTERVENTION EFFECTS

Table of Contents

Page No.

PROLOGUE .................... * * * * * *............ i-v

EXECUTIVE SUMMARY . ........ . ...................... ... vi-x

I. INTRODUCTION ........ .................................. 1

II. MEASURING MARKET DISTORTIONS: ACCOUNTING PRICES FORTRADEABLES ................................................ 3

III. MEASURING MARKET DISTORTIONS: ACCOUNTING PRICES FORNON-TRADEABLES ....... ..................................... 8

Accounting Price of Labor ..... .............. 8

Accounting Price of Land ............. .. ............... 10

Accounting Price of Capital ........................... 10

IV. MEASURING MARKET DISTORTIONS: SUMMARY MEASURES ......... .. 13

Nominal Protection Coefficient ........................ 13

Traded Goods .......................................... 14

Non-Traded Goods ................. ..................... 14

Value Added ........................................... 15

Effective Protection Coefficient (EPC) ........ ........ 15

Effective Subsidy Coefficient (ESC) ................... 17

Domestic Resource Cost (DRC) ..... ..................... 17

Net Economic Benefit (NEB) ............................. 20

Producer Subsidy Equivalent (PSE) and Consumer SubsidyEquivalent (CSE) ................ .. .................. 22

V. EFFECTS OF PRICE DISTORTIONS: A MORE GENERAL PARTIAL

EQUILIBRIUM METHODOLOGY .................................... 24

VI. EVALUATION OF METHODOLOGIES ............................... 45

VII. GUIDELINES FOR ANALYZING PRICE INTERVENTION EFFECTS ... .... 48

APPENDIX I: A Mathematical Model of Accounting Prices

APPENDIX II: Price Effects of Agricultural Supply

APPENDIX III: Price Effects on the Demand for Agricultural

Products

APPENDIX IV: Empirical Background and Main Text References

METHODOLOGIES FOR MEASURING AGRICULTURAL PRICEINTERVENTION EFFECTS

PROLOGUE

i. In most less developed countries (LDCs), the domestic terms of

trade for agriculture vis-a-vis the other sectors of the economy, particularly

the manufacturing sector, are unfavorable. The agricultural terms of trade

have been lowered in LDCs as a result of a variety of Government interventions.

Some of these interventions are the instruments for applying an implicit

theory of economic development which had its heyday 20-30 years ago, but

which is less fashionable at the present time. Other interventions stem from

a variety of ad hoc beliefs, concerns and considerations, such as the ease of

raising revenue (from export taxes on agricultural products and the appropria-

tion of the profits of monopsonistic agricultural marketing boards); the need

to placate pressure groups, particularly urban ones where, it is believed,

that political power primarily resides (keep the cost of urban wage goods,

i.e. food and some clothes, low); the importance of stabilizing agricultural,

particularly food, prices; the belief that infant manufacturing industries

need to be heavily protected, and the propensity of politicians to look only

at short-run effects and to ignore longer-term considerations.

ii. The implicit theory of economic development, which has been the

inspiration of so much intervention, put simply, runs like this: productivity

and productivity growth rates are (much) higher in non-agricultural, parti-

cularly manufacturing, sectors than in the agricultural sector, ergo priority

in development must be given to manufacturing, and to do this resources--the

agricultural surplus and foreign exchange--must be tapped and transferred to

finance manufacturing investment. Underlying this theory are two suppositions

1/ of considerable importance: first, is the assumption that prices do not

matter because farmers are not price responsive and, therefore, they can be

taxed without a major effect on overall output and without disturbing production

patterns. Put in economic jargon, this means that supply is inelastic, both

in the short-run and in the long-run. Second, is the heroic belief thatbureaucrats are capable of solving what are in essence a myriad of simultaneous

equations, which are normally solved, however one may dislike the solutions,

in the market place. It is a heroic belief even in the more developed countries;

how much more heroic is it in LDCs with their poor information services and

very scarce managerial resources.

iii. This development theory, together with the other beliefs, concerns

and considerations mentioned above, have spawned a veritable host of interven-

tions: administrative prices, physical controls, taxes and subsidies, which

have tended to proliferate over the years in response to the relative ineffec-

tiveness of partial intervention. There are signs, however, of a rethinking

of the theories, beliefs and concerns and a greater propensity, if not to

allow prices to allocate scarce resources, at least to guide the decisions of

bureaucrats. There is also a growing realization that the theories of market

intervention have either not achieved what it was believed they would or,

perhaps, that in a number of countries the planners may have killed, or made

ill, the goose which lays the golden egg.

1/ There are others, which are not discussed here, and we do not discuss

the soundness of the theory as a whole.

-ii-

iv. It is curious that so little attention has been paid to or researchundertaken into the effects of administered agricultural prices, taxes andsubsidies on the patterns of production, consumption, income distribution,agricultural research and development generally. I/ Little or no attemptshave been made by developing countries to monitor and evaluate the effects ofgovernment interventions. The Bank's attitude towards administered agriculturalprices, taxes and subsidies had not been spelled out in any kind of policypaper and no guidelines had been provided to country economists, agricultural/rural sector economists or project analysts as to how they might analyze thesituations in their countries. It was for these reasons that a researchproject was started in the second half of 1976. It was based upon 8 countrycase studies. The basis for selection was partly geographic spread, partlyinteresting examples of different kinds of interventions, partly the avail-ability of data--a more or less serious constraint in all countries, andpartly the willingness of governments to permit the studies. It is notclaimed that the resulting list--Argentina, Egypt, Kenya, Mexico, Pakistan,Portugal, Thailand and Yugoslavia--is a fully representative sample of LDCs:they are simply examples of interesting cases of countries with widely differ-ing political philosophies, policies, resources and technologies. Mexico andPortugal were chosen because agricultural sector models exist 2/, which couldbe used to contrast with the informal, more partial equilibrium methods beingused for the other countries. No West African countries were included becauseof the four case studies made earlier by Balassa and Associates. 3/

5. The central issues which the research project was designed toaddress were the effects of interventions on a) the efficient allocation ofscarce resources, and b) the distribution of income--between individuals,between rural and urban areas and between agricultural and non-agriculturalsectors. In order to analyze the central issues of efficient resource alloca-tion and equitable distribution effects, it is necessary to ask:

- What are prevalent kinds and what is the extent of governmentintervention?

- What are the theories and objective functions (goals) whichunderlie the interventions?

- To what extent are actual interventions in harmony withstated objectives? And to what extent are interventionscomplementary (reinforcing) or offsetting (conflicting)?

1/ However, see Distortions of Agricultural Incentives, Theodore W. Shultz,Ed., Indiana University Press, Bloomington and London, 1978 for anexcellent analytic review of the current state of the art.

2/ See: A Development Model for the Agricultural Sector of Portugal,Egbert, Alvin and Kim, Hyung M., World Bank Staff Occasional Paper No. 20,and Multi-Level Planning: Case Studies in Mexico, Goreaux, Louis andMann, Allan, North Holland Publishing Company, Amsterdam, 1973.

3/ "Methodology of the Western Africa Study", February 6, 1976, and thefour case studies in Ghana, Mali, Senegal and the Ivory Coast, Balassaand Associates, unpublished mimeographs.

-iii-

What are the norms by which to judge whether interventionsare beneficial or harmful to societies--in the short-runand in the long-run?

In the short-run and in the long-run, how do interventions

affect:

- farmer's incentives and supply responses--for particular

crops/produce within agriculture via the own-elasticity

and the cross-elasticities of supply--in general andrelative to other sectors?

- resource allocation

- employment in agriculture

- income distribution

- savings, particularly public savings

- trade.

- How do we measure the effects of various kinds of intervention?

- Do governments have the data and the economic expertise available

to trace the effects of intervention?

- In general, do interventions discriminate against agriculture and

the rural areas?

- Is there a real conflict between cheap food for the poor and fair

prices for the farmers?

vi. Clearly, to give answers to all the above questions would be a

formidable task, and largely because of data problems the case studies concen-

trated on the efficiency issues.

vii. In respect of the informal, more partial equilibrium methodologies,

after discussion and review it was decided to:

(a) Undertake 6 country case studies (plus the Mexico and Portugal

formal agricultural sector models) from which it was hoped that,

as a minimum expectation, it would be possible to obtain:

(i) A description of and recent trends in the system ofadministered agricultural prices, taxes and subsidiesand other policy tools. The information would be obtainedfrom Bank economic, agricultural sector and special study

reports, supplemented by official government publicationsand field investigations.

-iv-

(ii) An analysis of the importance of the various policy toolswith respect to the government budget and public savings,agricultural output and consumption. The ratios of:taxes on agriculture to total government revenue;subsidies to total government recurrent expenditure andpublic savings; value of net taxes/subsidies to the valueof output; and the value of net subsidies/taxes to consumersto the value of consumption would be calculated.

(iii) Preliminary analysis of the adequacy of economic incentivesin relation to the country's comparative economic advantage.A limited number of key food and non-food commodities wouldbe chosen and initially no attempt would be made to differen-tiate between agro-ecological zones. The methodologies usedwould be those used by Balassa and Associates in their WestAfrican study estimating nominal protection coefficients(NPCs), effective protection coefficients (EPCs), effectivesubsidy coefficients (ESCs) and the domestic resource costs(DRCs)--and estimating net economic benefits (NEBs), measuredin economic (efficiency) prices, using a simplified (time-collapsed) version of the Little-Mirrlees (L-M) 1/ Squire-van der Tak (S-T) 2/ methodology. Data would be drawnfrom Bank Project Appraisal Reports, official governmentpublications, farm management studies and field work.

(iv) Some broad analysis of the incidence of existing systemsof administered prices, taxes and subsidies on thedistribution of incomes as between agriculture and non-agriculture and between rich and poor.

(b) Carry out a literature search of farmers' responses as reflectedby demand and supply elasticities for selected agriculturalcommodities. 3/

(c) Undertake a functional review of the treatment of administeredprices, taxes and subsidies with respect to agriculture in Bankeconomic and sector reports. 4/

1/ I.M.D. Little and J.A. Mirrlees, Project Appraisal and Planning forDeveloping Countries, Heinemann (London) 1974 and Basic Books (New York)

1974.

2/ L. Squire and H. G. van der Tak, Economic Analysis of Projects, JohnsHopkins University Press, Baltimore and London, 1975.

3/ See "Agricultural Price Policy and Supply Response: A Review of Evidenceand an Interpretation for Policy", 1977 AGREP Division Working Paper No. 2,1977.

4/ See J. Graves and Y. Kimaro, "Functional Review of the Treatment of Agri-cultural Pricing in Bank Group Economic and Sector Reports - Progress Report",AGREP Division Working Paper No. 1, 1976.

(d) Estimate from a selection of Bank Project Appraisal reports:

(i) Retention Coefficients--ratios of farmgate financial

prices to farmgate economic efficiency (border) prices.

(ii) Cost Coefficients--ratios of financial input costs to

financial value of output to see if there are any

discernible similarities/differences as between

crops and countries. 1/

(iii) Compare the informal, partial equilibrium methods

in (a) (iii) above with the use of formal agricultural

sector models in Portugal and Mexico, and evaluate

the usefulness of both.

viii. An overview and integrated summary of the results of the country

case studies and the complementary desk studies are given in:

"Agricultural Prices, Taxes and Subsidies:

a Review of Experience", Staff Working Paper

(forthcoming), prepared by Gilbert Brown and

Graham Donaldson.

ix. For various reasons only four of the country case studies have so far

been written up in a suitable from for publication. These are:

"Argentina - Country Case Study of Agricultural Prices,

Taxes and Subsidies", Staff Working Paper No: 386,

prepared by Lucio Reca (Consultant).

"Agricultural Price Management in Egypt", Staff

Working Paper No: 388, prepared by William Cuddihy.

"Prices, Taxes and Subsidies in Pakistan Agriculture,

1960-1976", Staff Working Paper No: 387,

prepared by Gilbert Brown and Carl Gotch (Consultant).

"Thailand - Case Study of Agricultural Input and

Output Pricing," Staff Working Paper No: 385, prepared by

Trent Bertrand (Consultant).

x. This paper describes, analyses and evaluates the informal, more partial

equilibrium methodologies used in the country case studies.

1/ See Graves and Kimaro, op. cit.

-vi-

EXECUTIVE SUMMARY

Introduction

i. Research into the extent and effects of government market interven-tions within the agricultural sector of developing countries has revealed thatthe interventions are widespread and have important effects upon the efficiencyof agricultural production, trade, investment and employment, upon croppingpatterns and resource use and upon the distribution of incomes, both withinthe agricultural sector and between the agricultural sector and other sectors.

ii. The research involved inter alia, the application of formal andinformal methodologies, a literature search into the results of econometricstudies of supply responses and an analysis of the treatment of agriculturalmarket intervention in Bank economic and agricultural sector reports. Thispaper reviews and evaluates the use of a number of informal partial equili-brium methodologies in six country case studies (Argentina, Egypt, Kenya,Pakistan, Thailand and Yugoslavia) and in subsequent desk studies. Because ofdata problems the studies concentrated on the efficiency rather than on thedistributional aspects of intevention policies.

Reference Points

iii. In order to measure and then evaluate the effects of governmentintervention which affect incentives, efficiency and income distributionthrough the price mechanism, it is necessary to have a set of referencepoints. Such points can be regarded in either positive or normative ways.Border priced values for traded goods are selected, not as perfect, but as theonly realistic set of reference points available. It is recognized that theremay be special situations, for individual countries at particular times, whenone would not want to use border prices, but in general, border priced valuesare the most appropriate opportunity costs for most countries, provided theyare likely to be enduring as far as can be foreseen. At the very least, theuse of border price reference points provides an estimation of the net welfarelosses and transfers resulting from departures from free trade, which may or maynot be offset by contra advantages. Whether it is considered that they do or donot, involves both positive and normative judgments and eventually politicaldecisions.

iv. Very few countries monitor and evaluate the effects of their marketinterventions on a regular basis. Some "systems" have evolved consciously inresponse to industry-led theories of development, involving taxation of theagricultural surplus to finance (heavily) protected manufacturing development.Others have grown up in response to ad hoc reactions to specific economicevents or to placate pressuregroups, and different parts of the system areoften inconsistent with other parts or with other development objectives.Interventions which have caused considerable distortions are difficult toremove or modify.

Valuation of the Traded Goods

v. There are problems of identifying appropriate accounting (shadow)prices for traded goods, i.e. expressed in border prices or border pricesequivalent values, but, while care must be exercised to ensure one is compar-ing like to like, agricultural products are much more homogeneous thanmanufactured products.

vi. Almost all tradables, which are in fact not traded, and non-tradables

have some traded elements in them, if only the consumption of traded goods by

people. Decomposition of such goods will result in traded components and the

primary non-traded factors of production--labor, land and capital.

Accounting Price of Labor

vii. The accounting price of labor is determined by: (a) its opportunity

cost, i.e. the border price value of its weighted average marginal product in

alternative uses; (b) the cost to society of the increased consumption resulting

from the additional labor, measured in terms of the foreign exchange foregone; (c)

the social value of the increased consumption, modified as may be considered to be

necessary by income distribution and public income weights; (d) the value of leis-

ure foregone which may be expressed in the form of a reservation wage; and (e) the

social costs of migration if any.

Accounting Price of Land

viii. The accounting price of land is determined not only by its opportunity

cost as a factor of production but also by the fact that it is an asset held as

a hedge against inflation, for speculative purposes and, most important of all,

as a depository of value and secure means of subsistence.

Accounting Price of Capital

ix. Since the ways of measuring incentives and comparative cost advantage

of pricing policies which have economy-wide effects do not necessarily involve

discounting, there may be no need to estimate the accounting rate of interest

except for normative purposes. Instead an annual value of capital in various

uses has to be estimated. This can be done by calculating the depreciation and

decomposing it into its traded and non-traded components and valuing them

accordingly. Profit was ignored in the case studies for reasons of time and

simplicity. This simplification is not thought to have distorted the analysis

significantly.

Summary Measures of Incentives/Disincentives

x. Four coefficients were used to measure the incentive/disincentive

effects of administered prices, taxes and subsidies to producers. These are:

-- Nominal Protection Coefficient (NPC), which is the

ratio of the domestic price to its border price(measured at the farmgate).

-- Effective Protection Coefficient (EPC), which meas-

ures the effects of protection measures, not only

on traded outputs but also on traded inputs, i.e.the ratio of value added expressed in domestic

market prices to value added expressed in border

prices.

- Effective Subsidy Coefficient (ESC) in which the

EPC is adjusted by the sum of the differencebetween profits, taxes, interest and the price

-viii-

of non-traded goods actually paid and what isconsidered to be "normal charges", i.e. thevalue of direct and implied subsidies.

-- Producer Subsidy Equivalent (PSE), which isthe subsidy, net of indirect taxes, given toproducers, expressed as a percentage of themarket value of each commodity.

xi. The results of the country case studies show in all the countries stud-ied, that while there have been some variations over time, the NPCs, the EPCs andthe ESCs are typically less than one and the PSEs are typically negative, indicat-ing disincentives to the production of the main crops included in the study, andto agricultural production in general. The results of the country case studiesalso indicate that for agricultural products, where the value of the purchasedinputs are typically a small proportion of the value of output, there are in-significant differences between the NPCs, EPCs and ESCs. This is an importantfinding because EPCs and ESCs are much more difficult to calculate and one canrely on NPCs, which are easy to calculate, to obtain estimates of the incentive/-disincentive effects of trade taxes and subsidies.

Summary Measures of Comparative Cost Advantage

xii. Two basic efficiency measures can be used to measure the comparativecost advantage of producing various agricultural products and by inferenceagricultural products in general vis-a-vis other commodities, but both measureshave the same foundation:

-- Domestic Resource Cost (DRC), which can be expressedeither as a coefficient or as a percentage. In itsoriginal formulation it is obtained by computing thevalue of domestic resources (primary, non-tradedfactors of production) in domestic currency units,it takes to earn or save a unit of foreign exchange.Expressed as a coefficient it can be compared to anaccounting price (shadow price) of foreign exchange.If the DRC for a product is less than the appropriateaccounting price of foreign exchange a comparativecost advantage exists in producing the commodity inquestion and vice versa. If both the numerator andthe denominator are expressed in the same numeraire,e.g. border prices, then the coefficient which re-sults is essentially the same as the Net EconomicBenefit, but with the terms arranged differently.

-- Net Economic Benefit (NEB), is another way of look-ing at the comparative advantage in terms of econo-mic efficiency, using a time-collapsed form of theLittle-Mirrlees/Squire-van der Tak methodology,without applying social weights. For this reasonit is probably preferable to the DRC because it isin conformity with the Bank's own cost-benefit method-ology for projects. It is also simpler to estimateand less arbitrary in that one does not have to worryabout which items to put in the numerator and whichin the denominator.

-ix-

xiii. Where the land and labor requirements for different crops within speci-

fic areas do not vary substantially, the main determinants of the DRC and NEB arerelative yields and relative border prices. Hence, in such circumstances one can

simplify the analysis of comparative advantage by reference to a comparison of

the border price multiplied by the yield for each crop. Where, however, thereare significant differences in land and labor inputs full DRC and NEB calcula-

tions have to be made. The experience of the country case studies is that these

calculations are made much easier if there exist already sets of national para-

meters (accounting prices, conversion factors and commodity cost profiles). Thus

the requirements for agricultural sector price analysis (and other sectors as

well) and project cost-benefit analysis coincide.

xiv. It is interesting to contrast the Producer Subsidy Equivalents (PSEs)

with Consumer Subsidy Equivalents (CSEs). Whereas in the more developed coun-tries PSEs are positive and CSEs are negative, the reverse is generally true in

developing countries: farmers are taxed and they pay for the often considerable

consumer subsidies, from which some of them benefit in part.

Economic Gains and Losses and Welfare Transfers

xv. PSEs and CSEs are a simplified form of a partial equilibrium method-

ology derived from the theory of consumers' and producers' surplus. Although the

theory of economic surplus has its well-known limitations, nevertheless, as Hicks

has commented: "While it is easy to raise objections to the use of the concept

of economic surplus for providing answers for policy formulation, it is diffi-

cult to find workable alternatives." (Hicks, p. 791 of "The Rehabiliation of

Consumers Surplus". Review of Economic Studies, Vol. 8, Feb., 1960-61.) There

are also practical problems, the most important of which are estimating reason-

ably accurate demand and supply elasticities. Nevertheless, such estimates are

available for a large number of countries, and, subsequent to the country case

studies, calculations of the "Net Economic Loss in Production",l/ the "Net

Economic Loss in Consumption", the "Welfare Gain to Producers", the "Welfare

Gain to Consumers", the "Change in Foreign Exchange Earnings" and the "Change inGovernment Revenue" for a limited number of agricultural products in Argentina,

Eygpt, Kenya, Pakistan, Portugal, Thailand and Yugoslavia were made. These

calculations confirm the broad picture of discrimination against agriculture and

of net welfare transfers for agricultural producers to consumers and, to a lesser

extent, to other producers via the government.

xvi. The partial equilibrium framework briefly described above was also

used to evaluate foodgrain policies in a different group of developing coun-tries with the difference that an index of cost effectiveness in terms of nutri-

tional needs was used. This is defined as the ratio between the total amount of

foodgrains distributed through the rationing systems and the amount reaching the

population below the FAO/WHO calorie requirement. The results show (a) that

government intervention never closes more than 50% of the potential nutrition

gap; (b) that the costs of distribution are borne partly by producers (through

the loss in producers' surplus due to procurement at less than border prices)

and partly by the government; and (c) that, with the exception of Sri Lanka,

1/ The literature generally refers to these as the "Net Social Loss in Pro-

duction, etc.", but we prefer to call it "economic" rather than "social" inorder to distinguish the economic from the social.

the loss of producers' surplus per unit of additional consumption of the mal-nourished is equal to or greater than the corresponding fiscal cost. In allcases, the sheer magnitude of the losses imply that the food procurement anddistribution policies could stand considerable improvement, both because of highproduction losses and because of leakages of benefits from low food prices to un-intended beneficiaries.

Evaluation of Methodologies

xvii. While the informal, partial equilibrium methodologies employed havetheir limitations and drawbacks, used with discretion they can illuminate some ofthe efficiency effects on agricultural production and broad welfare transfersfrom agriculture to consumers, and to other sectors through governments. A numberof short-cuts are feasible and sufficiently accurate for broad-brush analysis.Thus, for agricultural products in developing countries, where input costs are arelatively small proportion of output values, incentive/disincentive effects canbe measured by nominal protection coefficients alone. This is fortunate becauseNPCs are much easier to calculate than EPCs or ESCs. Where the land and laborrequirements within specified agro-ecological zones do not differ very much asbetween different crops, the comparative - cost advantages of different crops canbe measured by comparing relative yields and relative border prices. However,where land and labor requirements do differ DRCs or, preferably, NEBs have to beestimated. This task would be greatly faciliated if country sets of nationalparameters (accounting prices, conversion factors and crop profiles), updatedperiodically, were made available. Such national parameters are required notonly for policy analysis but also for project analysis.

Conclusions

xviii. Research has shown that market interventions have moved the domesticterms of trade substantially against agriculture in many developing countries tothe detriment of farmer's income, foreign exchange earnings and food production.This has particularly important implications for the fight against poverty andfor the growing world food shortage. It implies that governments should reviewand evaluate their intervention policies, and aid agencies, including the WorldBank, should assist them to do so. In turn this probably implies that agricul-tural policy analysis and sector work generally should receive priority overother forms of country economic and sector work. A large part of the invest-ment in research and extension and the transfer of knowledge and technologiesthrough technical assistance and through projects and programs will continue tobe thwarted if farmers are not given reasonable incentives to produce food andfibers in accordance with relative factor endowments and comparative cost advant-ages. This paper concludes with a brief outline of a three-stage recommendedprogram of analysis.

EVALUATING METHODOLOGIES FOR MEASURING

EFFICIENCY EFFECTS OF AGRICULTURAL PRICE INTERVENTION

I. INTRODUCTION

1. The objective of this paper is to present methodologies for the

evaluation of agricultural price intervention in developing countries. This

presentation of suggested methodologies is based in turn upon a description and

an evaluation of the use of micro-economic, informal methodologies used in six

country case studies (Argentina, Egypt, Kenya, Pakistan, Thailand and Yugo-

slavia) 1/ of agricultural prices, taxes and subsidies, which formed part of a

wider World Bank Research Project, which commenced late in 1971. 2/ The

research has confirmed that there is extensive interference by governments in

developing countries with market forces which has turned the domestic terms of

trade against agriculture--severely so in many countries--with adverse con-

sequences in terms of economic efficiency and income distribution. 3/ This

discrimination against agriculture in developing countries is in marked

contrast to the over-generous treatment of agriculture in most of the more

developed countries.

2. Given the bias against agriculture and the unsatisfactory and often

internally inconsistent agricultural pricing policies in most developing

countries, the methodologies presented here may prove useful (a) to provide

Bank staff with operationally useful instruments of analysis; (b) to offer

appropriate advice to client country governments and their planning institutions

concerned with pricing policies; and (c) to help bridge the methodological gap

between project cost-benefit analysis on the one hand and sector and country

economic work on the other hand. 4/ It is recognized that there are difficult

communication and education problems in convincing policy makers in developing

countries to make their development policies less discriminatory against

1/ Some of these methodologies were also used in a Bank study of four

West African countries (Ivory Coast, Ghana, Mali and Senegal).

See "Methodology of the Western Africa Study", February 6, 1976

and the four country case study by Balassa and Associates, unpublished

mimeographs.

2/ For an outline of the raison d'etre of the study and its various elements,

use the Prologue to this paper. Formal agricultural Sector Models were

used to study the situations in Mexico and Brazil, but this paper does

not attempt to evaluate these or to compare the formal with the informal

methods.

3/ For an overview and analysis of the various kinds of intervention and

their effects see a companion paper: "Agricultural Prices, Subsidies

and Taxes: A Review of Experience", prepared by Gilbert Brown and

Graham Donaldson.

4/ From this point of view see also "Application of Shadow Pricing to

Country Economic Analysis with an Illustration from Pakistan", World

Bank Staff Working Paper No. 330, prepared by Lynn Squire, I.M.D.

Little and Mete Durag, June 1979.

-2-

agriculture in terms of both economic efficiency and income distribution. Wecan hope only that this paper will make a contribution towards providing abetter basis for communication and for education.

3. It is also recognized that applying the type of analysis and method-ologies proposed is not costless in terms of the time needed to be devoted tosuch work by country, sector and project economists: if more time is spent onsuch agricultural policy analysis, almost certainly less will be spent onother activities. On the other hand, we would argue, the marginal net benefitsfrom this type of work is higher than that of much of the current activitiesof operational economists and would be more operationally useful and supportiveof the Bank's development objectives and lending operations.

4. The paper concentrates on the efficiency aspects of agriculturalprice intervention. This is more by reason of the inadequate data base inmost developing countries, and the consequent difficulty the researchersfound in tracing through all the income distribution effects of the marketinterventions, rather than by choice. There are also other important aspectsnot covered here, such as the effects upon employment, institutions and agri-cultural research. Neither do we examine the constraints on changing policies,such as the need to devise alternative, less discriminatory and more equitableways for governments to raise revenue. Instead this paper concentrates onpricing issues, some of which can be tackled by the new cost benefit methods ofeconomic efficiency pricing.

5. In Section II, market interventions are characterized as inducingdistortions from a system of reference or accounting prices, and this sectionillustrates the problems involved in using border prices as a reference pointfor traded goods. Section III discusses ways of computing accounting pricesfor non-traded goods/factors. Section IV presents concepts and definitionsrelated to summary indicators of comparative advantage and incentives.Section V discusses implications of cross-country evidence on price distortionand general trends emerging from a set of empirical estimates, on the basis ofa more general partial equilibrium methodology based on the theory of consumersand producers' surplus which is first described. While a number of qualitativeapproaches to the study of agricultural pricing are also discussed, the mainfocus of this section is on quantitative partial equilibrium methods and onillustrative applications for a sample of developing countries. Section Vevaluates the usefulness of the various methodologies used, and the paper endswith putting forward some general guidelines for analyzing agricultural priceinterventions. The paper is complemented by four appendices detailing thetheory underlying the methodology proposed, some of the numerical componentsof the estimates presented in the main text, and the sources of the theoreticaland empirical material used.

-3-

II. MEASURING MARKET DISTORTIONS: ACCOUNTING PRICESFOR TRADEABLES

6. This paper makes an assessment of agricultural price intervention on

the basis of three premises: (a) price interventions are widely used in almost

all countries; (b) governments will continue to find it useful to use a

variety of such price interventions in the future; and (c) that the use of

some interventions is a legitimate option of governments provided they are

internally consistent and provided their consequences in terms of economic

efficiency and income distribution have been analyzed in some detail.

7. There are, however, two corollaries of (c). First, there must

be a much better data base or information system plus a system for monitoring

and evaluating the effects of price interventions. Secondly, there has to be

a system of reference points against which the effects of market interventions

can be assessed in terms of development objectives. Border prices -- foreign

prices converted into a country's currency units by applying the official

exchange rate -- are used in this paper as such reference points. It is not

argued here that border prices are perfect reference points, but they are used

to judge the rationality of national pricing systems for two main reasons:

(a) except for the prices of an arbitrary group of trading partners (such as a

common market, a free trade area, etc.) no other common, observable system of

market values exists that can be used to compare prices of different countries;

(b) from the point of view of the individual country, border prices represent

the true opportunity costs of the country's tradeables. Where imports are not

expected to affect the foreign supply price or exports the foreign demand

price, the appropriate border prices to use are the projected c.i.f. import

and f.o.b. export prices. Where, however, an increased demand for imports

would, at the margin, affect the supply price or where the increased supply ofexports would, at the margin, affect the foreign demand price--i.e., the

demand and supply are less than perfectly elastic, the appropriate prices to

use are the projected marginal import cost or marginal export revenue. The

former is the c.i.f. price multiplied by (1+1/e) where (e) is the elasticity

for foreign supply; the latter is the f.o.b. export price multiplied by (1+1/n),where (n) is the elasticity of foreign demand.

8. While existing international markets can provide a useful standard

against which to measure economic performance, it is also important to underline

some of the limitations and problems that go along with such an approach.

First, identifying the appropriate border prices is often difficult where

products are heterogeneous. In agriculture, however, agricultural products are

typically much more homogenous than most industrial products. Nevertheless

care has to be exercised to ensure that the border prices are taken for

basically comparable produce in terms of quality and use. Second, where

international prices fluctuate quite widely, it is difficult to select an

appropriate trend price in real terms. Third, there are a number of problems

connected with the fact that world prices, and hence border prices, are not

free-trade prices, but prices which are often distorted by monopsonistic

situations in other supplying countries, by trade barriers and pricing policies,

-4-

by dumping and by other imperfections in world markets. Some of these distor-tions have resulted in some agricultural produce being produced in the wrongplaces -- sugar beet growing in developed countries is a classic example. Inthese senses, border prices are themselves often distorted, so how can theybe used as reference prices? The answer to this question depends upon one'sviewpoint. From a global point of view one might not want to take internationalprices as one's reference point in many instances, although it is far fromclear what reference points one would substitute for them. Looked at it fromthe point of view of each individual country, however, it is irrelevent thatsome international/border prices may be monopsonistic, dumped or otherwisedistorted ones: what matters is whether or not they are true and lastingtrading options. For example, most milk powder entering into internationaltrade is dumped, but from the point of view of a country deciding whether or notto go ahead with a domestic liquid milk production project, the relevant questionis not whether milk powder is dumped, but whether the dumping is likely tocontinue over the estimated life of the project. If it is, then the borderprice of reconstituted milk from imported milk powder is the appropriate oppor-tunity cost of liquid milk. If not, then the appropriate price to take is theeconomic value of the domestic milk production, valuing the traded components attheir border priced opportunity costs and including depreciation and profitelements.

9. Thus, while border prices are not always and everywhere the mostappropriate accounting prices to use, in general they do appear to capturethe real opportunities open to countries through trade and thus provide aconvenient reference point.

10. There are four ways in which border prices can be used to evaluate acountry's system of domestic prices and to formulate policy recommendations.First, trade prices can be used to evaluate the rationality of the pattern ofspecialization of a country in the light of its trading opportunities. Second,they can be adjusted to producing points and compared to producer (farmgate)prices to evaluate the consistency of private and public incentives in production.Differences between prices faced by domestic producers and prices faced bypublicly-run marketing boards, export-import agencies, and similar agencies ininternational markets imply differences in the rates of investment in thevarious sectors of the economy and may have important consequences for thelong-run pattern of specialization of the country, as well as on the balancebetween the public and the private sector. Third, consumer prices and borderprices can be compared to provide insights into the pattern of consumptionforced upon the country by government market interventions. Fourth, thedeviation between domestic and international prices for commodities or sectorsin the same country can be used to evaluate internal distortions in the domesticterms of trade between such commodities or sectors (e.g. between agriculture andindustry).

11. In all cases, a systematic comparison of international prices anddomestic market values can provide the basis for an analysis of consistencybetween instruments and targets of government policies. While the principlethat the price system should serve national goals is generally accepted, it isclear that most countries have developed over the years fragmentary andchaotic systems of subsidies and taxes. For many countries the prevailing

-5-

market "distortions" may not be related to present goals, and in many cases ithas been shown that they run counter to present goals. Thus, without denyingthe possibility of market interventions, we suggest that international borderprices be considered the most convenient benchmark to design more rationalpolicies.

12. Although the origin of market distortions can be traced to actionsof the government or private entities or to specific deficiencies of themarket system, the ultimate effect of the distortion is by no means easy toidentify or measure. The main difficulty is that, despite the importance ofpricing for planning and policy making, designing market intervention policiesis a process often occurring under the severe constraint of many pre-existingdistortions. These distortions are in part due to factors outside governmentcontrol and in part are the direct results of past actions taken by thegovernment. In all cases they permeate the existing price system, so that thefirst logical step of any attempt at designing market intervention policiesmust depend on the identification and measurement of their net effect.

13. Market interventions can be looked at from two viewpoints. First,they can be measured as the end result of all interactions with the free playof market forces. Second, their impact on selected important characteristicsof the economy can be quantified. In both cases a quantitative evaluationrequires estimates of the effects of the distortions on a distortion-freepricing system and, indirectly on the pattern of allocations of resources,incomes and consumption. For this purpose, as in cost-benefit analysis, asystem of accounting (shadow) prices 1/ and conversions factors is necessary.The accounting prices and conversion factors used in the country case studiesand hence our description of them, were based more on the Little-Mirrlees(L-M) 2/ and Squire-van der Tak (S-T) 3/ methodologies than on the UNIDOmethodology. 4/ This is because the World Bank has opted for the formerrather than the latter, but it is generally agreed now that, provided thebasic economic analysis is the same, in most practical situations the twomethodologies give the same accept/reject decisions.

1/ We prefer to use "accounting prices" rather than "shadow prices", asthe latter has a connotation of something ephemeral, vague, unreal andesoteric, and seems to induce negative reactions in some people.

2/ Project Appraisal and Planning for Developing Countries, I.M.D. Littleand J.A. Mirrlees, Heinemann, London, 1974 and Basic Books, New York,1974.

3/ Economic Analysis of Projects, Lynn Squire and Herman van der Tak,Johns Hopkins University Press for the World Bank, Baltimore andLondon, 1975.

4/ Guidelines for Project Evaluation, Project Formulation and EvaluationSeries No. 2 UNIDO, Vienna, 1972. See also Guide to Practical ProjectAppraisal - Social Benefit-Cost Analysis in Developing Countries,Project Formulation and Evaluation Series No. 2, prepared byJohn R. Hansen for UNIDO, New York, 1978.

-6-

14. The analogy with the process of accounting pricing used in projectcost benefit analysis may be useful in clarifying some of the characteristicsof the systems of accounting prices sought. Cost benefit analysis can itselfbe conceived as a pricing process where rules are sought to impute "undistorted"prices to outputs and inputs of a particular class of projects. Since pricepolicies represent an attempt to impose prices on all market transactions ofa particular class, using accounting prices for cost-benefit analysis is thusa special type of pricing policy - one whose ultimate intention is to reducesome of the consequences of existing distortions.

15. Despite such a close resemblance, there are two important differencesbetween the two pricing methods that make cost benefit analysis a particularlysimple sub-case of the more general pricing problem. First, accounting pricesin cost benefit analysis are used to impute money values to predeterminedbundles of outputs and inputs. Because the projects being appraised are bothsmall and few as compared with the allocative decisions guided by marketprices, accounting prices have only marginal effects on resource allocation inthe rest of the economy. Government pricing policies, on the other hand, arelikely to have a major impact on resource allocation, investment decisionsand, indeed, project appraisal criteria of private and public economic units.Second, project appraisal techniques may either assume away the fiscal impli-cations of the accounting prices adopted, or they may consider fiscal measuresas independent of the net economic benefits estimated. Clearly this is notpossible in the case of economy-wide prices, and accounting prices based onborder prices have to be confronted with their fiscal consequences. This isalso true of cost-benefit analysis where the project financial analysis indicatesa deficit.

16. The selection of a reference system of accounting prices dependson four main factors: (i) the extent of the policy measures evaluated; (ii)the nature of the economic system and the development pattern of the country;(iii) the national goals sought by the government; and (iv) the relationshipbetween the country's economy and its (actual or potential) trade partners.

17. On the first point, although in principle the government can beheld responsible for many, if not most, of the existing distortions - in thesense that it is the direct cause of such distortions or could make efforts toremove them - in practice there will be a number of market imperfections orexisting policies beyond the control of government. Also if the sectordirectly affected by the policy measure is large, the number of taxes andsubsidies substantial, or the impact of the contemplated intervention farreaching, pricing recommendations should be directed to offset some of theundesirable effects of other distortions and to improve the efficiency andequity features of the economy. This is a very complex task and can only reston a careful case-by-case economic and social analysis of each country. Whenit cannot be accomplished without the danger of generating further negativeeffects on the economy, international and/or equilibrium domestic prices stillprovide a "least imperfect" norm. Discrete changes of large magnitude,however, will significantly change the accounting prices of domestic resourcesand of foreign exchange. More will be said on this account in the part of thepaper discussing ways of computing these accounting prices.

-7-

18. On the second and third points, patterns of development in countries

may differ greatly in the emphasis given to accelerated growth, income distri-

bution, industrialization, the weight of the export sector, the importance of

the internal markets, and so on. Often national goals as seen by the government,

or as expressed in the national plan, are a rationalization of a particular

pattern of past development. The importance of prices in this context is that

they can be used to provide incentives to the private sector and partly to the

public sector to achieve specific growth targets or in a manner broadly consis-

tent with the model of development sought for the country. Patterns of develop-

ment and growth targets determine the level of some key prices for domestic or

foreign resources: wages, interests, rental rates and exchange rates.

19. Finally, the relationship between the country and its actual and

potential trade partners needs close scrutiny before a system of reference

prices can be chosen. With the possible exceptions of Hong Kong and Singapore,

no country has moved to a completely free trading position. Across countries,

multilateral and bilateral trade agreements and geographical and political

constraints shape trade patterns and limit the scope for substitution and

competition. Thus border prices are true measures of the country's opportunity

costs, if they take account of likely changes in distortions and of the volumes

of trade and the markets involved. Similarly, border price-based marginal

revenues for goods with finite elasticity rest on the hypothesis that the

country's trade partners will tolerate the imposition of a tariff without

retaliation.

20. All the above considerations suggest that there exists no absolutely

unique system of accounting prices for use in policy formation. For most

countries, however, the rule for specifying accounting prices for cost-benefit

analysis are adequate to provide: (i) a benchmark to judge the extent of

price distortions involved in the current system of market interventions; and

(ii) to suggest a gradual and complete removal of all these distortions where

this is deemed to be feasible and consistent with the national goals.

-8-

III. MEASURING MARKET DISTORTIONS: ACCOUNTING PRICESFOR NON-TRADEABLES

21. If it is accepted as a general rule that border prices provide theappropriate reference point for tradeables, what about non-tradeables? Therule here is that non-traded goods should be broken down once, and probablyonly once in practice, into their traded and non-traded elements, with thetraded elements valued directly at their border prices. 1/ All non-tradedelements should, in theory, be valued as the weighted average of the values ofthe factor marginal productivities in the production of all products using thefactors as inputs. In practice, of course, not all activities can be takeninto account and a "representative" or "typical" activity is taken. Themarginal value product of the input factors are then revalued at their borderprice equivalent values by applying appropriate conversion factors, such as aconsumption conversion factor for labor. 2/

22. Estimating the accounting prices of labor, land and capital presentssome specific problems due to the fact that wages and interest rate levels,particularly, relate to four areas of government concern: (i) employment,(ii) growth, (iii) inflation, and (iv) income distribution. Government'spreferences and objectives in these areas could find direct expression in"merit want" accounting prices for labor, land and capital. In general,however, although it may be quite unrealistic to expect policy makers to beable to translate their targets into price variables, policy recommendationson factor prices can be based on an opportunity cost criterion and appropriatelymodified to reflect policy choices.

Accounting Price of Labor

23. Consider first the accounting price for labor. The pricing prescrip-tion of project cost-benefit analysis is based on four points. First, as forall inputs, the opportunity cost of employing one additional unit of a particularskill depends on the value of the product foregone in its alternative uses.In general such a value is a weighted average of marginal products evaluatedat border prices, the weights being the marginal shares of labor in each ofthese alternative. Second, the giving of a wage represents committing socialincome to consumption. If the government has a high target rate of growth or

1/ This process of decomposition is made much easier if a set of nationalparameters (conversion factors and accounting prices) for a country isavailable, as grouped or individual conversion factors can be applied torelatively unimportant tradeables, rather than hunting for the appropriatedirect border prices. See, for example, National Parameters for ProjectAppraisal in Malaysia in 3 vols. State and Rural Development Project,Economic Planning Unit, Prime Minister's Department, Kuala Lumpur,November 1977 and June 1979, prepared by Michael Veitch and Hamzah Bakar.

2/ It should be noted that if, for example, the UNIDO methodology is beingused, the estimated marginal value products of non-traded inputs willalready be expressed in the UNIDO numeraire -- domestic prices -- sono further conversions are required.

-9-

a low target rate of inflation, a penalty should be charged to incorporate the

cost to society of increased consumption. Alternatively, an expansionary

policy of increasing aggregate demand would give a premium to wage earners for

their high propensity to consume. Third, a wage commits social income to

subgroups of individuals so that a premium (or penalty) may also incorporate

income distribution preferences for a particular group of workers. Fourth,

the wage remunerates a labor effort that requires the worker to forego leisure,

to modify his quality of life, to alter his consumption basket, etc. Thus, a

lower bound to labor opportunity cost is an evaluation of workers' foregone

welfare (the "reservation" wage). These aspects of the accounting price for

labor are summarized in the following equation for the conversion factor for

labor:

CF = [MO + AC$ - aC d /v] c c 2. w

where: CF = conversion factor for labor accounting price for labormarket price for labor

M marginal product of labor at domestic prices

ac conversion factor for consumption

AC = increase in consumption valued at domestic prices

dg = weighted average income distribution coefficient

v = social value of public income relative to private consumption

w = market wage rate.

21. The first term of equation (1) is simply the value of the marginal

product of labor in the next best alternative use, measured at border prices.

The second term is the value of-the increased consumption at border prices.

It is a cost in terms of foreign exchange foregone and, therefore, added to

the marginal product. It is also a benefit; hence the third term is the

social value of the extra consumption and, therefore, is subtracted. It is the

value of the consumption at domestic prices multiplied by an income distribution

weight and divided by the value of public income coefficient to express the

benefit in terms of the numeraire being used. If there is no premium attached

to public income v=l/Oc and, if benefits are not differentially valued, d = 8 v

and equation (1) collapses to CF. = M 8 /w , which is what was used in the

country case studies. Additional terms can be added to equation (1) if the

disutility of effort and migration effects are considered important. The

country case studies assumed (a) a lower bound for CF. equal to an estimated

reservation wage and (b) that migration was not important in the agricultural

sector.

-10-

Accounting Price of Land

22. Not unlike labor, land is a basic depository of value and productivepotential for agriculture. Its value is difficult to assess since it is notonly a production factor, but also an asset and related markets are oftenthin, unreliable and complex. On the one hand, land markets in developingcountries are not very active as land in the hands of poor people is scarceand they tend to hold on to it as the only secure means of subsistence. Onthe other hand, the market for land services tend to reflect often complexland tenurial institutions and distortions in the labor markets, so thatrental rates are not necessarily related either to the price of land or to itsmarginal products.

23. Because land is also held as a hedge against inflation, and a pre-cautionary reserve of wealth, land accounting prices should take into accountthe availability of markets for alternative assets, the structure of theinterest rates and the rate of inflation. Land taxation is often desirable todiscourage (i) underutilization of productive land, (ii) rent accumulation,and (iii) use of landownership as a base of political power. Although difficultto enforce in practice, land taxation may also be an effective instrument toappropriate agricultural surplus. It can also be made neutral as betweencompeting uses, thus avoiding distortions between different cropping patterns,and it can be made progressive and more equitable.

24. The "reference" accounting price for land should be, as in the caseof labor, the opportunity cost given by its marginal product evaluated atborder price plus a component reflecting its function as a depository ofvalue. If renters save more than workers, this component may be positive whenthe government is committed to growth and higher savings are at a premium overconsumption. Income distribution weighting of the marginal product and theasset component, however, may induce substantial modifications in the overallresult.

Accounting Price of Capital

25. While in project appraisal the accounting price of capital is basicallyused to discount future benefits and costs, when we examine the more generalpricing issue, several other possibilities arise. First, an accounting interestrate may be useful to evaluate the interest rate prevailing in domestic marketsas a consequence of government fiscal and monetary policy. Second, it may beused to judge the optimality of foreign and domestic borrowing by the government.Third, it may be used to evaluate the productivity of the investment undertakenby the government.

26. The accounting rate of interest (ARI) is defined as the rate of fallover time of the value of public sector income measured at border prices. Ifsavings and consumption are considered to have equal value and income distri-bution considerations are neglected, such a value is the same as the marginalproduct of capital of government investment (q) measured at border prices:

ARI = qa (2)

where:

q = marginal product of capital of governmentinvestment at domestic prices

a = a border price conversion factor for investment.

More generally, however the ARI is calculated applying the formula.

Av = ARI - CRI (3)

where CRI is the consumption rate of interest and measures

how much more valuable is a unit of consumption now rather

than a year hence, and Av is the rate of fall of the value

of the public sector income at border prices relative toprivate consumption. Since CRI expresses the rate of fall

in value of private consumption over time, expression (2)

is an identity as it simply states that the rate of fall

of the value of public sector income relative to private

consumption is the difference between the rate of fall of

the value of the first variable (i.e. public sector income)

minus the rate of fall of the value of the second variable

(i.e. private consumption).

27. In practice the CRI is estimated using a formula based on one

objective parameter, the growth rate in per capita income g, and two subjective

ones: the elasticity of marginal utility of income n and the rate of pure

time preference e:

CRI = ng + e (4)

Together these three factors fully account for the rate of fall of the welfare

value of private consumption over time since "(a) if people are richer in the

future, the additional rupee of consumption will not mean the same to them" 1/

and, (b) people attach a premium to enjoying things today rather than in the

future, even if their income stays the same.

28. As for the other terms of expression (2), q. is measured as the rate

of return on the "marginal" government investment, while ARI is estimated as a

residual. The ARI, of course, is the rate of discount which should be applied

to all investments whose inputs and outputs have been evaluated at the account-

ing prices illustrated above. In economy-wide terms, it also represents a

1/ Hansen op. cit. p. 102

-12-

long-run rate to which the domestic interest rate may be compared to judge itsoptimality.

29. Use of the ARI, of course, is specially relevant where the discount-ing of costs and benefits can be done. In analyzing the effects of agriculturalprice interventions, however, where the focus is a broad sector or subsector,one cannot discount since the allocative problem concerns the use of existingresources, rather than any significant additional investment. The countrycase studies used, amongst other measures, Bruno/Balassa domestic resourcecost coefficients (DRCs). Here no discounting has been used in practice, andhence the DRC is a time-collapsed coefficient. As Hansen puts it: "thecapital costs of a project can usefully be viewed from two perspectives forshadow pricing. For lack of better terms, let us call them the asset and therent components. When Rs. 100,000 is invested in project X, for example, twothings happen. First, Rs. 100,000 of financial resources is converted intoreal physical assets. Secondly, the investor removes Rs. 100,000 worth offinancial resources from the national pool of savings that might be used forinvestment in alternative projects. Once invested in project X, therefore,these assets should yield a benefit, or rent, at least equal to what theywould have otherwise earned." 1/ There is no problem of valuing the realassets at their border price or border price equivalent values. From therental point of view, the economic cost of the capital, ignoring any premiumon savings, is the marginal product of capital. This is typically not easy toestimate in practice.

30. While a basic reference mark for the evaluation of interest ratepolicies is given by the opportunity cost of capital, the difficulties associatedwith differential inflation rates therefore suggest that more definite indica-tions on interest rates would be given by farmers' implicit discounting behaviorand their implicit or explicit willingness to pay. Equivalent subsidies andtaxes should be more appropriately judged with reference to these endogenousrates.

31. Since determining aggregate opportunities from trade does notdepend, per se, on the welfare function of the country, social considerationsare of no consequence in determining the required system of primary "accounting"prices. As in cost-benefit analysis, however, this system should be viewedonly as a convenient point of reference and not as a normative yardstick to beapplied under all circumstances. Thus, social prices can be used to formspecific goals of, say, growth or equity, and the divergence between them and''opportunity" border prices will directly give a measure of the equivalenttariff (or subsidy) needed for enforcement. Note, however, that unlike thecost-benefit analysis project case, economy-wide social pricing impliesknowledge of both the consumption pattern of the different income groups andtheir sources of income.

1/ Hansen, Ibid. p. 40.

-13-

IV. MEASURING MARKET DISTORTIONS: SUMMARY MEASURES

31. Once a system of accounting prices was worked out, the countrycase studies estimated four basic reference coefficients: the nominal pro-tection coefficient (NPC), the effective protection coefficient (EPC), the

effective subsidy coefficient (ESC) and the domestic resource cost (DRC).In most cases, the ESC was not used for reasons which will be explained later,

while in Egypt the net nominal protection coefficient was also calculated.

In two cases--Egypt and Yugoslavia--these coefficients were supplemented by

estimating producer surplus equivalent (PSE) and consumer subsidy equivalent(CSE) coefficients as in a pilot FAO study. 1/

Nominal Protection Coefficient (NPC)

32. The nominal protection coefficient (NPC) of any commodity is the

ratio of its domestic producer price to its border price. 2/

NPC = i ... ... (5)pb

where NPC, = nominal protection coefficient of the ith commodity

pd = domestic price of the ith commodityi

pbi = border price of the ith commodity, with the borderprice being its foreign price times the official rate ofexchange.

1/ See T.E. Josling, "Agricultural Protection and Stabilization Policies:

A Framework of Measurement in the Context of Agricultural Adjustment",

FAO Mimeograph, October 1975.

2/ Strictly speaking, it should be worked back to the appropriate producer

point, i.e., adjusted for internal transport and distributive margins,converted to their border price equivalent values, but in practice this

is not usually done as the differences are unlikely to be great.

-14-

33. The difference between the domestic producer price and the border

price of a comparable product is the ad valorem tariff rate 1/ (calculated

as a percentage on c.i.f. prices), where the product is not subject to

quantitative restrictions. Where there are quantitative restrictions,

domestic producer and border prices have to be related directly. Where

there are differential indirect taxes, the differential tax rates, expressed

as percentages of c.i.f. prices, have to be added to the tariff rates.

Thus, if the tariff is 20% of the c.i.f. price and the indirect tax is the

same for the domestically produced as for the imported commodity, the NPC 1.20.

But, if the indirect tax is 10% on the domestically produced commodity and 25%

on the imported commodity, the NPC - 1.20 x 1.25 = 1.36.1.10

34. The NPC can also be expressed as a percentage difference between

domestic and border prices, in which case it is called the nominal rate ofprotection (NPR):

pd pb

NPR. = i1 (6)pb

i

35. Before defining the EPC and ESC, it is necessary, first, to define

what is meant by "traded" and non-traded" and by "value added."

Traded Goods

36. Traded goods are those which are either directly imported or

exported or whose domestic sale (purchase) results in the goods being

exported (imported) by some person or firm. They are to be valued at theirf.o.b. export (c.i.f. import) prices so long as the exports (imports) are

not thought to affect the foreign demand price (foreign supply price).

Where, however, an increased supply of exports would, at the margin, affect

the foreign demand price or where an increased demand for imports would,

at the margin, affect the supply price, i.e., the demand and/or supply are

less than perfectly elastic, the appropriate border prices to use are the

projected margin export revenue or the projected marginal import cost.

Non-Traded Goods

37. All those goods which are not traded, are to be broken down intotheir traded and primary non-traded components, with the traded componentsvalued at border prices and the primary non-traded goods valued at their

opportunity costs, and converted into border price equivalent values usingappropriate conversion factors (see below).

1/ The expression "tariff" used here includes the ad valorem equivalent

of specific duties.

-15-

Value Added

38. Value added is normally defined as the value of the output at anypoint in the production-distribution process in any period, less the value of

the purchased inputs in the same period, less depreciation. It should be

noted that, as normally defined in national income accounting terms what is

subtracted from the value of the output are the purchased inputs which can be

both traded and non-traded. While one can redefine value added to be the sales

value less the value of all traded inputs, with the residual being the value ofall non-traded inputs, where does depreciation fit into this definition?

Depreciation might represent assets which are both traded and non-traded.

Strictly speaking depreciation should be split up into its traded and non-traded

components, but in practice, as in the case studies, it is typically treated as

traded. Using the traded/ non-traded definition, value added is equivalent to

the net return to non-traded resources used in producing a good or providing a

service plus (minus) any surplus (deficit) over and above "normal profit". In

equilibrium and constant returns to scale this decomposes into the marginal

value of land, labor and capital: otherwise the decomposition will include an

economic surplus (deficit).

39. Since value added is a residual concept, clearly what is purchased,and hence the value added of any commodity, will vary according to the time

period being considered and the level of analysis, e.g., whether viewed fromthe point of view of the country as a whole, the sector or the farm. It is

usual in most analysis to take a year (or a crop season); what is purchased

at the farm level might form part of value added at the sector level, and

what is purchased at the sector level might form part of value added at the

national level.

Effective Protection Coefficient (EPC)

40. The effective protection coefficient (EPC) measures the effects of

protective measures not only on traded outputs but on traded inputs or, toput it another way, on value added. It is measured by the ratio of value

added expressed in domestic market prices to value added expressed in border

prices:

EPC = Va

Vab

I/ This ratio can also be put in a percentage form:

Vad Vab

a_ _ x 100

Vabi

in which case it is called the effective rate of protection(EPR).

-16-

where EPC, = effective protection coefficient in the ith activityor commodity

dVa1 value added per unit of output in the ith activityor commodity at domestic prices

bVa1 = value added per unit of output in the ith activity orcommodity at border prices

This decomposes into its output and input components:

d k dp E (a ~P)

EPCi b -k b (8)P1 - (a P;)j J=1 ij i

where ai = quantity of the Jth input used to produce one unitof the ith output

P() domestic price of output/input for the ith commodityim or jth input

Pb( ) = border price of the output/input for the ith commodityi(j) or jth input

J = l...k = all traded inputs, direct and indirect

41. If the good which the denominator of (8) refers to is nottraded, but is potentially tradeable, the question to ask is whether or notit is likely to be traded in the future. If it is, then it should betreated as a traded good and some estimate made of the c.i.f. (or marginal)import or f.o.b. (or marginal) export prices; if it is not, then it shouldbe treated as a non-traded good and decomposed into its traded and primarynon-traded components (see paragraphs 36 and 37 above).

42. An EPC > 1 means that, at the existing official exchange rate, pro-tection measures provide positive incentives to produce the commodity orcarry out the activity under consideration, and vice versa, an EPC < 1 indi-cates that protective measures discriminate against the commodity in ques-tion. An EPC < 0 signifies an absolute loss of foreign exchange to theeconomy. It is important to point out here that an EPC less (greater)than one is not necessarily an actual disincentive (incentive), only apotential one.

-17-

Effective Subsidy Coefficient (ESC)

43. The ESC is the EPC adjusted for subsidies. It is calculated byadjusting (+) the numerator of equation (8) by the sum of the differencebetween profits tax, interest and the prices of non-traded goods actuallypaid and what is considered to be "normal" charges, i.e. the value of directand implied subsidies.

Domestic Resource Cost (DRC)

44. The domestic resources cost coefficient (DRC) is an adaptation ofproject evaluation methodology to country-wide policy evaluation. It is obtainedby computing the cost of foreign exchange saved for imported competing goods andthe cost of foreign exchange earned for exports. Traded final goods are thusevaluated at border prices. Non-traded goods are decomposed into traded inputsand non-traded primary factors (domestic resources). The former are evaluatedat border prices while the latter are evaluated at accounting prices. DRC's arethen measured as the ratios between the cost of domestic resources (evaluated ataccounting prices) and net foreign exchange earnings (value of traded ouputminus value of traded inputs).

45. DRC's can be interpreted as ratios between the cost of a dollarearned or saved through domestic production and an accounting rate of exchange.

For a given commodity i, denoting by Pi its border price, domestic resource

costs (value added) evaluated at accounting prices at a given stage of fabrica-

tion by V, the value of imported input per unit of output by Nb and the

input requirement coefficient by ajiV the cost of a dollar earned or saved

(C i) can be written as:

j V; aji

ci = pb E Nb ~~~~~~~~~~~~~~~(9i- E N aij

46. For example, if the cost of domestic resources is 400 cruzeiros, thecommodity border price 25 dollars and the expenditure on imported inputs is5 dollars, the domestic resource cost of a dollar in the i-th industry willbe 20 cruzeiros. If the accounting exchange rate is 18 cruzeiros to the dollar,the industry can be considered noncompetitive, in that the cost of a dollarearned through its operation exceeds the rate of substitution between domestic

currency and dollars (the accounting rate of exchange) by 11%.

47. According to Bruno, the originator of the DRC measure, the cost of

foreign exchange criterion "clearly measures comparative advantage". At thecore of the criterion, however, lie two basic estimates of values: (i) theaccounting prices of domestic resources, and (ii) the accounting exchange rate

to be used to convert these prices in units of international currency. Because

-18-

the DRC measures concern entire sectors of production rather than marginalproducts, accounting values of resources and of foreign exchange are notindependent of the composition of production. For example, the accounting priceof unskilled rural labor in rice production may be defined as the value ofthe production foregone in the next best alternative by a marginal unit oflabor employed in rice sector. Ideally, this value would be unique, sincemarginal productivities of domestic resources of all sectors should be equalizedby competition in labor markets. In practice, however, marginal productivitiesof labor, land and capital vary widely across sectors. They also obviouslydepend on the relative size of each sector.

48. For all policy changes of significant proportions, therefore, account-ing prices of factors and shadow exchange rates should be calculated forthe situation existing before and after the suggested policy changes and thecorresponding two sets of DRC's considered to evaluate the policies. As insimilar index number problems, this two-fold evaluation would be conclusivein some cases (where the DRC's before and after the change offer the samequalitative assessment for an expansion or contraction of a given sector)and inconclusive in others.

49. From the practical point of view, the definition of the DRC suffersalso from the problem that it is not always clear what are domestic andforeign resources. A more useful dichotomy to use is that between tradedand non-traded, because this dichotomy is important in the valuation ofinputs and outputs. This is used in the equations given below. Two formulaewere given to the researchers carrying out the six country case studies:those given in ejuations (10) and (11) below. In equation (10) the valueof the numerator is expressed in domestic prices and the denominator inforeign prices.

JdE a . mW . p

DRCi j=k+ k _J (10

i j E1 (a; Pf i 1 jk

where MPPY. marginal physical product of the jth input in its(weighted) average alternative use (y) 1/ adjusted asnecessary to reflect other factors as in equation (1)for labor and in paragraph 24 for land.

P = domestic price of the yth output (average alternative use)y

Pf = foreign price of the ith output

1/ See Appendix (1) for the interpretation of the weights.'

-19-

Pjf = foreign price of the jth inputjj = 1.. .k = inputs of directly traded goods plus the traded

elements of non-traded goods after decomposition.

j = k+l...J = inputs of primary, non-traded factors--includesthose obtained as a result of decomposition ofnon-traded goods.

kE a.. . P in the denominator is the unit foreign valuej=k j j

of the directly traded inputs, plus the indirect, traded elements of non-traded items left after decomposition, i.e. the foreign exchange value of alltraded costs per unit of output. Subtracted from the unit value of the tradedgood and multiplied by the output level gives the net foreign exchange earnings.

JJ a MPPY p d

j=k+l aij * ; y in the numerator is the domestically priced, oppor-

tunity cost unit value of the direct, primary, non-traded factors, plus theindirect, primary, non-traded elements of non-traded items left after decom-position. When multiplied by the output it gatves the total value of the non-traded, primary domestic resources used.

50. Equation (10) gives us implicit foreign exchange rates for specificactivities, which can be compared directly with an accounting price of foreignexchange. This price, however is not unique as it can be seen by consideringa second formula for the DRC based on a decomposition of nontraded goods intotheir tradable components:

E a.. * MPPY *P

DR.-J=k+l 'J 1 y (1

pb pbPi £(a ij* )

51. The only difference between equations (10) and (11) is that in (11)the prices of both the numerator and the denominator are border ones, whereasin (10) the prices in the numerator are domestic and the prices in the de-nominator are foreign. By comparing equations (11) and (10), however, wecan see that the accounting price of foreign exchange, to which the DRC in(10) has to be compared, will be sector specific. Such a price is simply given

by P /P , the ratio of the unit domestic value to the unit border value ofthe tunXle of goods that the resources employed in the ith sector would generatein their alternative use.

52. In both equations (10) and (11), MPPY is often difficult to estimatein practicg, anddin sur, cases consultants were iadvised to replace MPPY multi-plied by P or P by P. or P'J, respectively, where P! is the estimated value

y y j J

-20-

of the marginal product of the jth resource in its next best use, net of taxesand subsidies. In the absence of adequate information about marginal valueproducts, it was realized that market prices of the factor (resources) wouldhave to be used as a surrogate. Additional complications in estimating thenumerator of (11) can be solved by the use of a conversion factor. For example,if one of the goods is transportation, sixty percent of it can be consideredtradeable and priced accordingly by using a miscellaneous conversion factorbased on the border price of trucks and gasoline, while fourty percent canbe considered a nontradeable whose value is embedded in the wage payment tothe labor used.

53. It,ahould be noted in equation (10) that, if the MPPY . P isreplaced by Pj, and, if the resulting expansion is divided by tie acJounting

exchange rate, we have the Balassa formula used in his Western Africa Study. 1/The Balassa formula gives a pure number: A DRC <1 means that the domesticproduction of the commodity would be economically unprofitable and vice versa.

54. Equation (10) also gives us a pure number. It gives us the borderprice value of the resources (factors) in their (weighted) average alternativeuse per unit of border priced value of the resources in their existing use.If the number is greater than one, it tells us that the resources would be putto better use in an alternative way, and, if less than one, that this is arelatively sound use of resources.

Net Economic Benefit

55. Another way of looking at comparative advantage in terms ofeconomic efficiency is to appraise the existing use of resources in agricultureusing a simplified form of the L-M/S-T or the UNIDO methodologies. The use ofthese methodologies in fact makes clear that the name "comparative advantage"foreshadows at least two different questions. First, as we have indicatedbefore, one can ask whether the resources used in broad sectors or sub-sectorsof the economy could have been put to more profitable use elsewhere. Second,we can ask whether it would be profitable to expand the production of a par-ticular commodity (or a bundle of commodities) and to what extent it might bedesirable to do so. In practice this entails the appraisal of a specificmarginal investment whose ultimate effect is to increase the size of the sector(i.e. both its resource use and its production).

1/ Balassa in fact used two formulae--one where the foreign prices were con-verted into border rupees by applying the official exchange rate, and asecond where the "shadow" price of foreign exchange was used to convertforeign prices to domestic rupees.

-21-

56. Because the second question is likely to presuppose a positive answer tofirst and because it really addresses the issue of project evaluation ratherthan sector-wide comparative advantage, we concentrate here on the firsttype of question. In this case it would be inappropriate and probablyimpractical to consider the discounted value of a specific stream of benefitswhen looking at the comparative advantages of broad sectors or sub-sectorsand evaluatin the comparative advantage of the existing stock of agriculturalinvestment. L-/ Rather, the situation is similar to what in project analysisis referred to as "full development" net project benefits. In other words,we consider the present make-up of a productive sector as the result of alarge number of past investment decisions and we ask whether now, at fulldevelopment of those many investment projects, total benefits exceed totalcosts (including capital consumption allowances or as they are currentlyreferred to, depreciation). Equation (12) summarizes this approach:

i i j-l (ai k+l (aij MPP p b (12)

1/ Similarly, the converse applies: when analyzing the effects of amarginal investment, additional to the capital stock, the differentialincidence of benefits and costs and the time values of money are important.This may be one reason why the DRC has not been used in cost benefit analysis.

-22-

where Bi = net economic benefit (NEB) per unit of output

bPi = accounting value at border prices of one unit of output

a.. = quantity of the jth input need to produce one unit of theith output

MPPy = marginal physical product of the jth input in its averageJ alternative use (y)

pb = border price of the jth input

Pb = border price of the yth output (average alternative use)

j=l. .k = inputs of directly traded goods plus the traded elementsin non-traded goods after decomposition.

j=k+l ..j = inputs of primary, non-traded factors--includes thoseobtained as a result of the decomposition of non-tradedgoods.

56. (cont.) It can be seen that the only difference between equations (12)and (11) is that the different elements of each equation are differentlyarranged, but this is not so for equation (10), where different units areused in the numerator and denominator. While equations (12) and (11) areformally identical, they could be different in practice because of thearbitrariness involved in the tradeables versus non-tradables classification.Going back to the example of transportation, depending on how much ofthis "complex" input is considered tradeable or non-tradeable, the DRC in(11) may be greater or smaller than 1 and the conclusions on comparativeadvantage mwv turn out to be quite different for different classifications.No such ambi!quities arise from the use of the Bi measure.

Producer Subsidy Equivalent (PSE) and Consumer Subsidy Equivalent (CSE)

57. Producer Subsidy Equivalents (PSE) and Consumer Subsidy Equiva-lents (CSEs) were derived and first applied to agricultural products(wheat, barley, maize, rice, sugar and dairy products) for some of thedeveloped countries (Australia, Canada, European Community (six), Japanand the United Kingdom) where trade in the commodities is important.The FAO study 1/ showed that over the period 1968 through 1974, whichincluded periods of both high and low prices, producers received consid-erable net subsidies, while consumers received substantial negative sub-sidies or paid substantial net taxes. 2/ Although not all the six country

1/ Josling, op. cit.

2/ In the period of relatively high prices, developed countries tookadvantage of the rise to reduce the subsidies to producers and thetaxes on consumers.

-23-

case studies of the World Bank research project calculated PSE, and CSEs,

the general picture in LDCs is the reverse of the more developed countries:PSEs are negative and CSEs are positive, i.e. producers are taxed and

consumers subsidized, and often the net taxes on producers and the net

subsidies to consumers are high. There are some indications, however, that

the world food shortage, together with the relative failure of industry-

led development strategies, have caused some LDCs to rethink their agricul-

tural pricing, tax and subsidy policies.

58. The PSE is the subsidy, net of indirect taxes, given to producers,

expressed as a percentage of market value (price x quantity per annum) of

each commodity. In estimating the market value, care has to be exercised to

ensure that direct payments to producers are added onto the market value. The

CSE is the subsidy, net of indirect taxes, given to consumers, expressed as a

percentage of the total consumer value (volume of consumption x retail price).

Since PSEs and CSEs can fluctuate considerably with changing price levels it is

desirable to estimate them for a number of years to show trends and to graph the

results. Having done this, ideally, it would be desirable to update the

calculations yearly.

59. PSEs and CSEs have the advantage that: (a) they portray vividly and

simply the net effect of taxes and subsidies on producers and consumers of

agricultural products; (b) they can be estimated fairly simply from national

income and public sector accounts. On the other hand, they have to be inter-

preted with caution because: (i) they measure only the explicit subsidies and

indirect taxes; and (ii) like other partial equilibrium measures, they take no

account of "own-elasticities" and "cross-elasticities" of demand and supply.

-24-

V. EFFECTS OF PRICE DISTORTIONS: A MORE GENERALPARTIAL EQUILIBRIUM METHODOLOGY

60. The methodology outlined in the previous section can be considered asimplified form of a partial equilibrium methodology derived from the theory ofconsumers' and producers' surplus. 1/ The latter was not employed in theindividual country case studies, but we have since applied it to the countriesconcerned. 2/

61. The basic analytic structure of the model is given by the follow-ing formulae. (In the large country case, the border price would be themarginal revenue converted into domestic currency units at the official exchangerate.) The general formulae measure the gains/ losses/welfare transfers inrespect of producers, consumers, and governments for any kind of taxes and/orsubsidies on products. Here we use the net difference between border anddomestic prices since border prices are our reference points.

(A) Net economic loss in production (NEL )

= 1/2(Qb - Qd b -d

= 1/2 t2 n Vp s

(B) Net economic loss in consumption (NEL )

- = 1/2 (Cb - Cd)(Pb Pd)

= 1/2 t2 fdw

(C) Welfare gain of producers (WG )

QQ (P - ) - NELd d b p

1/ See Currie, J.M., J.A. Martin and A. Schmitz, "The Concept of EconomicSurplus and Its Use in Economic Analysis," Econ. J. 81 (1971): 741-99.for an excellent review of the concepts, their uses and limitations.

2/ See also E. Lutz and P.L. Scandizzo, "Price Distortions in DevelopingCountries: A Bias Against Agriculture", forthcoming in the EuropeanReview of Agricultural Economics, and Price Distortions in Agricultureand Their Effects, World Bank Staff Working Paper No. 359, prepared byM.D. Bale and E. Lutz, October 1979.

-25-

(D) Welfare gain of consumers (WG )

= Cd (Pb Pd) NEL

(E) Change in foreign exchange earnings

= -Pb (Qb - Qd + Cd Cb)

(F) Change in government revenue

-- (NEL + NEL )-WG - WGp c p c

' -(A) -(B) -(C) -(D) -(E)

where:

Qb= production at border prices

Qd= production at domestic prices

V =Value of production at domestic prices

Cb= Amount of consumption at border prices

Cd= Amount of consumption at domestic prices

W = Value of consumption at border prices

Pb = border prices

Pd = domestic prices

iB = elasticity of domestic.supply

d= elasticity of domestic demand

t , t = proportion of net trade taxes in the domestic pricec ~ at the consumer (t ) or the producer (t ) levels

c p

62. The first two formulae can, perhaps, be more easily understood by

reference to Marshall's graphic illustration of tax effects, given in Figure 1.

-26-

Figure 1. EFFECTS OF A TAX ON CONSUMERS' AND PRODUCERS' SURPLUS

y

Price Si

D

P2 -- - -

Pi -___ - -- -,-r-

SI ~~l

xO 02 Q 1 Quantity

World Bank - 21231

-27-

S S ' is the supply curve prior to the imposition of, say, an export tax

equal to AG, which raises the supply curve to S S'. DD' is the demand curve.1 1

The loss in consumers' surplus is equal to the area PI P2AE. The loss in

producers' surplus is equal to the area FP EG. The tax (foreign exchange)

revenue is equal to the area FP 2AG. The net economic loss in consumption is,

therefore, equal to the area P 1P2AE - PP 2AC = AEC. The net economic loss

in production is equal to the area FP EG - FP CG = CEG. AEC and CEG are the

famous Depuit Triangles. For consumers' surplus, if the demand curve is a

straight line within the relevant range, the area of the triangle AEC = 1/2

ABEC = 1/2(Q1 - Q2)(P2 - P1). In terms of the notation used in Equation (2)

above Q-Q2 (Cd -Cb) and P2 -P = (Pb - Pd). So we arrive at

Equation (B) NELc =1/2(Cd -Cb)(Pb -Pd). But (Cd - cb)/Cd= -ndt where

t is the tax rate (Pb - )/Pd. Substituting ndt into (2) and multiply-

ing out we get NEL = 1/2 t n W, as in equation B.c c d

63. To obtain an alternative explanation of the same formulae, note that,

assuming that prices equal average costs (either because of long run technical

efficiency, input-output technology or as an approximation),the average increase

in cost of production following a tariff is (Pb d (Qd Qb)- The foreign2d b

exchange gained because of the same production increase, however, is only

Pb (Qd - Qb)- The net efficiency cost is then the difference between the two

terms which equals NEL p- On the consumption side, we can make an analogous

computation by noting that the average fall in consumption value as a con-

sequence of the tariff is b d (C - C ) while the foreign exchange savings2 b d

amount to only Pb (Cb - Cd). Again substracting one term from the other we obtain

the net efficiency loss in consumption NEL . As for welfare changes, producersc

-28-

gain is given by additional foreign exchange earnings Pb(Q d Qb) plus increases

in revenue Q(Pd P minus the average increase in cost (Pb d (Qd2

or Qd(Pd - P b NELp. An analogous reasoning holds for consumers.

64. The net welfare loss of a market distortion for a traded goodsthus depends on two main factors: (i) the trade (import demand or exportsupply) elasticity and (ii) the net amount that would be traded without thedistortions. Because import demand and export supply elasticities are aweighted average of supply and demand elasticities, with a negative weightassigned to the demand elasticity, their value tends to be large even forcases where the absolute value of demand and supply elasticities appear tobe low. As a consequence, if the amount exportable (importable) at world pricesis significant, the welfare loss from an export (import) tax may indeed growlarge.

65. To give a further idea of the orders of magnitude involved, Table 1presents value of the welfare loss from a 30% tax as a percentage of the valueexportable at world prices for a range of elasticity values and under theconservative assumption that 30% of domestic production would be exported atworld prices.

Table 1: NET WELFARE LOSS INDUCED BY A 30% EXPORT TAX

(Percentage of the Value Exportable at World Prices)

DemandElasticity Supply Elasticity

0.2 0.4 0.6 0.8 1.0 1.50.4 0.07 0.10 0.13 0.16 0.19 0.270.6 0.09 0.12 0.15 0.18 0.21 0.290.8 0.11 0.14 0.17 0.20 0.23 0.311.0 0.13 0.16 0.19 0.22 0.25 0.331.5 0.19 0.22 0.25 0.28 0.31 0.38

The magnitudes indicated in Table 1 are expressed as a percentage of thevalue of the commodity that the country would be able to export under freetrade. If this value is large, the cost of the export tax may be largeindeed. If the country accounts for a large market share of the commodityin question, on the other hand, an export tax may raise world prices andtherefore be beneficial for the country. In both cases, however, transfersof welfare would be high.

-29-

66. While one may conjecture that welfare costs of present agricul-

tural policies are substantial, clearly ad hoc analysis is needed to appraise

the efficiency losses of any particular case. Despite the limitation inherent

to any partial equilibrium model in respect of income distribution and effects

on other sectors of the economy, the elasticity framework has been found

useful in analyzing the impact of market interventions on prices, quantities,

government revenues and consumers and producers welfare. This is especially

true where (i) the good considered does not affect a large part of the

economy, and (ii) producers and consumers may be treated as separate classes

of economic agents.

67. Some interesting expansions of partial equilibrium analysis concern

a number of factors that have traditionally been considered as falling out of

its domain. These are: (i) market failures (monopolies, increasing costs,

etc.); (ii) risk and uncertainty; (iii) multi-market effects and (iv) income

distribution. Measurement of the welfare impact of the distortions in these

special cases requires, in addition to demand and supply elasticities, quan-

titative information on the distortions that are the cause of market failure

(for example, the degree of monopoly power as expressed by the elasticity of

the demand curve facing the monopolist), on the sources of uncertainty and on

the income distribution mechanism. The modifications required to compute the

impact of the distortions under such different conditions are not great. In

general (See Appendix IV) they mainly consist of correcting the values of the

elasticities by appropriate factors reflecting the additional market char-

acteristics taken into account.

68. In all cases, the basic parameters needed to evaluate the impact

of price interventions are the elasticities of supply and demand. These para-

meters quantify the responsiveness of producers and consumers to price move-

ments and provide the basis for a forecast of production and consumption

changes induced by the market intervention. Numerous studies have attempted

to make econometric estimates of demand and supply elasticities and a survey

of these attempts is contained in Appendices II and III. Tables 2 and 3

summarize some of the most recent restimates for a sample of developing coun-

tries.

69. The evidence on agricultural supply elasticity is unfortunately both

biased and weak. Most of the successful attempts at estimating behavioral

supply functions concern acreage response equations. An indirect account of

the response of other factors such as capital and labor through yield response

functions has been tried much less successfully by econometricians. Production

function and more recently, profit function analysis has been used to account

directly for response in factor use and substitution. This analysis suggests

that supply elasticities may be very high in agriculture, since the response

to price increases is particularly strong in the non-land variables: labor,

fertilizer, pesticides, etc.

70. While the evidence is diverse, most of the successful attempts

report estimates of acreage response in the range of 0.1 to 0.8 for short

run elasticities and 0.3 to 1.2 for the long run ones. 1/ Yield responses

1/ See Appendix II.

-30-

Table 2: PRICE ELASTICITIES OF SUPPLY IN DEVELOPING COUNTRIESBY NUMERICAL RANGE

Countries or RegionsFood Less 0 1/3 2/3 More

Category than 0 to 1/3 to 2/3 to 1 than 1

Agricultur.l Products ArgentinabIndia

Rice India India India Taiwan W. MalaysiaaEgyptb,c Pakistan Peru Iraqa

Bangladesh Indonesia(Java) ThailandThailand Iraq (Thonburi only)W. MalaysiaPhilippdnesEgyptD,'

Sri Lanka

Wheat India India India India SyriaaIraq Pakistan Pakistan Egypt' New Zealar.d

Hungary Kenysa d ChileaJordan EgyptLebanon SyriaEgyptb,c Lebanon

ArgentinaChile

Other Foodg rains India India India India IndiaaSyria Pakistan Hungary Lebanon Thailand5

Iraq Lebanon Sudan Kenyaa SyriaaJordan Egypt Lebanon Kenya bEgypt Sudan Jordan Brazil

Philippines SyriaHungary BrazilJordan

Vegetables & Spices Syria India India India IraqbJordan Jordan Lebanon SudanaLebanon EgyptPakistan Syria

Bangladesh

Milk & Meat:s Greece Greece Greece

Sugar & Oi:s India India India IndiaNigeriaPhilippines

Coffee, Tell, Cocoa Pakistan Brazil Brazil Brazil Brazil& Tobacco India Nigeria India Nigeria

Bangladesh Malawi Jamaica CameroonEcuador Ghana GhanaDom.Republic Venezuela Ivory CoastVenezuela CameroonNigeria ColombiaColombia NigeriaKenya

Cotton, Jute, Egypt India India Egypt IndiaSisal & Wool Iraq Uganda Uganda Pakistan Tanzania

Tanzania Tanzania S. AfricaBangladesh Nigeria Uganda

Su4an

a - Long-run elasticity.b - Short-run elasticity.c - Pre-World War II.d - Post-World War II.

Sources: 1) Askari, H. and J.T. Cummings, "Estimating Agricultural Supply Response with theNerlove Model: A Survey", International Economic Review, Vol. 18, No. 2,June 1977, pp. 257-292.

2) Sobhan, I., Agricultural (Price) Policy and Supply Response - A Review of Evidenceand an Interpretation for Policy, World Bank, AGREP Division Working Paper, mimeo,July 18, 1977.

-31-

Table 3: PRICE ELASTICITIES OF DEMAND IN DEVELOPING COUNTRIESBY NUMERICAL RANGE

Food Count':iesCategory Less than 1/3 1/3 4 2/3 2/3 - 1 More thain 1

Total Food (exp.) Ireland** Chile*** Panama I Ceylon** IKorea (urban) Greece*** Nigeria Dom. ReniblicRhodesiai Hondurasi Ecuador

Jamaica Peru***Korea (rural) Thailand"**PhilippinesPuerto RicoS. AfricaTaiwanW. PakistanIsrael***Yugoslavia

Foodgrains Bangladesh** India***

Total Cereals India*** Ghana*

Rice W. Pakistan Argentina** India*** Ghana***Bangladesh

Wheat InliaIW. PakistanArgentina

Rice & Wheat Bangladesh***

Pulses W. Pakistan India***

Other Foodgrains Argentina Ghana*

Vegetables & Fruits Pakistan Argentina**

Meats Nigeria*Argentina**Pakistan

Milk & Milk Products PakistanArgentina**

Sugar & Oils Pakistan

Coffee & Tea PakistanArgentina**

* Statistically significant at 30% level.** Statistically significant at 10% level.

*** Statistically significant at 1% level.I Statistically insignificant.

Source: Extracted from Appendix III, Table 1.

-32-

tend to have lower ranges and reported estimates appear much less reliableand numerous. While reported attempts may be biased, there seems to be littledoubt that: (i) farmers appear to respond to price signals, and (ii) theirresponse is conservative in the short run but tends to build up and to becomeconsiderable if the price variation persists for a sufficiently long periodof time.

71. The elasticities estimated through the production function approachtend to be high, particularly if profit maximization assumptions are used withfunctions showing only minor decreasing returns to scale. For the Cobb-Douglascase, for example, output supply elasticity is equal to the ratio between thecoefficient of scale and its complement to one 1/ so that it tends to infinityas returns to scale approach one. If a major factor of production such as,for example, capital or land for small farmers are assumed to be given,however, even with constant returns to scale supply elasticities would fallto the level shown by direct acreage and yield estimates.

72. On demand elasticities, the evidence, also diverse, points to arange of values not very far, in absolute terms, from the estimated values ofsupply elasticities. In general, basic food items (foodgrains, pulses,staples, etc.) show own price elasticities in the range of 0.2 to 0.7; highincome items such as meat and fruits show higher elasticities often in excessof 1, and special items such as coffee, cocoa and sugar show very low elasti-cities.

73. The low measurable values of supply and demand responses to pricechanges tend to limit the damaging consequences of price distortions in agri-culture. But despite these low values, the size of the present distortionsis such that, according to our estimates, a production shortfall in the rangeof 10% to 50%, depending on the country, is likely to be the effect of thepresent direct and indirect taxation policies in agriculture.

74. Before reporting our own estimates, it is interesting to look atsome of the results of recent studies on economic losses due to distortions inagriculture. Table 4 briefly reviews some studies reported in *the AmericanJournal of Agricultural Economics. For all studies, net economic losses appearto be substantial, both in absolute and relative terms. Real effects andforeign exchange losses not reported in Table 4 were also estimated to belarge. For example, Valdes (1973) in his study of the trade policy of Chile,estimated negative effective rates of protection averaging: -.38 for wheat,-.35 for beef, -.30 for lamb and, -.46 for wool. Because of these rates, hefound that production was below equilibrium levels from 3 to 10% for wheat, from4.5 to 14.5% for beef, and from 6.8 to 23% for wool. The loss in foreignexchange earnings ranged from 24 to 39 million dollars, depending on the supplyelasticities. Accordingly, Chile's agricultural trade deficit would have beenreduced by 76% during 1951-55 and would have pratically disappeared during1956-60.

Eai1/ The formula is n i = where ai is the elasticity of production

with respect to the i-th input and ni is supply elasticity under profitmaximization.

-33-

75. We have investigated the effects of price distortions in agriculture

in two different ways: first, for a sample of countries for which detailed

studies of the tax and subsidy system were available, 1/ the partial equili-

brium methods illustrated above have been applied to evaluate the real and the

monetary effects of price intervention. Second, for a smaller sample of

countries, data provided by World Bank reports have been utilized to evaluate

more specifically the intervention policies on markets for foodgrains(including food distribution programs).

76. Tables 5 and 6 present the empirical estimates for the first sample.

These estimates are based on the elasticity ranges shown in Tables 2 and 3

and on the estimates of nominal protection coefficients provided by thereference material. 2/ Because of the limitations of the data base, the esti-

mates in Tables 5 and 6 represent average annual value 1973-75 for Argentina,Egypt, Portugal and Yugoslavia, average annual value 1974-75 for Thailand, and

annual values for Kenya and Pakistan. Data from FAO Production and Trade

yearbooks were used to compute production, consumption and trade figures.Using as a reference the intervals of variation tabulated in Tables 3 and 4,

plausible elasticity ranges were used for supply and demand elasticities. In

most cases supply elasticities were assumed to range from 0 to 1/3 or 1/3 to

2/3 whereas demand elasticities were assumed from 0 to -1/3 and from -1/3 to

-2/3.

77. In order to separate the effects of the currency overvaluation from

the one of the specific tariff, the nominal protection coefficients were also

adjusted by multiplying them by the following Bank estimated conversion factors:Argentina .8, Egypt .69, Kenya .76, Pakistan .91, Portugal .83, Thailand .88

and Yugoslavia .98. These adjusted estimates compare the effect of protectionon the individual crop in a second best situation, and permit measure of the

costs of market intervention whose removal is contemplated only for one speci-

fic sector. These estimates are presented in Table 7 and 8.

78. The real effects of the price distortions (Tables 5 and 7) are often

sizable. Since production and consumption effects have opposite signs they

are additive with respect to trade effects. Where no government subsidies areinvolved, price distortions cause a reduction in trade; on the other hand,export or import subsidies cause an expansion of trade amounting to the sum of

the absolute production and consumption adjustments. For Pakistan, forexample, under the hypothesis of high demand and supply elasticities, the

negative protection for wheat resulted in a reduced production of 1,599

thousand metric tons and increased consumption by 940 thousand metric tons.Thus, as compared to no price intervention, imports increased by 2,539 thousand

metric tons. Since, in actuality, total imports amounted to 1,345 thousand

metric tons, a removal of the import subsidies would have caused Pakistan to

become a wheat exporter.

79. In Tables 6 and 8, the monetary effects of price distortions arepresented. The net economic losses in production and consumption depend cri-

tically on the extent of protection and on the elasticities. In some cases,

1/ See Appendix IV.

2/ For additional details see: E. Lutz and P. L. Scandizzo, op. cit.

Table 4 E2iPIRICAL STLDIES ON litE COST OF PP.ICE DISTORTION OF AGRICULTURAL COZIODITIES

(From the LQornal of Porm Economics or the American Jotirnal of Agricultural Economics)

Estima tedSubject Supply Domay,d Aninual Net 3/

Coraimodities Flasticities Elasticitics Social L-csses-Study (Sce Appendix ) Cotintrv 'Lo'u High Low Hieb Low Hi.,h Comments

ilailace (1962) Cocl;rane plan (US) .05 1.0 -.12 -.4G .06 6.5 In X of the valueBrannon plan (US) .05 1.0 -. 12 -. 40 .03 4,7 of farm output'

iohnson (1965) Tobacco (US) .2 .4 -. 4 -. 6 1 25 In % of the valueof production

Dardis (1967) Grains (Ui.) .27 .75 1/ I/ 1 2.5 Equivalent tariff rate of.01 .03 In % of GNP 22-25%

Sl:ape (1969) Sugar (developedcountries) 0 0 n.a. n.a. n.a. n.a. Losses in absolute terms range

from $176 to $1,175 million

Raw wool (US) 2/ 2/ -.5 -1 1 3 In % of the value of.rnn,;ron (' 569) ........... domeetic consumption

Cramn-atte/ Raw wool and wool 1 2 -1 -2 n.a. n.a. Losses in absolute terms ratcgei'ardis Y1970). p-oduce (US) from $68.5 to $130.0 million

.'a1des (1973) Beef, larbh, wool, .15 .5 -. 3 -. 3 n.a. n.a. Effects or. production, consumption*Wheat (Chile) and foreign exchange earnings estimated

Ea 1 e! Eigh: coammodities .12 1.61 -. 14 -1.23 .1 2 In % of GEP*,reenshi6lds (1978) (Japan)

'?org (1978) Rice (Thailand) ,4 .9 _.43 -.47 Slight net Large country assumption used,social gain exporc tax level 25-30%

; N-ot relevart for d._fici ency payr.:oitt aystam.2/ -ot appli.cable since producer prices are held cotnstant.,/ The tariff rates and the welfare effects on producers, consumers, and the government are not shown since only a few studies make reference to them.

-35-

the net economic losses in production and consumption exceed 10% of the value

of production or consumption, using the high elasticity assumption. While

elasticities are important for the determination of the social losses, their

impact on the welfare gains of producers and consumers is small. Thus, a

doubling of an elasticity e.g., from 1/3 to 2/3 seldom changes the amount of

the transfer by more than 10%.

80. The welfare transfers are in all cases much larger than the res-

pective welfare losses. In one third of the cases, the welfare gain (or loss)

of producers is more than 10 times as large as the net economic loss in pro-

duction using the high elasticity assumption. Since positive protectionexists for only three out of the 24 cases, the producers incur welfare losses

in 21 cases. Even though not all major commodities were taken into consider-

ation, this is indicative of how heavily penalized agriculture is in developingcountries. The sum of the welfare losses for producers are substantial even

in per capita terms. Analogously, consumers make substantial welfare gains in21 out of 24 cases.

81. All calculations imply a small country assumption. This was con-

sidered to be justified since the world market shares of the countries and

commodities in question are small. A case could be made for using the large

country assumption for cotton exports from Egypt because of the distinguish-able characteristics of long staple cotton. This however, would have no

effects on any of our estimates except on the government revenue which would

be increased by an amount of exports times the world market price increase ofcotton due to the reduction of exports.

82. There are several limitations of the above analysis. First, theperiod of analysis is between 1973 and 1975, at a time when world market

prices for some commodities were high in comparison with the long-run average.

This tends to overestimate gains and losses for commodities with negativerates of protection and tends to underestimate gains and losses where positiveprotection exists. Second, the assumptions on the ranges of the elasticitiesresult in corresponding ranges for the estimates of efficiency losses and not

in specific estimates which might be desirable. Third, no distinction wasmade between prices facing producers and consumers. Fourth, the effects on

factor prices and employment are not estimated.

83. The impact of policies designed to extend the distribution of key

foodgrains has been investigated in five Asian countries: Bangladesh, India,

Pakistan, Sri Lanka and Indonesia over a three year period beginning in 1974.A large percentage of the population in all five countries would have clearly

suffered from malnutrition if domestic prices were as high as the world pricewhich prevailed from 1974-75. Even at the relatively reasonable world price

levels of 1976-77, a sizable number in all five countries could not haveafforded to buy adequate quantities of foodgrains. During this periodBangladesh, India and Pakistan all operated ration shops that made low quality

foodgrains available at relatively constant and low prices to coupon holders

(in the case of Bangladesh and India), or to the entire population, while in

Indonesia, although rationing did not exist, domestic market prices were kept

from reaching the high world levels in 1974 and 1975 by the official release

of subsidized imports and stocks.

Table 5 GROSS REAL EFFECTS OF PRICE DISTORTIONS OF SOME COMMODITIES(in OOOs m.t.)

Estimated EstimatedChange in Change in Reduction in

Country/ Bordern/ Exports Production A/ Consumption Exports (Imports)Cownodity Price NPC EPC / Production Consumption (Imports) Low High Low High Low High

ArgentinaWheat 122 .54 .51 7,033 4,746 2,287 -1,977 -3,954 0 h,334 1,977 5,288Haize 107 .58 .57 9,100 4,618 4,482 -2,175 -4,349 0 1,104 2,175 5,453Beef 1810 .69 .64 2,254 2,096 158 -334 -668 311 622 645 1,290

Egypt Rice 428 .37 .38 6,966 6,787 179 0 -3,914 0 3,814 0 7,728Cotton 2439 .43 .43 441 207 234 0 -193 91 181 91 374Wheat 192 .57 .68 1,918 4,524 -2,606 0 -955 0 1,126 0 -2,081

KenyaMaize 113 .96 .92 1,600 1,479 121 -22 -44 0 20 22 64Coffee 1400 .94 .88 75 7 68 0 -2 0 0 0 2Beef 975 .76 .64 113 109 4 -12 -24 11 23 23 46

PakistanWheat 185 .84 .80 7,673 9,081 -1,345 0 -965 0 567 0 -1,531Cotton 777 .68 .67 510 311 199 -158 -240 48 97 207 337Rice 487 .75 .74 3,804 3,326 478 0 -418 0 366 0 784Sugarcane 351 .68 .67 21,242 21,242 0 -3,299 -6,598 0 3,299 3,299 9,896

PortugalWheat 199 .84 .75 566 873 -307 -36 -71 0 55 -36 -126Olive Oil 2335 .78 .64 325 321 4 0 61 0 30 0 90Maize 140 1.20 1.27 501 1,506 -1,005 28 55 0 -83 28 138Beef 1268 1.75 .45 85 112 -27 12 24 -16 -32 28 56Rice 407 .70 .64 140 180 -40 0 -20 25 51 -25 -71

ThailandSugarcane 351 1.19 1.24 13,965 13,445 520 736 1,472 -708 -1,417 -1,444 -2,888Rubber 605 .77 1.03 363 14 349 0 -72 1 3 1 74

YugoslaviaWheat 201 .69 .59 5,143 5,499 -356 -763 -1,525 0 815 -763 -2,340Maize 224 .78 .63 8,559 8,394 165 -797 -1,593 0 781 797 2,375Beef 2065 1.21 -.41 301 261 40 17 34 -15 -30 -32 -64Pork 2060 .93 1.71 374 383 -9 -9 -19 10 19 -19 -38

1/ In $ per m.t. border prices are country and commodity specific due to quality differences of the commodities andthe differences in transportation costs.

2/ Nominal protection Coefficient.3/ Shown for purpose of comparison.4/ Low indicates that the lower bound for demand and supply elasticity has been used. High indicates the upper bound.

Table 6 GROSS MONETARY EFFECTS OF PRICE DISTORTIONS(annual million US dollars)

Welfare WelfareNEL NEL NEL Gain of Gain of Change in Foreign

P c Producers Consumers Inc. in Exchange EarningsGov't

rountry/Cowmditv RPC low RHish T.ow Rieh *oU Usih Low-'/ Hih1/ LoU-/ HiRh!/ Revenue Lo!'/ High!/

ArgentinaWheat .54 55.5 110.9 0.0 37.4 55.5 148.3 -450.2 -505.6 266.3 228.9 128.3 -241.2 -645.2Maize .58 48.9 97.7 0.0 24.8 48.9 122.5 -457.8 -506.7 207.5 182.7 201.4 -232.7 -583.4Beef .69 93.8 187.5 87.2 174.4 181.0 361.9 -1,358.5 -1,452 1,088.9 1,001.7 88.6 -1,167.3 -2,334.7

Total 198.2 396.1 87.2 236.6 285.4 632.7 -2,266.5 -2,464.3 1,562.7 1,413.3 418.3 -1,641.2 -3,563.3

Egypt .304

Rice .37 0.0 527.7 0.0 514.1 0.0 1,041.8 -1,878.3 -2,406.0 1,830.0 1,315.9 48.2 0.0 -3307.4Cotton .43 0.0 134.1 62.9 125.9 62.9 259.9 -613.1 -747.2 224.8 161.9 325.3 _220.8 -912.2Wheat .57 0.0 39.4 0.0 46.5 0.0 85.9 -158.3 -197.8 373.5 327.0 -215.1 0.0 -399.6

Total 0.0 701.2 62.9 685.5 62.9 1,387.6 -2,649.7 -3,351.0 2,428.3 1,804.8. 158.4 -220.8 -4,619.2

KenyaMaize .96 0.0 0.1 0.0 0.0 0.0 0.1 -7.2 -7.3 6.7 6.6 0.1 -2.5 -7.3Coffee .94 0.0 0.1 0.0 0.0 0.0 0.1 -6.3 -6.4 0.6 0.6 5.7 0.0 -2.4Beef .76 1.4 2.8 1.3 2.6 2.7 5.4 -27.8 -29.2 24.2 22.8 0.9 -22.5 -45.1

Total 1.4 3.0 1.3 2.6 2.7 5.6 -41.3 -42.9 31.5 30.0 6.7 -25.0 -54.8

PakistanWheat .84 0.0 14.3 0.0 8.4 0.0 22.6 -227.1 -241.4 266.9 285.5 -39.8 0.0 -283.3Cotton .68 19.7 29.8 6.0 12.0 25.7 41.8 -146.5 -156.6 71.3 65.3 49.5 -160.6 -261.5Rice . .75 0.0 25.4 0.0 22.3 0.0 47.7 -463.1 -488.6 404.9 382.7 58.2 0.0 -381.9Sugarcane .68 185.3 370.5 0.0 185.3 185.2 555.8 -2,571.1 -2,756.4 2,385.9 2,200.6 0.0 -1,157.8 -3,473.6

Total 205.0 440.0 6.0 228.0 210.9 667.9 -3,407.8 -3,643.0 3,129.0 2,934.1 67.9 -1,318.4 -4,400.3 (.)

PortugalWheat .84 0.6 1.1 0.0 0.9 0.6 2.0 -18.6 -19.1 27.8 26.9 -9.8 -7.1 -25.1Olive Oil .78 0.0 15.5 0.0 7.7 0.0 23.2 -166.9 -182.5 164.9 157.2 2.1 0.0 -211.0Maize 1.20 0.4 0.8 0.0 1.2 0.4 1.9 13.6 13.2 -42.2 -43.3 28.1 3.9 19.3Beef 1.75 5.7 11.4 7.5 15.0 13.2 26.5 75.1 69.4 -4L4.O -121.6 25.7 35.3 70.7Rice .70 0.0 0.1 0.2 0.3 0.2 0.4 -17.1 18.3 20.4 18.9 -4.9 -10.4 -28.8

Total 6.7 28.9 7.7 25.1 14.4 54.0 113.9 -137.1 56.9 3R.1 41.2 21.7 -174.9

ThailandSugarcane 1.19 24.5 49.1 23.6 47.2 48.1 96.3 906.8 882.2 -920.3 -943.9 -34.7 506.9 1,013.8Rubber .77 0.0 5.0 0.1 0.2 0.1 5.1 -50.5 -55.5 1.8 1.7 48.6 -0.8 -44.9

Total 24.5 54.1 23.7 47.4 48.2 101.4 856.3 826.7 -918.5 -942.2 13.9 506.1 968.9

YugoslaviaVhest .69 23.7 47.5 0.0 25.4 23.7 72.9 -344.2 -367.9 342.6 317.2 -22.2 -153.3 -470.4Maize .78 19.6 39.3 0.0 19.2 19.6 58.5 -441.4 -461.0 413.7 394.4 8.1 -178.5 -531.9Beef 1.21 3.7 7.4 3.2 6.4 6.9 13.9 126.8 123.1 -116.4 -119.7 -17.3 66.5 132.9Pork .93 0.7 1.3 0.7 1.4 1.4 2.7 -54.6 -55.2 54.5 53.9 -1.3 -38.7 -77.5

Total 47.7 95.5 3.9 52.4 51.6 148.0 -713.4 -761.0 694.4 645.8 -32.7 -304.0 -946.9

1/ The "low" and "high" refer to the lovw and high elasticity assumptions and do not necessarily correspond to the lows and highs of the respective monetary effects.

Table 7 NET REAL EFFECTS OF PRICE DISTORTIONS OF SOME COMMODITIES

(in 000s m.t.)

Estimated EstimatedChange in Change in Reduction in

Exports Production Consumption Exports (Imports)Country/Commodity NPC Production Consumption (Imports) Low High Low High Low High

ArgentinaWheat .43 7,033 4,746 2,287 -3,077 -6,153 0 2,076 3,077 8,229Maize .46 9,100 4,618 4,482 -3,525 -7,051 0 1,674 3,525 8,725Beef .55 2,254 2,096 158 -566 -1,132 523 1,047 1,089 2,178

EgyptRice .25 6,966 6,787 179 0 -6,896 0 6,719 0 13,615Cotton .30 441 207 234 0 -340 159 319 159 658W4heat .39 1,918 4,524 -2,606 0 -1,980 0 2,335 0 -4,315

KenyaMaize .73 1,600 1,479 121 -195 -391 0 181 195 571Coffee .71 75 7 68 0 -10 0 1 0 11Beef .58 113 109 4 -27 -54 26 52 53 106

PakistanWheat .76 7,673 9,018 -1,345 0 -1,599 0 940 0 -2,539Cotton .62 510 311 199 -206 -313 63 126 269 438Rice .68 3,804 3,326 478 0 -591 0 517 0 1,107Sugarcane .62 21,242 21,242 0 -4,296 -8,593 0 4,296 4,296 12,889

PortugalWheat .70 566 873 -307 -80 -160 0 123 -80 -284Olive Oil .65 325 321 4 0 -116 0 57 0 173Maize .99 500 495 5 -2 -3 0 2 2 5Beef 1.45 85 112 -27 9 17 -11 -23 20 40Rice .58 140 180 -40 0 -33 43 86 -43 -119

ThailandSugarcane 1.15 13,965 13,445 520 601 1,202 -579 -1,157 -1,180 -2,360Rubber .68 363 14 349 0 -113 2 4 2 117

YugoslaviaWheat .64 5,143 5,499 -356 -955 -1,909 0 1,021 -955 -2,930Maize .73 8,559 8,394 165 -1,045 -2,089 0 1,025 1,045 3,114Beef 1.13 301 261 40 11 23 -10 -20 -21 -43Pork .86 374 383 -9 -20 -40 21 41 -41 -81

T.ile 8: ST M')SETARY EFFECTS OF PRICE DISTORTIONS

(anniual million US dollars)

Welfare Welfare

NPC NEL NEL Gain of Gain of Change in ForeignC C Producers Consumers Inc. in Exchange Earnings

Countrv/Commodity NPC* - Low High Low High Low High Lo-V High2/ Lo'2/ High2/ Revenue Lowe' High2/

Areentina:;heat .48 107.0 213.9 0.0 72.2 107.0 286.1 -596.0 -703.0 330.0 257.9 159.0 -375.3 -1,004.0

Maize .46 101.8 203.7 0.0 48.4 101.8 252.1 -627.6 -729.5 249.7 201.4 276.1 -377.2 -933.6

Beef .55 230.5 460.9 213.1 426.2 443.6 887.1 -1,937.7 -2,168.1 1,365.4 1,152.3 128.7 -1,971.4 -3,942.8

Total 439.3 875.5 213.1 546.8 652.4 1,425.3 -3,161.3 3,600.6 1,945.1 1,611.6 563.8 -2,723.9 -5,880.4

E: v?Rice .25 0.0 1,106.S 0.0 1,078.4 0.0 2,185.2 -2,236.0 -3,342-9 2,178.6 1,100.2 57.4 0.0 -5,827.4

CoCton .30 0.0 269.8 136.0 272.1 136.0 562.0 -752.9 -1,042.7 217.3 81.2 399.5 -388.7 -1,605.7

Uitear .39 0.0 115.9 0.0 136.7 0.0 252.6 -224.6 -340.5 529.8 393.1 -305.2 0.0 -828.4

Total 0.0 1,512.5 136.0 1,487.2 136.0 2,999.8 -3,213.5 -4,726.1 2,925.7 1,574.5 151.7 -388.7 -8,261.5

Ken, aXsize .73 2.9 5.9 0.0 2.7 2.9 8.7 -51.7 -54.7 45.1 42.3 3.6 22.0 -64.5

Coffee .71 0.0 2.0 0.0 0.2 0.0 2.2 -30.4 -32.5 2.8 2.6 27.6 0.0 -15.4

Beef .58 5.5 11.0 5.3 10.6 10.8 21.7 -51.8 -57.3 39.3 33.9 1.6 -51.7 -103.4

Total 8.4 18.9 5.3 13.5 13.7 32.6 133.9 -144.5 87.2 78.8 32.8 -29.7 -183.3-

Paki stanWheat .76 0.0 35.5 0.0 20.8 0.0 56.3 -340.6 -376.1 400.3 379.5 -59.7 0.0 -469.7

Cotton .62 30.4 46.1 9.2 18.5 39.6 64.7 -181.0 -196.7 82.5 73.2 58.7 -209.2 -340.6

Rice .68 0.0 46.0 0.0 40.2 0.0 86.2 -592.8 -638.8 518.3 478.0 74.4 0.0 -539.2

Sugarcane .62 286.5 573.0 0.0 286.5 286.5 859.5 -3,119.7 -3,406.3 2,833.2 2,546.7 -0.1 -1,508.0 -4,524.0

Total 316.9 700.6 9.2 366.0 326.1 1,066.7 -4,234.1 -4,617.9 3,834.3 3,477.4 73.3 _1,717.2 -5,873.5

PortricalWR.eat .70 2.3 4.7 0.0 3.6 2.3 8.4 -36.1 -38.5 52.1 48.4 -18.3 -15.9 -56.4

Olive Oil .65 0.0 47.1 0.0 23.3 0.0 70.4 -265.6 -312.8 262.3 239.0 3.2 0.0 -402.9

Maize .99 0.0 0.0 0.0 0.0 0.0 0.0 -0.2 -0.2 0.2 0.2 0.0 -0.1 -0.2

Seef 1.45 2.4 4.9 3.2 6.5 5.6 11.5 46.0 43.5 -67.1 -70.4 15.4 25.6 51.2

Rice .58 0.0 2.8 3.7 7.4 3.7 10.2 -23.9 -26.8 27.1 23.4 -6.8 -17.5 -48.6

Total 4.7 59.5 6.9 40.8 11.6 100.5 -279.8 -334.8 274.6 240.6 -6.5 -7.9 -456.9

ThailandSugarcane . 1.15 15.8 31.6 15.2 30.4 31.0 62.1 719.4 703.6 -723.1 -738.3 -27.4 414.1 / 828.2

Rubber .68 0.0 10.9 0.2 0.4 0.2 11.3 -70.2 -81.1 2.5 2.3 67.5 -1.3/ -70.8

Tatal 15.8 42.5 15.4 30.8 31.2 73.4 649.2 622.5 -720.6 -736.0 40.1 412.8 757.4

Yui- sla viaWheat .64 34.5 69.0 0.0 36.9 34.5 106.0 -406.7 -441.2 397.9 360.9 -35.7 -191.9 -588.9

Maize .73 31.5 63.1 0.0 30.9 31.5 94.1 549.2 -580.8 507.7 476.7 9.9 -234.0 -697:5

Beef 1.13 1.5 3.0 1.3 2.6 2.8 5.7 79.3 77.7 -71.4 -72.7 -10.7 44.0 88.1

Pork .86 2.t 5.7. 2.9 5.9 5.8 11.7 -110.7 -113.6 107.5 104.5 -2.5 -83.7 -167.5

Total 70.3 140.8 4.2 76.3 74.6 217.5 -987.3 -1,057.9 941.7 869.4 -39.0 -415.6 -1,365,8

1/ %PC* - NPC times standard converwion factor.

2/ The "lou' and "high" refer to the low and the high elasticity assumptions and do not necessarily correspond to the lows or highs of the respective monetary effects.

-40-

84. The analytical framework used to evaluate the foodgrain policies isthe same partial equilibrium model described in para. 75 above. In the caseof food policies, however, it is not satisfactory to evaluate costs andbenefits through real increases in aggregate consumption and production and/orthrough consumers' and producers' surplus estimates. While the governmentobjective may be broad and include the need to assure a minimum subsistencelevel to the poor, avoid excessive wage increases in the urban areas, etc., asuitable index of cost effectiveness may be provided by a nutrition basedconcept of food needs. This is defined as the total amount of foodgrains thatwould have to be distributed through the domestic market and rationing systemsin order to guarantee that the population below the FAO-WHO calorie requirementbe adequately nourished.

85. Based on the above definition, Tables 9 and 10 show the real con-sumption effects of the lower average price faced by consumers as a conse-quence of government policies. 1/ As Table 11 shows, government interventionnever closes more than 56% of the potential nutrition gap. In both Sri Lankaand Pakistan the commitment to external foodgrain availability appears rela-tively constant due undoubtedly to the policy of the distribution of substan-tial quantities at low ration prices. Bangladesh appears to be increasing itscommitment as it adjusts to its independent state. In India, the commitment,although substantial, seems designed more to respond to crisis situations thaneveryday needs; while in Indonesia, extended foodgrain distribution seems tooccur only during periods of high world prices and may even be partiallyfinanced by reduced distribution in periods of low world prices.

86. The above definition of "real" effects of foodgrain policies permitsmeasuring costs effectiveness of government intervention in terms of allev-iation of malnutrition. Table 11 shows how the costs of the distributionpolicies are borne partly by producers (through the loss in producers' surplusdue to procurement prices lower than world prices) and partly by the governmentand reports measures of costs effectiveness as the ratio between the increasein consumption of the malnourished (i.e., the people consuming below theFAO-WHO requirement) and total cost (col. 5) or between the same increase andfiscal cost (col. 6).

87. The results reported show that with the exception of Sri Lanka,where the beneficiary of the rationing policy is the population at largerather than any specific target group, the loss of producers' surplus per unitof additional consumption of the malnourished is equal to or greater than thecorresponding fiscal cost. In all cases the sheer magnitude of the two lossesimply that the food procurement and distribution policies are costly andinefficient both because of (i) high production losses and (ii) because ofleakages of benefits from the low food prices to unintended beneficiaries.

88. To sum up: the partial equilibrium analysis of market interventionpolicies in agriculture has been illustrated by the use of the consumers' -producers' surplus framework for several developing countries. For the case

1/ For details on these computations, see: P. L. Sandizzo and J. Graves"The Alleviation of Malnutrition: Impact and Cost Effectiveness ofOfficial Programs", World Bank (mimeo), Working Paper No. 19 of AGREPDivision, Nov. 1978.

-41-

Table 9:

REAL CONSUMPTION EFFECTS OF FOODGRAIN CONSUMPTION POLICIES

Ratio of Average Per Capita Increase % of Potential

Retail Price to in Consumption Nutrition Gap

Country/Year World Price (kg/person) Closed LL.

Sri Lanka: 1974 .23 12.02 59%

1975 .29 9.85 50%

1976 .31 9.48 51%

India: 1974 .60 26.08 51%

1975 .74 13.96 33%

1976 .77 12.89 39%

Pakistan: 1974/75 .31 12.60 37%

1975/76 .43 8.80 28%

1976/77 .54 6.38 22%

Bangladesh: 1974/75 .78 8.01 16%

1975/76 .45 12.05 27%

1976/77 .39 13.94 30%

Indonesia: 1974/75 .49 24.36 43%

1975/76 .91 2.66 7%

1976/77 1.10 -0.55 -2%

/1 The potential nutrition gap is defined as the difference between the food-

grain distribution required to meet market demand and then provide the

additional foodgrain necessary to eliminate malnutrition completely and

the level of domestic consumption which would prevail if domestic consumption

were exclusively a function of world prices.

Source: Pasquale L. Scandizzo and Judith Graves "The Alleviation of Malnutrition:

Impact and Cost Effectiveness of Official Programs", World Bank (mimeo),

Working Paper No. 19 of AGREP division, Nov. 1978.

Table 10: GOVERNMENT POLICY IMPACT ON FOODGAP, MAINOURISHED POPULATION, DEGREE OF MALNOURISh1fENT AND TOTAL FOOD CONSUiPTION

Degree of TotalFoodg_ap_/ M4l_nourt_5h_ed Popul_a_t_i_on_ Malnouri shment Con______ t_i_e ____

With Without 2/ With Without With Without With WithioutPouiatLion Intervenifon Intervention Change Im ' InM.nf Inron 7k lntiv n -CtV Intervention InoErvention InterventioP mt n n_Sa

('000) ------------- ('000 tons)-------- -------------------- 0)--- - - - - --------- ------------- ('r)------------- ----------------- i'000 tons)--------

Sri Lan'ea1974 13392 70 171 - 101 - 59 49 70 - 21 2.3 5.6 3123 2

9°2 I1r

1.75 13603 77 154 - 77 - SO 51 68 - 17 2.5 4.9 3166 3032 1341976 13819 71 144 - 73 - 51 49 66 - 17 2.2 4.5 3227 3096 131

-i7-' 958i056 9421 19341 - 9920 - 51 40 58 - 18 7.3 15.0 135775 120488 152871975 580707 9551 14158 -4607 - 33 41 52 - 11 7.3 10.8 133514 125',08 ei .0!'1976 592521 7304 11905 - 4601 - 39 35 46 - 11 5.6 8,9 140401 132976 7

? 8a4 s t an

19'C 75 67211 1399 2220 - 821 - 37 89 94 - 5 8.8 14.0 1"573 13726 R.-1975 76 69229 1535 2127 - 592 - 28 90 93 - 3 9.4 13.0 14909 14300 eoQ:9-A 77 71306 1531 1966 - 435 - 22 89 92 - 3 9.1 11.7 15414 14959 455

is,zl-jeshIlS 7.9 76202 2611 3100 - 489 - 16 79 84 - 5 14.8 17.6 15365 14757 610:97.576 79603 2095 2872 - 777 - 27 71 81 - 10 11.5 15.8 16590 15643 94-1976177 80399 2068 2948 - 880 - 30 70 81 - 11 11.1 15.8 1706S 1S°4' 1121

iS74 75 129080 3022 5344 - 2322 - 43 57 72 - 15 10.6 18.7 27720 2A5-6 31': 1*975 7S 132113 3418 3676 - 258 - 7 59 61 - 2 11.7 12.6 27947 27595 3;2 219767 135191 2897 2846 + 51 + 2 54 54 0 9.7 9.5 29509 29584 75

I Defined as food neaded to assure adequate intake of calories by all income groups less actual amount of food consumed.a Change in the foodgap as % of foodgap without Government intervention. This show how much of the foodgap has been reduced (increased) due to Government policy.?rc;ortion of the population with daily intake of calories below the minimum daily calorie requirement.' ith''ut intervention less with intervention.P Per capita shortfall from the minimum daily calorie requirement expressed as % of the calorie requirement.

§ Total food consumption in cereal equivalents.

Source: See Table 10.

Tabl- 11: COST EFFECTIVENESS OF FOOD PROCUREMENT AND DISTRIBUTION POLICIES

Cost of Alleviation of EfficiencyRatio of mal-nutrition Ratio ofAverage in US$1//kg Total Cost of Fiscal CostProducer Loss in Loss of Alleviation of Efficiency (with loss

Price to Government Producer Malnutrition to in Producer Surplus

Country/Year World Price Revenue2/ Surplus3/ Total World Price Set at Zero)

Sri Lanka 1974 .55 1.39 .61 2.00 4.55 4.201975 .66 1.57 .23 1.80 5.63 5.59

1976 .75 1.18 .15 1.33 5.54 6.00

India 1974 .59 .06 .42 .48 1.23 1.10

1975 .69 .08 .43 .51 1.82 1.27

1976 .70 .03 .43 .46 2.09 1.18

Pakistan 1974/75 .55 .36 .32 .68 2.83 2.541975/76 .60 .42 .38 .80 3.48 3.701976/77 .76 .52 .29 .81 4.26 3.74

Bangladesh 1974/75 1.30 .55 -.24 .31 .94 .97

1975/76 .67 .29 .27 .56 2.24 1.92

1976/77 .56 .16 .26 .42 1.83 1.87

Indonesia 1974/75 .49 .07 .33 .40 .98 .88

1975/76 .91 .08 .34 .42 1.45 1.411976/77 1.10 .16 1.78 1.94 7.46 9.65

1/ Shadow exchange rates have been used.2/ Equals the difference between expenditure on a retail basis for officially procured foodgrain and the

resultant receipts divided by additional consumption of mal-nourished (concessionary imports being costed

at world prices).3/ Equals the surplus foregone by producers due to average producer prices held below world prices divided

by additional consumption of mal-nourished.

Source: See Table 10.

-44-

where nutritional targets may be of concern, an extension of the same frameworkto evaluate cost effectiveness of government policies has also been shown fora smaller sample of countries. In all cases presented intervention policiestend to put a heavy burden on the economy of the country and a heavier anddirect one on its agricultural producers.

-45-

VI. EVALUATION OF METHODOLOGIES

89. From a general point of view, it can be shown (see Appendix I) that

DRCs, NPCs and EPCs are all in their different ways measures of product-specific

accounting exchange rates. Specifically, the DRC for the ith product equals the

marginal rate of transformation between domestic and foreign currency in the

production of the i-th good (Appendix I, p. 62); the NPC equals the marginal

rate of substitution between domestic and foreign currency when a foreign

exchange increase is turned into extra consumption of the same good (Appendix I,

p. 61); and the EPC is an appropriate weighted average of DRCs and NPCs (Appendix I.

p. 64).

90. As Bruno has shown 1/, the DRC is a special way of looking at the

NEB coefficient. Using the NEB, one gets an absolute value figure--positive

or negative. One can also pick up one primary factor, say capital, estimate

an internal rate of return to that factor and then compare it to the accounting

price of capital. What is done in the DRC is to "pick up the input and output

of foreign exchange (in the case of an export or an import substitute) and

compare a corresponding ratio of domestic resource costs (domestic, non-traded,

primary inputs) per unit of net foreign exchange output with the accounting

exchange rate." 2/ (Square brackets added.)

91. Data problems make it difficult to estimate accurately DRCs and EPCs

and they can vary substantially and quickly with changes in world priees, but

NPCs, EPCs and DRCs can be used quite confidently for a qualitative assessment

of comparative advantage and the distortionary effects of trade policies. But

it takes time to estimate a set of EPCs and DRCs for all the more important

agricultural products. Fortunately, on the basis of the six country case

studies, it would seem that the EPCs do not differ significantly from the NPCs

for agricultural products. This is basically because purchased inputs are a

relatively small proportion of the value of outputs of most agricultural cropsc6mrpVed to many manufactured products. For similar_reasons_ESCs do not

differ significantly from EPCs. Although producer input subsidies are often

large in terms of budgetary resources, relative to the value of output and the

taxes on output they are small. It follows that, unless purchased inputs are

a high proportion of output, which is unlikely in most developing countries,

calculating NPCs for the more important agricultural products is probably a

sufficiently accurate measure of the degree of distortion and disincentive--at

least for an initial or quick review of any situation--caused by a country's

trade policies.

92. Although it is partly a question of taste and familiarity, we prefer

the NEB measure for the following three reasons: first, one does not have to

worry about which element goes on the top and which on the bottom of a co-

1/ "Domestic Resource Costs and Effective Protection: Clarification and

Synthesis," Michael Bruno, Journal of Political Economy, #80 (1972).

2/ Ibid., p. 20.

-46-

efficient as in the DRC one; 1/ second, all values are expressed in the sameunit of account or numeraire; and third, this is the formulation, albeitwithout the collapsing of the time profile, which the World Bank has adopted inrespect to its project planning work. Nevertheless, we recognize that othersmay prefer the DRC to the NEB, and, provided the estimation of the nationalparameters and project-specific opportunity costs are the same, the twomeasures will give the same answers. Whichever measure is used, estimating aset of national parameters--accounting prices and national conversion factors--would make the task of estimating a set of NEBs or a set of DRCs from timeto time, much easier. Thus, the requirements for sound policy analysis andgood project cost benefit analysis coincide.

93. Calculation of PSEs and CSEs not only supplements the informationobtained from calculating NPCs and NEBs, but provides some very broad inform-ation about the income distribution effects of market interventions, i.e.,between producer and consumer, and, if calculated also for a range of manu-factured products, would provide an indirect comparison of the net effects asbetween agriculture and manufacturing industry. The PSEs and CSEs would haveto be broken down by income groups to obtain a fuller picture of the incomedistribution effects. To do this accurately would be difficult and involvesample surveys being undertaken. Almost certainly such surveys would confirmthe impressions one gets from examining the structure of the agricultural andindustrial sectors and odd pieces of information--namely, that the explicitand implicit taxes on agriculture bear most heavily on those least able tobear them, that producer subsidies tend to go to the richer farmers ratherthan the poorer farmers and consumer subsidies often go to people who do notneed them. In short, that the net effect of tax-subsidies is typicallyregressive.

94. The use of PSEs and CSEs is a simplified, special case of the wider,more detailed, partial equilibrium model for estimating the allocation in-efficiencies and welfare transfers as between consumers, producers and govern-ments, outlined in Section VI. The results are an interesting application ofapplied welfare economics. However, the data requirements for applying themodel to any real world situation are both larger and subject to a higherdegree of error (particularly the demand and supply elasticities) than usingthe simpler Joslin Methodology. If time, resources and data permit, the morecomprehensive methodology should be used, but if time, resources, and data area serious constraint, the Joslin Methodology gives an insight into roughorders of magnitude of the allocative inefficiencies and welfare transfersinvolved in any system of agricultural administered prices, taxes andsubsidies.

1/ E.g., capital has to be divided between traded and non-traded inputswith corresponding allocations to the numerator and denominator of theDRC equation. See Bruno, op.cit., p. 27, and Corden, W.M., "The Structureof Tariff Systems and the Effective Protection Rate", J.P.E. #74, June1966, p. 232.

-47-

95. To conclude, it has been shown that the summary measures, partic--ularly the NPCs and the NEBs/DRCs, are useful in illustrating the generaleconomic efficiency effects of agricultural administered prices, taxes andsubsidies. To a lesser extent, the PSEs and the CSEs, particularly if supple-mented with other distributional data, illustrate some income distributioneffects of the same prices, taxes and subsidies.

-48-

VII. GUIDELINES FOR ANALYZING PRICE INTERVENTION EFFECTS

96. An analysis of a large sample of World Bank Country Economic Reportsand Agricultural Sector Reports revealed that the analyses of the effects ofadministered agricultural prices, taxes and subsidies was unsatisfactory inmany respects. 1/ The practical question is whether country economic work andagricultural sector work can be improved significantly without a deteriorationin other aspects of this kind of work which might be equally or more important,since a trade-off is clearly involved.

97. It is largely a question of judgment whether there would be a netbenefit from substituting some analysis of the effects of administered agri-cultural prices, taxes and subsidies on economic efficiency and income distri-ution in place of other analysis and other country and agricultural sectorwork. We believe rather strongly that the substitution or application ofadditional resources to such analysis on a systematic basis would reap handsomedividends, from the point of view of LDCs and aid agencies. We believe thisfor the following reasons:

(a) The case studies reveal that the distortions are substantial inmany LDCs: the domestic terms of trade have been turned heavilyagainst agriculture vis-a-vis the rest of the economy; croppingpatterns have been distorted against what long-run comparativecost advantage would indicate is the optimum pattern; incomedistribution has been made worse; and policies are internallyinconsistent and not achieving what it was believed they woulddo. Thus, growth rates in agriculture and hence the economy asa whole are lower than they could be, and the poor bear a dis-proportionately heavy burden of the consequences; and the balanceof payments has been made worse.

(b) Nowhere are the above consequences more important than for foodproduction. Self-sufficiency in food in a world of growing foodshortages is a political not an economic criterion. Neverthelesscomparative cost advantages are such in the major food deficitcountries, that substantial increases in the output of subsistenceand marketable food crops could be effected if farmers were givenadequate incentives in terms of removing many of the existing price/tax/subsidy distortions, particularly if, at the same time, inputdelivery systems were improved and made available to the poorerfarmers.

(c) Some of the more important constraints on effective projectformulation and implementation which one comes across so often haveto do with inappropriate agricultural pricing, taxation and subsidypolicies. It is generally unrealistic to expect that leverage toremove the constraints can be applied with success through indi-vidual projects--a sector-wide or, indeed, economy-wide dialogueneeds to be entered into in order to effect an improvement in

1/ See Brown and Donaldson, op. cit.

-49-

policies. It is difficult to enter into such dialogues without astudy of the policy effects and this may be possible only whenadditional information has been collected. Thus, an analysis ofpricing effects on a sectoral/national basis is important both toimprove economic performance generally and to remove constraintsin respect to designing and implementing sound projects. In turn,analysis of pricing effects is made simpler if a set of nationalparameters (accounting prices and conversion factors) is availablefor each country.

98. Preferably the analysis and the back-up national parameters shouldbe estimated by each country and reviewed periodically, but whether this isdone by LDCs themselves (by the national planning agencies and/or the min-istries of agriculture) or by or with the assistance of external aid agencies,the trade-off problem can to a large extent be avoided by employing consultantsto do the initial work and to train nationals to undertake periodic reviewsand revisions.

99. Whoever does the work, how does one set about doing it? Are there

any short-cuts that can be employed? It is suggested that a three-stageexercise be undertaken. The first stage would consist of fact-finding anddescription, including such things as ascertaining and describing all the

administered agricultural prices, direct and indirect taxes and subsidies andphysical controls on agriculture there are; undertaking a literature review;summarising any previous studies; and obtaining realistic cost of productionprofiles of the main agricultural products.

100. The second stage would consist of estimating the key parameters such

as a set of accounting prices and conversion factors, elasticities and crosselasticities of demand and supply for the main agricultural products, orreviewing existing sets of key parameters and doing any necessary updating.This is probably the most time-consuming stage in most LDCs and the one whichcould well be carried out by consultants. For example, a foreign consultant,working with a local civil servant in the Economic Planning Unit, PrimeMinister's Department, estimated a set of national parameters for projectappraisal in Malaysia at a cost of under US$20,000 (1977).

101. The third and final stage would consist of the calculation of thesummary measures of distortion and disincentives, analysis and formulation of

conclusions and recommendations. In respect of the summary measures, NPCs can

be easily calculated for almost all countries 1/ because, even if border anddomestic prices cannot be compared directly, the NPC can be calculated byusing the foreign trade taxes and any discriminating indirect sales taxesthere may be and applying them either to the border price or to the domesticprice. If cost profiles for the major agricultural products are available for

a reasonable range of technologies--irrigated and rainfed, estate, settlementand smallholders, etc.--then the national parameters can be used to estimate

1/ The retention coefficients mentioned in the Prologue are similar to NPCsexcept they were estimated at the farmgate, and the farmgate financialand border prices were estimated ones taken from World Bank StaffAppraisal Reports.

-50-

NEBs. Alternatively, one can look at any recent appraisal reports (includingWorld Bank Staff Appraisal Reports) for the commodities in question and reworkthe figures, applying the national parameters to come up with net presentvalues at market and at accounting prices.

102. In most developing countries it should be possible to come up withsome rough and ready calculations of PSEs and CSEs, which will give additionalinformation of the burden of taxation on farmers and the extent to which theyare directly subsidising consumers. Subsidies for agricultural productsshould also be expressed as a percentage of total government revenue and ofthe national income/expenditure.

103. None of the summary measures described above throw any detailedlight on the income distribution effects of administered taxes and subsidies,and it may be desirable to add a further stage and try and trace through theincome distribution efforts with further studies, including sample surveys.

104. Once all these measures are calculated, the next step is toutilize them in policy analysis, planning and operational work. If NEB'sor DRC's are available, they should be used first to judge the optimalityof the present allocation of resources. Large NEB's (or DRC's much in excessof 1) would indicate sectors where the best project opportunities may ariseand where the Government and the Bank should concentrate resources and projectidentification activities. At the same time, however, large NEB's mayindicate the existence of extraneous factors hampering the expansion of thesector and the full exploitation of its comparative advantage. Majorhampering factors are likely to be, as shown by the studies, all the variouscombinations of market interventions reflected in the NPCs, EPCs, PSEs and CSEs.

105. A second step to use these measures operationally, therefore,should involve a comparison of the equivalent taxes and subsidies with theconclusions on comparative advantage reached after the computation of theNEB's. The basic question to be asked at this stage would be: is thepresent system of interventions an obstacle to the expansion of the sectorswhere comparative advantage lies? Clearly, implicit or explicit taxeslevied on producers or on consumers are a disincentive to production andmay entail a direct cost to development because they may discourage theexpansion of promising sectors (or they may encourage the expansion of un-profitable ones).l/

1/ Note, however, that in practice it is often necessary to analyze theoptimality of taxes or subsidies on specific commodities, without regardto the comparative advantage question. Such an analysis can be convenientlycarried out in terms of the producer/consumer surplus technique describedin Section V.

-51-

106. Whether an elimination or a reduction of these interventions shouldbe recommended, however, depends on a third step, which will consist of theevaluation of the effects of the measures on several policy goals. Fourimportant sets of these effects have been illustrated in Section V of thispaper. They are consumers and producers surpluses, foreign exchange, govern-ment revenues and food consumption of the poor. Two additional effects thatit may be important to evaluate are the employment implications of the marketinterventions and the impact on selected population groups or regions. Thequantification of these effects will provide the main information for rec-ommending a new level of taxes or subsidies, the elimination of the previousones or alternative sets of policies.

107. Rather than considering a sweeping elimination of all marketinterventions, it may be often useful to evaluate a partial elimination ofthe existing taxes and subsidies. Given the surplus measures and the addi-tional effects, one can compute the benefits and the costs of decreasingsay, a production tax from 10 percent to five percent. The benefits wouldbe the avoidance of producers' surplus losses, while the costs may be afall in government revenue and an increase in malnutrition. If one canattach a value to these two effects, not only any given policy can be evaluated,but also an optimum change can be determined.

108. While it is advisable to avoid basing one's recommendations onoverly inflexible general principles, we would like to mention in closingthree general propositions put forward by economic theory that can also beuseful to guide policy analysis and/or policy prescriptions in this area.l/First, the system of accounting prices that we have described in the pastpages is, as a rule, associated with a social optimum. As a consequence,the prima facie case against all taxes and subsidies is that they are inprinciple undesirable. Second, this conclusion holds both in a first best

situation where everything is optimal as well as in a situation where sometaxes are exogenously fixed and cannot be changed. Third, while it may bejustified in some cases to raise taxes or pay subsidies to the privatesector to modify its behavior, this should be done without affecting publicsector efficiency.

109. These propositions are intuitively appealing and have the gift

of simplicity in that they simply recommend sticking to the basic systemof accounting prices as the optimal set of mharket signals to enforce. Whilea more deta.led consideration of the extent , the benefit and the costs of

market intervention is generally in order, this hard and fast rule is aconvenient point of departure for any pricing analysis. Once the analysisis performed it may also provide a long term goal for the pricing policiesof the country.

1/ The basic reference for these propositions is: Diamond, P.A. andJ.A. Mirrlees, "Optimal Taxation and Public Production", AmericanEconomic Review, March and June 1971.

-52-.

Appendix I

A MATHEMATIC MODEL OF ACCOUNTING PRICES

1. The General Pricing Problem

1. Pricing guidelines for the evaluation of market intervention

policies and, as a special case, for project evaluation, can be generated

through a mathematical model. Here we consider a simple static version of

such a model: additional complications such as many time periods, extern-

alities, etc. could easily be introduced and would result only in compli-

cating the algebra, not in modifying the substance of the results.

2. Consider a country, whose system of national preferences can be

described by a function of the quantities consumed of each good:

U = U (cl, c2 ... c '~'I+l ... cn) (1)

where U is an utility indicator for the country, cl, ... cn are quantities

consumed of traded commodities, and cn ... cn are consumptions of non-

traded commodities. Each good (traded and non-traded) can be produced

through a technology described by a neoclassical production function:

Xi= fi (yli)(i ........ y I) YMi) +1 *- m).......... YM (2)

i=1 ... n

where xi is the quantity domestically produced of good i, y1 , y are

quantities employed of imported factors, and Ym +1 X Ym are

quantities employed of domestic (non-traded) resources subject to the

constraint:

ziYk = Yk !!-Yk k = Ek+l1, ...+ m (3)

where Yk, k = ml + 1, .. , m represent domestic availability of the

k-th untraded factor.

-53-

3. Consumption and production levels of traded commodities and other

foreign exchange earnings (Y*) are linked together by the balance of

payment (B.O.P.) constraint:

ni mE P* (c - x,) + E w* y - y* = O (4)

i=l i' j=l ]

where Pi is international price of good i (assumed as given in a "smallI

country" framework) , and wJ is the international price of imported factor j.J

4. The problem of finding an optimal set of product and factor prices

can now be directly cast in terms of the utility maximization problem in

.(1) subject to (i) the BOP constraint in (4), (ii) the resource avail-

ability constraint in (3) and, (iii) a market clearing constraint for

untraded goods. Formally:

Max. U=U(cl , ... cn)

nl mlSubject to: E Pi (Cc - x)+ y Y* = i

= i i i =1JJ

n (il 1 (5E k ) Y 0 ,k = ml + 1 , ................m (5)

ce Xe < 0 ,e = n+l, n

e e-1 Xi - fi (y(i).. Ym ~)3 e i = 1, ... n

All quantities non negative.

5. The constrained maximization problem in (5) captures the main features

of the problem of price setting and the conditions for its solution can be

directly interpreted as pricing rules for all endogenous variables:

consumption and production levels, imports and exports, factor employment

etc. Forming the Lagrangean from (5), we can write the equivalent formulation:

-54-

Max L =U ( c,. cn) - i[~p ~-f (y*. Y my.-Y* 1 n) [i= i 1C fiiY m3 j=l wjy

nl m - n(e) (e)(6

i-lk~m+ kk k k en +1 e Ice - fe(Y I.. Ym

6. The Kuhn-Tucker conditions for the maximization of (6) imply -tht

each non-zero consumption and factor employment level respect the following

equalities:

(7)

3L = auc - Tr = O e = nl+ , ................. n (8)

-L = 'p( p* + w* ) _ e = 0, j=1,2 ... nil (9)ay e~~~~~~~~~~~yi e=n1+1 yj

aL nI af. n a~ ~ Wk+ P r= 0, k=mi+l ... m (10)9yk i=l ayk k e =n+l e yk a

7. Equations (7) - (10) can be directly interpreted as pricing rules

necessary to ensure maximization of (6). These pricing rules are expressed,

however, in terms of "utils" or units used in the specification of the

utility functions. In order to translate them in monetary terms note

that p, the Lagrange multiplier of the BOP constraint can be interpreted

as the marginal utility of the foreign exchange in the hands of the

government(Y*). Thus, if we use Y* as a numeraire, expression (7) can be

rewritten as:

au/aci = aU/aei = ay* = P(11

V au/ay* ac .________ ____ 1I

-55-

which states that the marginal rate of substitution between foreign currency

and a traded good equals the utility value of an additional unit of con-

sumption (the "first best" price) and this in turn equals the good's border

price.

8. Expression (8) can be treated in a similar way, dividing both

members by p:

aUIace = =e ;_ e.=nl+l, ... n (12)

3U/Dy* a ce au aY

Because the we 's are the marginal utilities of the market clearing

constraints, they also equal, for all non-zero domestic commodities, the

vector of undistorted market prices (p) expressed in domestic currency,

multiplied by domestic currency (Y) marginal utility. Thus (12) can be

written as

DY; e=n ... n (13)

e

where ca = au/Y ay* (14)

9. Equations (13) and (14) state another well-known rule of project

evaluation: prices of non-traded commodities in terms of the foreign excnange

numeraire equal undistorted domestic prices multiplied by a factor reflecting

the rate of exchange (the marRinal rate of substitution).between foreign and

domestic currency. If no distortions are present this factor is a f

constant depending on the units selected to measure domestic or foreign

currency. Thus, it can be conveniently set to equal the official exchange

rate. If trade or domestic distortions are present, however, the conversion

factor will be a function of these distortions and special assumptions are

needed, as we will see, to compute it.

-56-

10. Performing similar normalizations on equations (9) and (10) yields the

pricing rules for non-zero and fully employed factors. For non-zero imported

factors, expression (9) can be written as follows:

DU/ay ay*nf n afi n afe ae)

u/ay* =y= i=l i a(g) ayj e=n +1 e a(;) yi(i=l ~ J1

j = 1, 2 ... m1

where denotes the amount of factor y. used in the production of good i.

11. This result states that optimum domestic price of input j should equal

a weighted average of its marginal productivity in all alternative uses

and this in turn should equal its import price. The weights should equal

the marginal shares of each Rood in the use of the input. For ex-

pression (10) an analogous result can be written noting that for fully

employed factors in equilibrium Dk = -Y wk where wk is the domestic price

of the k-th factor.

Thus: f ifnu i Yk n e k6

W* = awk =* i Pi a(i aY e+ Pe a() ay aL6)k k i 3~i) 'yk e=n +1 kei=l ~k 1 3k

12. The rules implicitly dictated by the conditions for the maximization

of our simple model are thus clear and bear direct implications for pricing

policies and project evaluation criteria. As general guidelines they

can be stated as follows:

(1) For traded goods, (goods that are actively traded or

could be traded on the international markets) domestic prices should

equal international prices (Equation (11)).

(2) For non-traded goods, domestic prices should equal competitive

market clearing prices (Equation (13)).

-57-

(3) For traded inputs, import prices should equal domestic prices and

these in turn should equal a weighted average of the value of the

marginal productivities of traded and non-traded goods in whose

production the inputs are used. These values should be measured

as in (15) and expressed in the units of common numeraire.

(4) For non-traded inputs domestic prices should equal a weighted average

of marginal productivity values measured as in (16).

13. These rules are simple, but far reaching. With trivial modifications

of the analytical framework they can be extended to the case of intermediate

goods, large countries and intertemporal equilibria.

2. The Second Best Case: The General Conversion Factor and the Shadow Exchange Rate

14. The simple rules specified above can be convenientlv modified to account

for specific government actions, institutional constraints, or other dis-

tortions causing domestic markets to fail to properly allocate resources.

15. In order to analyze this "second best" case, assume that the s-th

consumer solves the problem:

Max Us(c s I c2s ' cms) (17)

Subject to: P C - y < ; s = 1,2 ..... Si=l i is 5

where U is his utility function, pi is the domestic market price of the i-th

good, cis the quantity of the same good consumed. and Ys his income. The

necessary condition to solve (17) is thus:

5 5au, au pa cis Y-! i 0 il= 1,2...n (18)

s= 1,2 ...

-58-

16. Indicating with Cs the consumption vector of the s-th consumer, the

Pareto optimal problem could now be stated as follows:

U = [ U (C)] is maximum if there exists no feasible C1 such that

US (C1) U U (C°) for all s and US (C1) > uS (C°) for at least one s.

17. As Negishi (Metroeconomica, 1960) has shown, solving the Pareto

optimal problem is equivalent to maximizing any convex combination of

consumers' utilities. In the context of model (5) this implies that the

objective function be written as:

SMax U = V Us (CV >0 (19)

s=l C5

18. The Vq can be interpreted as distribution weights. Any system of

weights will lead to a Pareto optimum if problem (19) is solved under the

constraints specified in (5).

19. Assuming equal weights for all consumers (Vs = 1, s = 1, ... S),

we can write from the first order conditions:

DU aU and aU = DU (20)~Cj5 ac a a

s= 1,2 .. Si = 1,2 ... n

Substituting (.20) into (18) we thus obtain:

aU = au p i = 1,2 ... n (21)aci ay i

For a traded good, substituting (21) into (7) yields:

au Pi -au Pi = 0 (22)

or, using the definition given in (14):

= a* P= a P i = 1,2 ... n (23)i y i i

-59-

For a traded good affected by a distortion such as, for example, an import

tariff, we also have by definition that:

P= rpi (1 + ti) (24)

where ti is the tariff rate and r the official exchange rate. Since ax is

a conversion factor common to all prices it will thus be impossible to meet

the pricing rule defined by (7) or (23) for a good directly affected by the

distortion. For unprotected goods, on the other hand, the rule can still

be applied in the sense that domestic prices are required to be proportional

to border prices. The factor of proportionality, however, is a function of

all price distortions and remains to be determined.

20. The interpretation of at in the general case when all sorts of distor-

tions may affect domestic markets is thus important to establish what rule

to follow to determine "second best" prices. Assuming foreign exchange

availability to be determined exogenously, the inverse value a has

a clear interpretation since it represents the increase in expenditure

in domestic currency generated by an exogenous increase in the availability

of foreign currency. Clearly both the increase and the ways by which

additional foreign exchange is generated depend on government policies.

A general definition can be written as follows:

nl aEi

_ aY 1 i-l (25)r Y* a n1 * aEi + n1 w* ay.

E p * -i__1i 3g j=l 3 a

-60-

Awhere r is used to indicate the shadow exchange rate nature of the inverse

conversion factor, Ei = ci - xi and g represents the particular policy

instrument used to increase foreign exchange availability.

21. Several cases may be distinguished for the evaluation of r that

depend on alternative government policies to increase foreign exchange

availability:

(A) Reduction in domestic consumption

(B) Variation of official exchange rate (devaluation)

(C) Variation of a particular price or set of prices

(D) Increase in investment

In all cases the expression for the shadow exchange rate can be formulated

in more familiar notation as follows:

^ X i 1 + t ) + X yi Mi ( 1 + t )ij i ix j i + im) (26)r=

i e x + i i M

where E. = elasticity of export supply with respect to policy instrument

n, = elasticity of import demand with respect to policy instrument

Xi = value of exports of i-th commodity

M. = value of imports of i-th commodity

tix= ad valorem tax on exports

ti = ad valorem tax on importsim

22. The problem with cul.putation of (26), however. is not only that import

and export elasticities corresponding to (A) - (D) are generally unknown,

but also that, depending on a particular government policy, they beg the

-61-

question of which is the best policy or set of foreign exchange procurement

policies. Fortunately, it turns out that in most cases a good approximation

of (25) is provided by

R fiE Xi ( 1 + tix) + E b (+ im)r ~~ i

(27)

i Xi + i Miii i

23. To evaluate the approximation involved in (26), we write the

Taylor expansion of r in (25) around Ei'ni = 1. After simple algebraic

transformations, such an expansion can be written as follows:

A ~r r (X Tx ) + ( M - Tm) (Tx + Tm) + (28)

( X + M)2

where TX = xtj = value of taxes on exports

T = M t = value of taxes on importsm j j jm

24. Expression (27) can be used to argue that r will be a fairly good

Aapproximation of r whenever, (i) elasticities are close

to 1, and/or

(ii) the value of trade is large relative to the taxes levied, and/or

(iii) taxes and subsidies tend to offset each other in aggregate value

(T + T Z. 0).x m

-62-

3. More on Conversion Factors

25. The simple formula obtained in (25) reflects only a limited range

of distortions and is implicitly based on the hypothesis that the country is

a price taker in the international market. Special conversion factors can be

obtained, however, to account for a wider range of distortions and/or

hypothesis.

26. In order to derive these expressions from a general formulation,

consider the case in which we seek to determine the optimum shadow price of

good j at consumption and production level. Maximizing (6) w.r. to pj is then

equivalent to maximize (19) under the equilibrium assumption Vs= 1. Taking

derivatives with respect to pj, yields, for the consumption side:

n Du ac. n dC n Dcefl n--1 1 n * iE 7 2z oU. DP i n pe ap

i=l i j i=l 1 e=n1+j

27. If consumers maximize utility in domestic distorted markets we

must have:DU Du c cac. = Y Pi U Pi (30:

where U is the marginal utility of income in domestic currency and pi is0

the price faced by domestic consumers (demand price). Note now that p is

the marginal utility of foreign exchange Y* at the constrained optimum of

U (c) obtained by shadow pricing the consumption level of the j-th good.

I/UO = DY/3Y* will thus reflect the marginal rate of substitution between

domestic and foreign currency at the same level. Denoting this conditioned

Cvalue of the conversion factor as acv we can thus write, for the small

country case: n1 * c

C Vil 1c nij (31

; =

i ij

-63-

where V4 = P ci c v = PC C and Nic is the cross demand elasticity between1 i i i J

the i-th good and the j-th price. For the Robinson-Metzler case of zero cross

price elasticities a is simply:

i

ac = Pi = r (32r=7 - ~~~~~~~~~~~~~~~~~(32)

Pi i

where r is the official exchange rate. Thus the consumption conversion factor for

the i-th good equals the ratio of the official exchange rate and the Nominal Protec-

tion Coefficient-. In the large country case p* is replaced by marginal revenue,1

that is, in the simple case of (32):

c P~~~1 (33)ai = c (1+ } )(33

Pi

where ni is the foreign excess demand elasticity.

28. For the production side, on the other hand, we obtain:

Dx m 1 y n ax m ay11 P* - E w* E~) 7T e wg )= (34)

i=l i apj k-l k apj e=n1 +l e 3pj g=mi+l j Pi

where iT =U px and w uw.Thus:e O e andwk ok

E V * rj. - I W* E;ax i= ix 1i k=l k kj

J V nX _ We.

e=n1 +l e ej g=mL+l g -g

where V= x x , Vi = P* xi, Vni is the cross price elasticity of supply, .kjee e i i j k

the elasticity between factor demand for the k-th factor and price of the j-th

output and Wk and W indicate respectively foreign and domestic values of inputs.

Expression (35) implies that unlike demand, supply shadow prices and conversion

factors depend on the existence of non-traded goods and/or factors. In the

1/ Computed at the retail level.

-64-

special case of zero cross elasticities, we have:

- m1

x ix i k-l k ki(36i m

2 W eg=m +l g gi

29. The conversion factor for the i-th production price can thus be

interpreted as an inverse marginal cost of foreign exchange (DRC) in the

production of the i-th good: it represents the increase in net foreign

exchange earnings that would follow an increase (or decrease) in supply of

the i-th good or an increase (decrease) in supply of any input used for its

fabrication. In the particular case of input-output technology (zero elasticity

of substitution among factors) the derivative ayk/DPj can be decomposed into

input-output and supply response coefficients:

ayk ayk ax i axi (37

-=i axi ap = -ki 3pi

Substituting into (34) yields, for the zero cross elasticity case:

(V* - E W )r Xx i ~~k-1 __ax = mk=l = C DRC (38

, Wg iig=m1 +l

where DRC indicates the domestic resource coefficient as defined by Bruno

and r is the shadow exchange rate. Thus, under partial equilibrium input-output

technology assumptions, the conversion factor for production equals the inverse

of Bruno's domestic resource cost coefficient. Again, the large country case

can simply be taken into account by using marginal revenues (expenditures)

instead of prices:

-65-

* 1 x E 1 k

x ~ ~~i k=l ki m

gm 1 +l g gi

where n* is the elasticity of foreign (demand) supply for the i-th good andi

e* is the elasticity of foreign supply for the k-th factor.k

30. Consumption and production conversion factors can thus be interpreted

as net foreign exchange values of increases in consumption or production of a

given good or combination of goods. When applied separately, they presuppose

a situation where pricing is applied only to consumption or production either

because the project acts only at either of these levels or because the possibilities

of reinforcing prices other than (distorted) market prices are limited. In

all other cases the appropriate conversion factor is a weighted average of the

two. For the small country general case, in particular, the conversion factorU

is obtained by summing expressions (29) and (34) and solving for -

n V c nm * mlE v n En1 T - k *E

ilic nij ilix ij kl k Ckgi=l i-l k=l (0a = -- ~~~~~~~~~~~~~~~~~~~~(40)

j n n mi-l i nij+ en+l eej g=m+l g gi

c xor ac. = Yc a y 'x (41)

J J J

nIV~flicc

where y =c n n in

i=l i ij e=n+l e ej g=m+l 9 gj

-66-

E - E *E

ial i ij k=l k kjand yx n5

E vc c. + n v r. - W e.i=l 1 ij e=n +l e ej g=m +l g gJ1 1

For the zero cross-price elasticity case, expression (40) becomes:

V* (n+ Tj W * * miEW*i ( + ni) k=l k ki i 1½ k=1 k ki (42

o ic m 4V ri.- W S.V n + V.nT E W s:c 1 =Z g gi Vi i i k=l k ki

where ni = ( - J.cj ). In the absence of domestic distortions and with

input-output technology this formula reduces to:

V _ m Wk

ai k-l EYCi (43m1 EFkk=l Wk

and equals the ratio between the official exchange rate and the effective

protection coefficient.

-67-

Appendix II

PRICE EFFECTS ON AGRICULTURAL SUPPLY

I. Introduction

1. Agricultural supply responses have regained much attention in the

last decade as development economics has focused on problems of dual economies

and food shortages. Furthermore, the question as to the responsibility of

exercising collective bargaining power for agricultural exports was also

raised after the success of OPEC's oil embargo and recent concern of govern-

ment and aid institutions on basic need and income distribution policies have

reinforced the necessity to increase our understanding on agricultural supply

responses. This Appendix presents evidence from two surveys on the subject

and from direct perusal of a number of recent studies, with the objective to

summarize the empirical information available in a way that may facilitate

its use in policy analysis.

2. In developing countries on-farm consumption is often a large share

of farmers' production. This portion can be subtracted from production to

measure supply response of marketed surplus. Alternatively, total acreage

(output) response can be defined to include on-farm consumption. In both cases,

it is important to consider the evaluation of demand responses in determining

the net supply responses. Own consumption response may be analyzed with a

different model framework than the one used for household consumption data.

One main difference is due to the fact that the farm income level is positively

related to price increase and, as a result, on-farm consumption responses to

crop price increases can also be positive. This positive demand response can

partly offset the production response and lower the net responsiveness of

marketed supply to price changes. Therefore, the distinction between production

responses and marketed surplus responses should be kept in mind when comparing

effects of various price intervention policies.

-68-

3. The majority of the supply response studies concentrate on measure-

ment of acreage response. Since production response can always be decomposed

into an acreage and yield response, the acreage response is a good measure of

total production response only if yield response is small. The latter condition

is subject to empirical testing. When substitution between land and other

inputs such as labor, fertilizer, irrigation and new seeds is possible to any

significant extent, yield responses may be considerable. On the other hand,

because actual production levels reflect the influence of factors such as weather,

plant diseases and infrastructure, measuring supply response with production

levels is not without flaws.

4. A further issue of importance in evaluating supply responses is the

relationship between output price and the profitability of crop production.

Clearly, relative rather than absolute prices are likely to affect profitability.

In principle there are three different ways to define the numeraire to compute

relative farm prices: price of other crops, prices of inputs, or prices of

goods farmers may consume. From a more general point of view, it can be even

agreed that the relative profitability of a particular production activity can

be compared not only with the alternative farm production activities, but also

with other nonagricultural activities--such as selling out the land for building

factories and migration to urban areas to be a wage earner. Because degrees of

freedom and data problems generally restrict the choice of the numeraire and/or

the number of decision variables to be included in the response equation, the

choices usuallv made are not necessarily satisfactory.

5. The supply response models used in the literature surveyed are typically

based on a reduced form composed of only observable variables:

-69-

Tn Tk

y = 0+ y -,+ E k bikx'kt-i, t a1 k=l i=l (1)

where yt is the supply of a given product at time t, and Xkt is the realized

value of an explanatory variable at time t (generally represented by the price

or the unit revenue of the crop y or of an altervative crop). The number of

explanatory variables is n, while the number of lags for each k variable is Tk.

6. Two basic formal structures can be used to give theoretical justifica-

1/tion to the reduced form in (1): Nerlove adaptive expectation model-= and

2/Koyck and Cagan distributed lag models.- The Nerlovian formulation rests on the

twofold assumption of: (i) a progressive and constant adjustment of short run

supply to its long run value and, (ii) a recursive formation of price expectations.

In algebraic terms these two assumptions can be written as follows:

Yt Ytl1 +YYtYt-l) ; y< 1 (2)

* * *t t- ( t-1 t-1 ) ; 0 < < 1 (3)

where the asterisks indicate the expected or long term value of the variable;

Yt denotes supply and Pt price.

7. Equations (2) and (3) can be used independently or jointly to provide

a theoretical justification to equation (1). Assuming expected or long run

supply to be regulated by an equation of the type:

* *

Yt = a + cPt (4)

we obtain, after substituting the values of yt and Pt given by (2) and (3):

Yt = a + (1 - Y) yt 1 + cy (1 _ i1 Pt i (5)

1/ Nerlove (1956, 1958).2/ Koyck (1954) and Griliches (1967).

-70-

Multiplying the expression for Yt 1 according to (5) by 1 - 6 and subtracting

from (5) yields the reduced form:

Yt =a + - (1- ) (l - Y) yt-2 + cyS Pt_1 (6)

where a. = a(l - A)

8. Koyck and Cagan's models are based on the equation:n X

yt a + i E ik xkt-i~~ k=l i=l

This equation can be related to (1) by assuming a particular polynomial

structure for the weights eik' The simplest formulation due to Koyck is based

on a geometrically declining form:

eik = 8k S(l - 6) i = 1, 2, . . . (8)

Substituting (8) into (7) yields after multiplying the corresponding expression

for yt-1 by (1 - a) and subtracting from t

n=a+ (1 -aZ 8 kXatl (9)

Yt a0 ( )t-1 - gk 'kt-1 9k=l

where a = a 8o o

9. As an alternative to (7), one can postulate a finite distributed

lag for expected price:

* T

Pt-= l p (10)i=l t-

and use it in conjunction with equations (2) and (4) to yield.

TYt =a + (1 -y) yt + yc E a t (11)

t-1 i=l

ln. Equations (6), (9) and (11) are the most common models for agricultural

supply response estimates. In all these models direct estimates of the price

coefficients yield estimates of "short run" parameters of supply response (in

the case of log-log functions, "short run" elasticities). "Long run" parameters

-71-

are obtained by dividing the short run parameters by the partial adjustment

coefficient or the product of it and price expectation coefficient. As it is

clear from the different formulations, the concept of long run varies with the

three models: it assumes completion of an adjustment process concerning only

planned supply in model (11), only adaptive price expectations in model (9)

and both supply and prices in model (6).

Tablel I 1null! p26ATC ITY (IF SuPP-IlY OF OAC iCn.ItcAi. P'IMiibTil lb LA.11 Appendix Ii

For- Price P.iottciniiaAdnint C-'f.Pr-l-, ~~Crc1nin Racle- Period S-evy Sitori-rnm I.9PnAe ri,e 2~ C.t-:,Ie al el.:.i& o 1' "ioi

- !r~~~~~~~~~eiee ~~~~~1950-74 .21 to .35 .42 to .70 Ag;grega.te oipot. Rena (1976) Ic S(e- M77

Poodgealna Icl. 1 dIstrIcts L. Punjab 1951-64 -.06 to .42 Agr-gotm colon lden of 11 mjor crops. -Iari (1970) In SaU;na

Rice Thailand Hiala 1951-65 *390 .21 1.27 .4.0 4 -1eitaim oi-er-im a-drl ore tried, (eIc.l. l9e$l' InA Cii,hot only t.O oro d -te?PdV e.j

RIce Talxa lll macInn 1937-63 .1700 .19 to .63... 1.0 in .25 t. 9 ihoneiand5 qala.Ohct 1c! A6Cdhr

no,ti Adyci-e tcEaroI-cn and PartialAdjosterot hyporhasas ar iop-sod.

pi.. India hlIa., (12 1953-63 .01 -o .22 bea-. ine in A 6 C 0l1 C

ndi.tielts)

Rice ni P.nn(ab- 1960-09 .19 to .24 .64 to .69 .26 i.. 5 PrdncioIs ma.ho oIente,d. A A C (19765ll-en-I.n model. OLd estimation

flint InI ltioa 1938-57 .22 ti-r-oa dynaoic mdel. PXAER (1962) lad A IC (1)76;

floe PakIstan NatIo 1949-68 .12 .17 .71 ran-i-c in A ii C hOTel

Rica Pahl.... xialo 1959 Tea 1.60 - Output elasticity of marked so-plne. tho 6 Ch,dIuep (19621 InSob,saepl.ia ro annoY accdle. Obha..

Rice. Bnoladesh Ration 1949-Rn .13 .09 C-nuaie. InA L C 0is7,1

51cm Pecgledeeh Ra.ttan 1950-66 .03 5.20 oet- o-aaei

A I Cii.

Rice Philippin.. Rciatl 1910-41 (AL) .10 Io .260 The booc model Ia to enpInia Aerenom (Segaha. (1966)

Rica Phillippia. Nation 1910-41 (Ii) .15n -prir n c c"0lc

ine philitppaae Ration 1910-41 (A) .00 to .000 .02 to 1.56 aad tcclaeegl-t -arlohlr (tInerrad or a yield roe. .ii plopi

RIco PhliIppine. Norton 1910-41 (0) .04 1.16 ho-me aedJ dit-bte lgcee ntried. Thlee ore 1clecenriC.:

Ric. Philippi-m Ration 194a-64 (A) RaSat1ea (A) Incl.de eo rerne od all alt-ranie-crap pr Ire. (R) fice lod o prIer en! i..

of -c prica to die etba- najoroteeta- ti-e pcico. acd (C) the clot. raia of

cnprica toancniaeaiIn indea of all

Rine 'Philippi.o. RatIon 1972-74 Ya. .4 to .7 .7 to 1.0 P.o aoplo a dec6to no estiota 1.th Tegoera (19;)) icn.lo

Rico Indonoa.la Jaa-Madoce 1951-62 .350 .73 A-crge rooponoen ore caynic-d itb tibyarte (196)) It A &.C (7->1

I... . .. . F ..

tlna btrcd iln Ilt ln...t form

Riom Indot.a.. L. aaiod 1951-62 .20- .s09p, .YIl ap alItc..ci by imla

Ric. K.I.ylayi i.&coc*nt Z.7.:. .61 (irec -a r adj_nc-nt. Ar1u9tail?P In A A C d0.!

R1cm l.iaoyal linds blame Valley Yea .61 Output o1.ntnIcty b.ned -n minro data,. Brhnc 4 Sqoee (19767 im Sln$iO-

Rice aT.o L-(lotIon 1962-72 .22 .9 iakte eoriortap ca.Chine (19765 in 6-lbonn

Rtica Tl.iqa " Ntion 1967-66 Tee 1.25 Onipot rePnnoao- RUne. a onmetod Potapanlom (1976) in t,'h-oteatrcted profIt niCijot.

Rice Korto RatIon 1960-71 f.2.00 Anr-ogu *ad yiold roync.lilolatrY of A11ln-ln- (79731 it

RIca Sri Lanlk. Ration 1953-74 .21 .99 A-ra~~ c. .yncAan o)(77IcSla

Rice Egypt NatIo 1920-40 .530, 50 *..62 Aeca_Ie.ooe. ereoe meda. AIsr C,enriag & uiarik IT!iIgyn ....drgn eaInoanic load A 6 C (1976)Rina Egypt titalo 1953-71 .02 .03 .79 reCer- programe. The po.rpono at thi.

Rice Egypt Ration, 1920-40 ..21n..i .56 YlaId ...p...e..

alce Egypt Nation 1953-72 .06 .08 .42 Yield .a. pnnea..

tAblo 11 PRICE ELASTICITY OF SfMVL. OP ACRICI.TtNaA. PPODICTS IN LDC. (c..ti..d) Appadint It

Farme _Price Elatitcitice Adjuetneet reef.r-d- Sh-t risn I.-, Arl. Fr ~ ~ ~ ~ ~ ~ ~~~~~* c* o SorcProwduct founere R ecroon Pert...! u rvey 55.,rc rl n L.w' run Are Pric -- R2 ____:_:_:_:':_--'______________

Rico Iraq Nation 6fl000 G ''0 ' I Pu -. ... v;; A.OVte *fl-Bs. dsn -o*l. A.ekri 6 Cu-ainC (1976)) Ir-q I..^ .. trop ... nIv I..d

R,ice Irq NRtIon 1961-71 .66* 1.57 39 ) reform progroo.

Rico lr.q Wation 1950-60 -.43' .61 .66 ) Yld re-poc....

Rio. tr-q 1Nciou 1961-71 L.820* 2.24 .8751e. P.N Mctlo 1946-66 0.50 0otp.t re-po.n.. .ith a N-ecovi.n pri- H..rrill (1967) in A A C (1976)

l'P.,.po onapproch. Rice prico 1.doflutd by .oot of lcvin- indon

iheac lrdla PonJob 1950-67 .10 .13 ACre--g re-pone C t.ng (1975) In Sc.hh.

Snect India Ponjob 1960-69 .02 to .0S- 1i5 te .sa- .14 to .17 1erloe - i.n del, OS. KVul 6 Sidhb (1971) in A & C k19O 0

71~hea Indil Natin 1931-57 .16 -Ierlauic dy i model. NCAER (1960) In A 6 C (Ii )

l.R e&t Pakl.tip Rainlon 1933-SS .10 to .20 Far Iroigoted cr. Stmplifted crc Falcon (1963) In A & C (1976)d1:u.t-men,t ; eodel R"iv p .... ofdeat to the mOno of -crops 1. usLd.

Pukilatc . ::.u- ;--35 _7 a. A 6 C In A 6 C (1976)

wheot Egypt NKtion 1920-60 .01 .02 .37 Ar-ng re-p...on. of -erl.osin roJ-. A ar-i. C..-., 6 14ri: cilblent C0ypt Nation 19)3-72 ..23e -.68 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~A 6 C (1976)Vh..' Ejgypt N^tL. 1953-72 -.23- -.68 .7d Thl. -Wy I1.o -.4-- ch. ..zPply

r-epo.en b-foro ond after enten- i-, lnd

'.'he. Egypt K-Nir. 1920-40 .01 01 67 ) YlioJ reeionea before acd after lnd A.khri 6 Cnnilng. (i976)) reof.rm Tie .cchore felt tl,at yielJ

67.000 CEsyt iN.tian 1953-72 91-* 44 .8 resoSn ara rulavoct aupply rouyOnrosVh..t Eup. In~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1 Egypt.

i'h.t SyrIl N aIn 1947-60 -.02 -.03 .74 ) A-reetn re-pou.on-f N-irlcin nod.l Ak-arl & Cu-Lngn (1976)W) hfor nJ ft-r land refom.

Oh.t Syria Nrtion 1961-72 .64' 3.23 .57 )

Vh.-t Sync Netlon 1947.60 1.32*' 1.44 .65 V Yeld reaponon before ..d after land ref. Ahaci 6 CurI.n. (1976) ;

shnot Syrl CatIon 1961-72 -2 37* -1.07 .57 ) Thn c 5.ati- yloid r-oaoc n_y la-a to AO.lri 6 Cunuinbo (1976)do 6ith a.pl -anotuic of nargincl Ilad.

vho t Ir-q CatIon 1951-60 .40 -4642 .5 ) Atc,r-o re-pono nf drloniun odn A.I.et & Cno.gn (9tuc.- ur^:d aftcr 1I d r:f-

*b+st.~ Ornq Yatioc tat7..71 - P1*0 -' 4 .77

Wh-cot lrq Nation 19S1-60 2.10..* 2.3) .91 YTled r..po.n.e bhforr and ft.e loud A.krt & Cu-ian.. (197.reform.

Uheat Ir-q Natin 1962-71 1. 59- 1.96 .70

-htet Jerdcn Wlio 1955-67 .20 .23 .66 Acre- p. r ..pon.e of .erla-l- nod.l. Aak.ti 6 C-uaicOc (1976)

whe.t Jordc Notion 19S5-67 -.66 -.77 .64 Y.ld r-.pae.. A61. norginal laud Aek-ri & Cucait. (1976)o. 'cvr-- . -- 19'!-'' ~ n7 .39 ,. -r ;.* ;, ::!: A9k-l otI6 C-noIos (1976)

Ub%e.t [Lb-nn -iatln 1951-72 .56-0 .08 .29 Yield r-pna....

Lheat inenya batltc 1950-69 31-0 .65 A.r-aSn r-pnson for lr." f-ruo.r ouly. y tha (197.) In S5b!anNeLvlnrd.1 Ath FI.h.r dli-rib-.sd

Barley ndl. Natinn 1938-57 .16 Nan1-ion dynanl1 -adnl. KCAER (1960) In A 6 C (19

7o)

B-r;y IrtJI PuaJab 19SI-64 53 60 .88 tAul (1967) In A 6 C (1976)

eo ale paakltan Notion 1951-68 .03 .02 Cuno.I. (1975) A 6 C

Barley Br-ili Rio Cr.ndn d. lol 1970-71 .22 to .6i 2.50t, 1.1 Arc oe ptncSe i 510g. 10610 6 Rnyrr in Sobh n

Barley Syria Nation ~~~~~~~~1947-60 -.15 -.24 .95 7 AfC .... ...npouncO. . olala Wnd.i A.k.ri. C-1ning & Harnk (1974) InB rley Syrlt iS ilo 1947-60 - IS - 24 5S ) before .nd fcor Iand rnf-o. A 6 C (1976)Barley Syria 9.0100 1961-72 -.57u -.57' .33

Earley Syria Ntlon 1947-60 1. 67* 2.69 .79 Yilld I c.pon.e..

Barley SyriS N-tion 1961-72 .27 .0 50

Appendix 11

Ta1ile i: PRICE LASTIClTY OF StPrLY OF ACRICYLTUaAL IlRODDgiCTS IN LDCc (continued)

°-~tlct Cr' ;t rv Retion Period Servey Si,rt. no Lo.:. r. . Are Prier R2

Ctr-,cts l-rLt Ndr I Rhode and dSlo_ Sterco

arl.o Iraq Nation 1951-60 .16 -.28 91 Acreags re-ponses before and after land Ackari, C-norings & torikretorto. (67n aoi wae a z. (ift.

Barley Iraq Kation 1962-71 -.05 -.19 .89

Barley Iraq Nation 1951-60 .51 .35 .74 Yield rasponsos before end after lendreform.

barley Iraq Nction 1962-71 .786' 21.51 .87

Barley Jordan Nation 1955-67 .61 -.45 .26 Acreoge rsponse.

Barley Jordan 11ation 1955-67 2.65 4.04 .52 Yield reapen.ec.

BZ-tey Lebanon N-tion 1951-72 .17 .22 .67 Acreage rcspone...

EBrley Lebanon Nation 1951-72 .09 .10 .63 Yield runponac.

Barley India PanJab 1960-69 .11 to .13 .14 to .16 .79 to .85 derlonian nodel Incldinge crop yield, Karl 6 Sidh. (1971) in A & Cr-infali. and trond varlable, cati- (1976)

Naice ?htlippir.e dAtlon 1910-41 (A) .09 Acr-onu roenen are cnplnined with N,,gahlc (1966)t112ietd nvl- Od of own primt, altrnatIve

Halos pbilieetn... Notion 19486-4 (87 .12 cyprino, factor price and ntooia'-' -~ ~~' ~~ ~; -: _:tire

trcnJ or a yield rato. Thrun -arintaen ..rtrc -nrc.. !r.: (A' i1 l. n 'sA pr:ceand oll alt-rnative crop prices.

eie K..enya Nation 1950-69 .95 2.4J AcreAge roapon.o for large farmers only. tlaichir (1974) In Sobltan

Paine Egypt Nation 1920-40 .03 .04 .61 5 A-reogo r-npoec-a of .ariovien Jodal Aektri e C-aing. end Hrikb.for a*nd after land reform. (1976)

aie Egypt Nation 1953-72 -.l8c -.13 .74

K.Lolo Egypt Nation 1920-40 -.16... -.25 .75 Yield response before and after lendre form.

"Moiso Egypt atien 1953-72 .04 .C9 bR

iaiae Syria Xation 1947-60 .51 .69 .84 Arcoga r-spon.on before and after lad

"a_ico Syria sttrn 1961-72 2.270 2.16 .56

Nale Sdyria Na. in. 1947-60 _.45 -.61 .42 Yield responses before and after lsaIref on.m

11ae Syric Kettin 1961-72 -.65' -.73 .67

aloe Jordan NetIon 1955-66 -.21 -.25 .91 Acreage responses.

Isiar tertlan Nation 1955-66 6.13' 6.40 .60 Yield reapona...

Y t Lb.r-on Nation 1953-72 .13 .29 .93 Acreog ceapoosen

..r^< lnan cotton snSc-t - oe. - 38 .78 Yield reeponas.

MaIne Sadan Nation 1951-65 .23 .56

:_::'!" Tl I tInd tsypoog 1955-63 1.09* 1.090 .14 -Ierlonlan edael with yield and price and Bhbrnon (1965) in A sod Ctboir niandard devi-tiott-, L procedoro. (1976)

Can..o,a productIon is -pert ori-ctod(75.), end Is .c.c-ntratod in 2 province.

Bajra India Punj (4 1951-64 -.01 to -.50 -.02 to -1.58 isal (1967) In A 6 C (1976)dlntricts)

M1ill,t Syria. 1947-60 94*00 .54 A Aer ang rop-at-ca of Nnr!o-ls nodel Arebri, Cnneigng & larikMillet 1961-72 l.2l~~~~~~~~~~~~O 1.60 ) boforo and after land reform. (1976)~~~bfor a,d aterlan (176M1let Sr 1961-72 1. 21- 1.60)

Millet Syria 1947-60 1.000* 1.78 Yield responses before and after land

Millet Syria 1961-72 .11 .14

Ml4 let

- -~~75

0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 0

* - 0 0 ]

E C O '

~C0O C 00 0 _ On 0 0

0~~ * ' O . 0_

0: Ct 0. S0 _0 < 0

0, 'I0.~ ~ _ _00

~~ io o4 e1 D~i = ;z. . 4

-O .0 - -S0_

'ot ,.C .0O U . C0 . J. EO .0 * O. . U00 0

*' 0 0. 0

JO O O . 0 . . .. .- . ~ . ., ., . . .1 .

0 0 0.0.1.01. toOl~~~.0 0 0 3. .. 1 00 0

s~~~~~ 0S 00000SC t o seD .O

4 ~ ~~~~ ~ '1. 0 0 0 0 0i ;

-76-

REFERENCES - APPENDIX II

Aromdee, Virach, Economics of Rice Trade Among Countries of South East Asia,Ph.D. Dissertation, University of Minnesota (1968).

Aromdee, Virach, "Can West Malaysia Become Self-Sufficient in Rice by 1975?",Malayan Economic Review, XIV (October 1969), 78-83.

Askari, Hossein and Cummings, John T. (1976), Agricultural Supply Responses -A Survey of Econometric Evidence, Praeger Publishers. A brief summaryalso appeared in the International Economic Review, 18:2 (June 1977),257-92.

Askari, Hossein and Cummings, John T., "Supply Response of Farmers withHeterogeneous Land," Indian Journal of Agricultural Economics, XXVI(January-March, 1976), 13-22.

Askari, Hossein, Cummings, John T. and Bassam, Harik, "Land Reform in theMiddle East," International Journal of Middle East Studies (publicationforthcoming 1977).

Askari, Hossein, Supply Response in Underdeveloped Agriculture: A CaseStudy of Four Major Annual Crops in Thailand 1937-1963 (Amsterdam:North-Holland Publishing Company, 1968).

Chinn, Dennis L. (1976), "The Marketable Surplus of Subsistence Crop: PaddyRice in Taiwan," AJAE, August.

Cummings, J.T., "The Supply Responsiveness of Indian Farmers in the Post-Independence Period," Indian Journal of Agricultural Economics, XXX(January-March, 1975), 25-40.

Cummings, J.T., "Cultivator Responsiveness in Pakistan - Cereal and CashCrops," The Pakistan Development Review, XIV (Autumn, 1975), 261-273.

Falcon, Walter P., "Farmer Response to Price in a Subsistence Economy: TheCase of West Pakistan," American Economic Review, LIV (Papers and Pro-ceedings) (May 1964), 580-591.

Griliches, Z., "Distributed Lags: A Survey," Econometrica, 35 (January 1967)16-47.

Hussain, Sayed Mushtaq, "A Note on Farmer Response to Price in East Pakistan,"The Pakistan Development Review, IV (Spring, 1964), 93-106.

Kaul, Jawahar L., "A Study of Supply Responses to Price of Punjab Crops,"Indian Journal of Economics, XLVII (July 1967), 25-40.

-77-

Kaul, Jawahar L. and Sidhu, D.S., "Acreage Response to Prices for Major

Crops in Punjab - an Econometric Study," Indian Journal of Agricul-

tural Economics, XXVI (October-December 1971), 427-434.

Koyck, L.M., Distributed Lags and Investment Analysis, Amsterdam, North-

Holland Publishing Co., 1954.

Madhavan, M.C., "Acreage Response of Indian Farmers: A Case Study ofTamil Nadu," Indian Journal of Agricultural Economics, XXVII (January-March 1972), 67-86.

Maitha, Joseph K., "A Supply Function for Kenyan Coffee," Eastern African

Economic Review, I (June 1969), 63-72.

Maitha, Joseph K., "Productivity Response to Price: A Case Study of Kenyan

Coffee," Eastern African Economic Review, II (December 1970), 31-38.

Mangahas, Mahar, Recto, Aida E. and Rattan, V.W., "Price and Market Relation-

ships for Rice and Corn in the Philippines," Journal of Farm Economics,

XLVIII (August 1966), 685-703.

Medani, A.I., "The Supply Response of African Farmers in Sudan to Price,"

Tropical Agriculture, XLVII (July 1970), 183-188.

Medani, A.I., "Elasticity of Marketable Surplus of a Subsistence Crop at

Various Stages of Development," Economic Development and CulturalChange, XXIII (April 1975), 421-430.

Merrill, William C., "Setting the Price of Peruvian Rice," Journal of Farm

Economics, XLIV (May 1967), 389-402.

Ministry of Agriculture & Fisheries, Government of Korea, "An Analysis ofSupply and Demand Structure for Rice in Korea," 1973.

Mubyarto, The Elasticity of the Marketable Surplus of Rice in Indonesia: A

Study of Java-Madura, Iowa State University, Ph.D. Thesis (1965).

National Council of Applied Economic Research, Long-Term Projections of

Demand for and Supply of Selected Agricultural Commodities 1960-61to 1975-76 (New Delhi, 1962).

Nerlove, Marc, "Estimates of the Elasticities of Supply of Selected Agri-

cultural Commodities," Journal of Farm Economics, 38 (May 1956),496-508.

Nerlove, Marc, "Distributed Lags and Estimation of Long-run Supply and

Demand Elasticities: Theoretical Considerations," Journal of Farm

Economics, 40, 1958, 301-11.

-78-

Nowshirvani, Vahid, "A Note on the Elasticity of the Marketable Surplus ofa Surplus Crop -- A Comment," Indian Journal of Agricultural Economics,XXII (January-March, 1967), 110-114.

Nowshirvani, Vahid, Agricultural Supply in India: Some Theoretical andEmpirical Studies, Ph.D. Thesis, M.I.T. (1968).

Rajogopalan, V., Supply Response for Irrigated Crops in Madras State, India,Ph.D. Dissertation, University of Tennessee (1967).

Singh, Anoop, "Sri Lanka: The Response of Paddy Producers to Price Stimuli,"1975/77, International Monetary Fund, Asian Department, Mimeo (1975).

Sobhan, Iqbal, "Agricultural (Price) Policy and Supply Response - A Reviewof Evidence and an Interpretation for Policy," The World Bank, AGREPWorking Paper, Mimeo, July 1977.

Squire, Lyn and Barnum, H.N., "An Econometric Model of an Agricultural House-hold," Mimeo, World Bank, 1976.

Toquero, Z. et al, "Marketable Surplus Functions for a Subsistence Crop: Ricein the Philippines," AJAE, November 1975.

-79-

Appendix III

PRICE EFFECTS ON DEMANDS OF AGRICULTURAL PRODUCTS

I. Introduction

1.01 Among its many wide-range applications, estimation of price

elasticity of demand is particularly useful for: demand projection andindustrial forecasting, sector economic planning, commodity marketstabilization, and in evaluating the effects of projects or governmentpolicies of protection and subsidy. Thus, statistical evidence of demand

elasticities is available for many diverse data samples in accordancewith different purposes of estimation. In this survey, we have includedanalyses based on: (i) single country time series data, (ii) singlecountry household survey data, (iii) single country panel data (i.e., surveyfollow-up data with the same households), (iv) cross-nation time seriesdata, and, (v), internationally aggregated data.

1.02 It is obvious that price elasticities estimated from these

diverse samples cannot be directly compared. It is then up to planners'judgment to choose appropriate estimates in analysis. It is commonlyassumed 1/ that elasticities obtained from the cross-section data characterizethe long-run behavior, while the results obtained from time-series dataare more appropriate in medium or short-run projection. However, this

assumption does not imply that cross-section elsticities are necessarily

greater than time series ones. For one thing, the measurements of variables

are generally different from two approaches: disposable income is typically

used in time-series analysis while expenditure level is often the onlyincome-like variable available in household surveys. In many cases, pricevariables are not even introduced in the demand specification of household

survey studies. In addition, the demand theory framework used in twoapproaches can be very different. As a result, the difference between

the magnitudes of demand elasticities so estimated cannot be totallyattributed to the distinction between long-run and short-run phenomena.-In fact, some authors even believe that both approaches describe the same

universal phenomenon. 2/ Mundlak (1964), for example, observes that the twosets of estimates tend to agree when broader classification of agricultural

products are used. 3/

1.03 The degree of commodity disaggregation should depend on the purposeof estimation. From the planner's viewpoint, a broader classification of

agricultural products is generally sufficient. But, from the industryforecasting aspect, more detailed breakdown is needed. Nevertheless,

1/ For example, Holmes (1968), Bussink (1970), Houthakker (1965), NCAER (1970),

and Weisskof (1971) all share this view. In fewer incidences the opposite

view is also held. One example of the latter is Anderson (1969) who views

the cross-section results as static results and a downward bias in

elasticity estimation.

2/ Such as Weisskoff (1971) and Mundlak (1964).

3/ Mundlak (1964), p. 120.

-80-

aggregation in commodity will reduce the magnitude of substitution effect.Even within the same industry, the substitution effects of demand forproducts of different processing stages may show a large variation. Forexample, demand for chocolate products will have different price elasticityfrom that of cocoa beans. Therefore, it is possible for a policy to havesubstantial effect on demand structure of food items although the aggregatefood demand remains rather inelastic to price changes. Also, it ispossible to change the magnitudes of agricultural intermediate inputs'demand by price policies, even if the final demands of these productsmay be price inelastic.

1.04 Readers are also reminded that demand elasticities estimated fromhousehold survey data refer to household consumption demand. Thus, industrialdemand for materials and intermediate inputs and farm demand for feed arenot included in estimating demand function 6f food items. This certainlycan be a source of important differences in the estimated size of theelasticities with time series estimates. Time series studies, in fact,measure demand for a given food item by the sum of production and imports(net of exports) adjusted by changes in stocks.

1.05 It is not our intention to compare the superiority of the time-seriesversus household survey approach. Rather, we would like to point out thateach type of study has its own advantages and shortcomings. For example,while it is easier to evaluate price elasticities in the time-series approach,these estimates tend to be biased because of the collinearity propertybetween income and price data over time. Cross sections, on the other hand,are more convenient to include the effects on demand of socio-economicfactors such as occupation and education of household head, age of familymembers, family size, geographic location of the household, etc.

1.06 A particular segment of the literature on aggregate demand analysisconcerns the so called commodity studies. Demand elasticities in thesestudies refer to the responses in world demand for a specific commodityto the changes in its price. 1/ These elasticities are directly applicableto commodity stabilization studies. While they are not directly usefulto individual country's analysis, if additional information such as the marketshare of the country and the supply elasticity of the rest of the world areprovided, from the world-wide aggregated elasticities one can estimate thedemand elasticities for a particular country. These elasticities in turncan be used in national policy evaluation.

II. Empirical Evidences of Price Effects

2.01 Our survey of elasticity estimates is systematized into two tablesaccording to the nature of data sample. Table 1 contains results forindividual Less Developed Country's estimates, while Table 2 summarizes worldwide aggregate commodity studies. These tables start with demand for all

1/ The commodity price is measured either by weighted price index (includingexport unit valued index) or the spot price in commodity market.

-81-

food, then continue to list the elasticitiesof major agriculturalcrops such as: rice, wheat, maize, root crops, pulses, vegetables, fruits,sugar, meats (beef, pork, poultry, fish), oils, milk, milk products, cocoa,coffee, tea, tobacco, cotton, wool, etc. The survey is by no means exhaustiveand is limited to the Anglo-American literature with only occasionalexcursions in studies carried out in LDC's {mainly India, Pakistan,Philippines).

2.02 The front section of Table 1 consists of total food demand.Lluch (1977) and Weisskoff (1971) are two major sources of these elasticities.Lluch (1977) has applied the ELES (Extended Linear Expenditure System) modelin estimate aggregate demands of 17 countries. This ELES model differsfrom the well-known LES 1/ (Linear Expenditure System) by allowing for theendogenous determination of total expenditure level. In the estimation, amaximum likelihood iteration method was used. The price elasticity isindirectly derived from the expenditure elasticity and other estimatedparameters such as average budget shares. While Weisskoff (1971) included15 countries in his study, only estimates for LDCs are presented. No systematicdifferences in the elasticity estimates appear to exist between the twostudies.

2.03 In the individual countries' demand analysis for major crops,(reported in Table 1) the most common models are the log linear singleequation and the Frisch scheme. The former generally gives rather goodstatistical fitting. The popularity of Frisch scheme, on the other hand, isprobably due to the simplicity of its computation; instead of estimatingdirectly all cross price elasticities, extraneous information are used toestimate the money flexibility parameter first, and then to derive all crossprice elasticities. To replace this method, a simultaneous system of demandequations is sometimes needed to estimate a set of interrelated demands.Haessel's (1976) study is one such example. Results are reported in Table 1.

2.04 In order to have an overall picture on the magnitudes of priceelasticities that we have collected, Table 3 summarizes some of the informationof the first two tables. This table shows that the majority of LDCs haveprice elasticity of total food demand within the range of one third to twothirds. For "necessities" such as rice, wheat, oils and sugar, morecountries have price elasticities lower than one third. For the same items,however, some countries have higher-than-one price elasticities. In thesecases the specific item is generally not a traditional food item in the nationaldiet. As for'other relative luxury or semi-luxury items such as meats anddairy products, price elasticities often fall within the range of one-thirdto two-thirds. Other non-necessity food items including coffee, tea, cocoaand tobacco appear to have very low price elasticities.

1/ References for LES include: Klein-Rubin (1947-48), Samuelson (1947-48),Stone (1954), Pollak (1969), and Parks (1971).

Fro!.d Col. 195 -rid6 -o (.ntn . .Fj.,.(3 ...10 ree -- 0t. IIe ln. th. dep-oIe7, t -".b on 0,1,0 0 SoofrI9I,oItt df.t

Food,~~~~ooIo0 lao,10d0,b 1951-70 o1O 0.00 01:00000 n~~~~~~~~~~~~~~~~~~~~Od, p', ib1,I i. if -t-.I . I---.. .7 A,I,..k.f (197)30Food Chn Eo,lod. 111-9.45- 0.450 .. I Fol-pldtto..e.tt-.-b0.1:1d .b- - Ailnoir (19736) 1(16!Food ChylunC t.119648-6 . -.4~0 ~ 0.934S-y Fo d r tpn,'If 13 -rbt. dopod p.I -oio inly,tdle itoh (1971 -d.l.io dS- to

Feed ..,.P.bIL~~~~~~~~~~~~~ 1950-64 -6ooorolb- r451 086lo. 937 fin...fro,0rlo077f og ol001:kSe 0970 Oce 1971(171

Food ChI..S 1586 -.45* 0.73~ -.- d1 - i-ioLnL oh(177dI tPooc-d-d i..(1965..)S.t

(ELES) no 61. p-g-c of Ba1:d -c aplIli:d In -atlo,ati-. 0.997 1.o-1, (1977)Food ....c 0950-65 0.412 1.176010 0.036 beIo-keff (1971)

.0.8~~9000 o~.oo a.. US noda1, OL5 nith It ...tion IFr ..olInI--- ftc.,o 0.996 Gol16b-g- (1966)3I .0.03400 0.73 can 1 1"

Food Et-oe- 1950-64 7.831 1 4 1 4 aa, 0.683 0Io,-(l?)Food Ilooduona 1954-64 0.373 0.916000 0.952 41:lk,,Uf ((975)F.noOlo SniI1).6 -.335000 0.48900.. Double Iog noda. Sloll I.... Its abtolnad It- o l.n.. nodal-0071 (9

Fodoona Sdla MM16 .90 0.7 I Dooble lag nodal. 0.61 P.od.y (1973)Food j-inolO 1959 -65 -1.278 0.69O0". 0.455 WelCehoff (1971)

* ~~~~~~~, ~~~ 1959-66 -.47 0.38 - ELlS 0.931 Ilo-ob (19773±/Food Fob - 1950.50 .l.1950-58 0.98700... 0.999 Olc-o,,c(f (1971)

0.26 0.75 . - C11b-.,uo-p-iod In W.t-obt (P)62)Foed Ir-l-d 1953-64 0.1 5700 0.6170 0.793 W,i,akoff (L971)

1955.68 -.43 0.64 rLES0 0.998 LI-c, (1977)Food 70..0 1553 -64 -.165 0.93200.. 0.331 Ocloi..k.f (1071)

1991-60 -.47 0.72 01.0 0.999 LI-ch (1977)Knrol -urban 1963-172 -.219 8710 - 1.1-1,

-Korca-r.1a 1963-72 -.457 871.87 L-,Fnod P- no 1960.68 -.70 0.92 E1.0S 0.976 L1.oo, (1977)Food Philippi.a 1953-65 -.35 0.52 - 87.87 0.996 L1.ch01 (1977)Food 08870000 9557050. 9 .94 .0.,635 W.eiookorf (1971)

.7' 01,90.271 41:laOkf! (1971)1955-67 -.50 0.49 ELES0 0.993 L1.oo, (1977)

Food Rhodacia 1954-63 -.128 0.737 .8 - ktf191Feed 00001 Afrlo- 1950-66 0.018 . 0.5000... 0.441 W.c 1,.6ff (19781

1955-68 -.3a 0.00 . ISS 0.998 LIo-h (1977)Feed Toloal 1958-65 -.07 1.

190l009 oocl 1971)

1955-68 -. I. 0.7 01.0L0 0,995 Lluc. 297MFood Thailood 195 7-66 1.lIS710n 0.635~ 0.999 Wei..kolf (1977)

19fO-69 -.69. 0'84 01.0 0.979 L1.1,6 (1977)Food W. FPki.tan 1963-64 ye. -.46 -2.51 0.63 Frloub ..bo,o(nn Indapenden 0001 .nd otilino the coo...pt oIf lo-m .1wi.t1 y - Boonlnk (8970) I/-

If narginal taIlly Ifeo-ey. E.n oditor. of 10 II,-o groups nra anolyrd. D.t. N.)

Food lOrnol 1952-64 -.106600 0.248 .o.didd t - d b. 0.793 Wolonboff (1971)Food I-roo .4630 0.64 ... KModlak (1964)Food 0. :0,1 1959.68 -31. 0.66 871.87 0.998 LIuh (1977)Foed Y.&.,lnv-n 1972 Ye.0 .58o.65 .79..90 Survey dt.a of 3 0001i0- ad 7 to.- typo, no- anly.ed 001th E7ES7 nodoI. - LI_ch(1977Food 4 Lo, 1. . .,

Meerlon L80. 1966.69 Ye. Total af 4112 ob-orotion, p1:11.- Jlo.t1ot1.a -c -tf-t.o ftin. E71.S nodel, nod HIon 6 Muogr-v Io L)u- (1977)B0401 (CoL.uI) -.002 0.66 thne aoodor l l Icoar ftro doobl. - log no l.7.

Coo.. no(O -oo) -.234 0.6Cupgol(tEoaC e -.614 0.61L.00 (Frn) -.05 0.47

Total Careals WInd 1951166 -.3700 0.61600.. Stellar o... I. obt.toed fron -t.oa nd doobla log node.I- NC4ER (1970)SodIn 195 1_68-5 ...50 0.9 lub Ie log noda 0.67 P.oodoy (1(973)Chnuo 1953-70 .2.320 0.79120 -2.32 (p-) Si.l u .,euy lao of derond q4uoti... too 3 go-loowroI [L.00 (6976)

geed l.. .o .t-tn,od ct16 51.0 0- hod.r'00000 or-lno Wlrleo - -.45 0.4 - Ilot`d elo..llotftI. I0 060 pr.jn,ti.. I1 c-n.dity trad. - World 1100k (1977)

I1. Sdie 19)1-68 ..71000 .94 Doublo log nodl. 0.57 Fandy 'Y(1973)W8. FP',',.n 06-64 Y.. (-.26-. 16) 0.05 - M.clh chton - Eouo 1970)Bongladealh 1951-70 -29 1.2 1 Soollg -ndel 0.02 A100481: (1971a)Meton 1953-70 -1.25 n. 0.0110 -550(oeoea) 4 -.oodity i.tau...... .700et0, 2S1.S. It....001(1976)

-1. 03' ly-n)-relua 000 po . ...11. nob-.--- J (1977)

World -.35 0.2 Q.otod.olu-tintiti I th l. projc-tin of -notdity t,nda - World So-k (8971)Itboot 12dL. 1950468 -.22 1.00 D..bl. 009 -od.1 0.87 Foodoy (1973) 1

W k. Fbltn 1905-64 ye -.10 0.21 -Foi.oh B ...o lu.ko (1970)-Ad~trgoltn 106)3 poe .03 0.16 0.6(c) Frf.,h Wueov, ol- .d oheoin .-: 100 - .b-g00up M 7 J-ouoy (8977)World . -.25 0.2 .1.1. tioonltil-. In th. pr.o)o..tlo of -o.dloy trade World na,k (7977)

01 Whec 671,: 1.loglodeob 1951-70 -.~7 0.. 1.97..0 1.00: ..&Ioo,ood:, -rdu-d to- -etlO,i-t 0.42 Alr.gtri W('17.)171110 li_n 1.53-700 -12.914 0.749- S -conodty Ol.uloooo 700, 312 -L,tue (1374)Cooroon Chon, ~~~~~1953-70 -.6410 .2 0.8461rioe), 6 o.--.dity LOtn, ...... s .070t, 151.S e lcol (976)

-2 .09iu...... 1)

Me. .,o. 1955.90 -.59530 0.5110 1. 5l9noi1. 5 eoondity ytolooo .t7.1 , 2SLS. - t-o-1 (8974)-.80(c t-)

Fo-otocO A,rg tOin 1963 Ye. -3..j0 0.038 Fooho.b -on,oo -- c -0c, (1972)___d_ 95148 -.49..0 0.32 Dnuhle lol: nodal07 ;,,ky(9

9.Fho,o 196 -64 yea. -0-6 0.16 F,P:".,1,07.0- .,:111 1770W00 I. Pob.tetn 1963)64 Y0e . 1-2 ~ 0:51 Flo,at,1 ,,,I, 17

Al-FeA I0,01 163 Yet 0.2 0 9a -.13 118r1 0c,,u-In1nur 1971)Wedll 1958468 -.05 Coo.odlly .-,dy -l.lo(1577)

App-di. I IIoo

* - ... C.-.. .-. ..... .-.~~~ ..,o,o ,o,ce Cc,xoo.ots sonur cojol an ?thod B2

Sourc

Noflipo-L. 1961-67 .50 S 70n .. -.31'(fi.h) Sioplo ti-,o o.d,t, -othly dot, 43 Olnyid: ,::0I1(1969)

l:IN-ri 1961-67 -51, 2. 00 ... ULo li1-1o -n~,od nootly u.L.t .47 Ooylde & Si

WorIld -1.25 (0.3 0.B) Q,oo elnsr-iolt,io .to 1. pr ~ o f -n.3dit7' t,08. ; oldBn (_7

Pe1, Arg,etino 96ye .41: 0. -0 I..) (o.6eI,oo yo of neat thlddi IeSkrnp)-tejaoy(92tork Argoo- n. 963 yn-.1 0.53 0.73 (b..f) Fol_ci _ro_ Ido Janvok (1977)

LWo.F.ktgcn tl. 1963-6 yen ( .40-.57 ; 2 I 0.13 (fi.h) Vole-h ghone D- J-ev`Y (1972)

I. Alg.ontio 1963 Y.. -.39'1 4.19 .0(pr Frinok -th-o Do J-o (1972)

Ed1bl. oil W1. Pokiet-n 1963-64 Ye. (-.030(p-k ooo, cle..Dl. Iek-- (1970)

C.....at 011 Philippino 1950067-.30' Siolo0oa5 imt f lon nl-, .9.1ko , (1971)(d-ti.Bl) 190o7.o0'1. li-. of0-.5 D'LJ-Y (19S2

Fa't. 6011l. W-rIl I.0 ot.dootIiyi k oje o of aon.odity _li0nd3 Borld.k (1977)

1t1k 8. P.ki.toa 1963-64 ye .2~3)10 ookeh.oLob...k (1971)

0:1 i olk . .1.a 1963-64 Yee (-.W -4- 0 .01. Foiso t aOtho,-Bnto:Ik.- Aoooi 1963 Yoe -.29 I.174- Prn-k .nhkWorl B -Jo ,r (1977)

oT-o Ao 0 .octin 1963 Yoe *. 24" 0. 17 Pt1cr6.tho-o- Ile.:Ikoo (1972)

Too '.3. Psktatan 1963.64 ~~~Ye. (..20-.29) 0.66e - oio.h sch- B.nik (90

Tohaco W. 1 k.61t.. 1963-64 Ye. (-.38- .08) 1.6 jII tb mo. 0060 De io ... k (19701

OM1 0000 5 0

0 1oco l06 j/yI t.opojoto eoaoit 0h3 - WooDo J--ok (1977)

iSY on Atg-tlln. 1963 Yee -:4.41840 - 2(.1 QFoitd oo..o 001,enoC.. tntn ssicuddi n nkrok oJn (19703

Banan AWooli -.3 -ele0.500 Quo-ed elnlt ltn a.dI r 30t -fc odity t oad .B ...l Bunk (1970)

Orator Ao~~~oat1na 1903 Yee -1.01 .. 0.52 .6 Vlc o,n.3tp.o ,l, ei 61. nodgroap W-I Jv-yk (1977)

~~DiO Aogeotioa 0963 ye. ~~~~~~~~~~~~0.30 0.446oa Frs) ?60h soes tyons of fril, so. Ln thi. C iyo de Be J.pvoy (1972)

roosotn Arg. t1(0 1963 yen.1.9? . .84 048 Vlshoooe Baol.nofo. ndolon*ndc. on abeop Do J-pc (90

Ao. 0 .oetn 1963.6 yes (-.18.9)0 0. 0.9 7O-I.. 1.okhouo-B.ik(90Vo..ooabioo 8. pokinIco 6963-66 ye. (-.0B-.32 0.84etsokt- - -i- -L thi. eou- -b_pD Bn.f-y (1972)

APto ArBe.k-t. 1963 y.-2 03 c4o g Do J00007 (1972)

ra t.istoll ngnft en .at 101 Isohn.h

yv~~~ 1 Teelstocyoei.alosofLlc's(973 od voik 190 .*1eored63ko ye. *taLi.theoBsinl(lakn o1970ioo2- / AS tcnuio yoeok(97) k o. ofPn0oehd. oo .- 10 ogtooosroe o,to 00 k nFr.i., o hapol ean w rceeototyoolupidk hedt oo-ait.3/ At ,die to Ioc (177) Ikedat ofiaplnicpol, Idoos *kt, erylitte vriaion. Tha, Ie rsolcni 184eta,pol. eeolcityshold oc e 1000 oolooly BoIdo, i Ik ta19tia2i

to. igolfiosec.~~~~~~~~~~0.

App-i1.i Ill

7 C)2: pjulc7 Iit icE V 1. II A6:41I ri (kO ECxiCOIiTUODPi EIAS7 PTiCITIF (IF ACR WUtLTUI1Al CI9IM1P!TiESl

____- LOC. 1956-71 -.9i7" **... '5:4050b.:gd..d wt "I .98 A-(96

Xe. 1957471 -Ig 86 2. 2.2 The nit.- pilot, ind- a caronetoI- ditiidat 'iRin, ThoIload 0.894 Ad... (1976)..9.Ui e It;.1 .Crop p,r.ice ludltittof detord oC

defatr ii histtdy. Local .a iiobitity I. as de soeothe edd!tico.i eI l aey out laba. ..e capita dowIud- ua

World -.35 0.2 Quoted itiitr coti-dity portin OD(97

Wheat ~~~~LOCe 1955Il -..l0g -.3l* .3 3 1.I whact price inotecaerted wit'h data of Argeetica. 0.94 Adoo. (1976)Vh..t I. 1957 ~ 35a * 2 t 200C-ado. USAhadIrn IIDul l~nde) with dintnlbuted lu, 0.71 Ad... (197h)

World -.25 0.3 Qu-ted -Inst Ic .. -sdi In -o-sility eeiln - IliRD (1977)

Coarne Goatee" World -.4.5 0.6 Quoted * litecticiie in cediodty P-ojCttl 00

Beef World -1.25 3.2 .8 Quoted. -i tictili. toe,1

In conendLty projection. Roge.it ncf..h.b - IDIRD (1977)welt -ae of n-porto i 'Jee.

B.n..a World -.3 0.hn .0.9 Quoted . loetlotica in c-modity proJectioo. e-eo enPoit Print - 1RD0 (1977)

W iurd 4 -.6--) (0,1- mIl) Cited Ii 101 1977). p. 42. - USD6 (1971)8.S. -.03 (0.2- 0.3) Cited 1.i lIt)Z (1977). p. 42. PADJapan -.95

boar LICe 19)5-73 -.05o .5o- 3.7 Double :..a model. Cematric log todd wee apied - DCe cod CPte 0.98 Ada.e (197hIDC% 1955-73 08 -.16 13.3l- 1.7 Su

6ac polo. indee W., c.oostn-otd with daeof UK. Cuha' eInan. 0.97 Ad.os (197h)

WorLd 1986 -.13... 68 4.0 and 11.3 0.97 Ad..a (1976)1928-65 -.03 - - Iloo~~~~~~~~~~~~~~~~~~~~~~~~~~~.thiy data. Co-ti-uao tte modeL 05.ic ejs ac.. ot s.oiy -inoIiho ppiouiateddiaret wod Iwoo estimated with PFlfl. o.thod. Lhs(9

Wlothd 0; ta ii co-ewptio-, producttlo so,d sLoth were cootrocted

1cm doia ol- 14 onotrill,.World .lm-4 3.1-4 Quoted lee.tlot.tIcs Ic t_idlty poJetti oo. (Soron..: PAD) - IBiD (1957)

Otle U1.S. - 1953-67 -.16 -. 5- - Annual data. Stoob odjoutnien .odel etlota-d with 2-stage 0.96 Lahys (1975)U.t. 1955467 -.42: -.0~ rodra ueite fothteooeu i sdpl i .9 Lha(95EtC 1953-67 -.20. - e te-dre u netiten o .2 eed1adn 0.73 Lehys (1975)

Jpn 1953-6 -64 -1.59* - elatic ty wa uputed for other coon.trin. 0.84 Lahys. (1975)

oils 0, Pats World -.13 -Price atote for Lodin iduu faean ils -o.eih - 18I1PI1 (1977Coconut 011 Wetld -.65 larger han that ior aggregate oIl denuod. - - 1015 (1977) h..tht .r Og, . I .-. d 1. I91

Palm hI Worl~ ~~d -.3- IRD D(1977)CI'a 611 Wonld -4.69 I 10R11D(1977)Soybean 01 World -8.601- IBID (1977tuoflnonrened 011 (uId -4.96 ' lOD (i 9771)

Cotnoo..eed oil World -2.0 1161(19771teoodn-t Oil World -1.3 - 1BR (177.Late World -2.6 R10 (1917)

L.r World -2.71 - B 011 (1977)

COcos I.Ce. 1953472 04 i'.A, .43* 1.01 1.3 Double .e ndol. The.na price iod-uwa coo.trcrtd with date 0.95 Adam (1976)DC.f 1955472 - -.

3ku to .1 of Tha.,llei. BaC- l. LCoroPad or oe .PatiIa 0.95 Adoa (I9)76

195 .122 .6d11 16adut. ot odl ealidoLC,an2yae'dirbnd0.99 - d 9t (96)loge wo s eppilold no DCs and TCt.,_ rt.LDCe -2 -.4 - -PrItnI reandhbyIhedollaed _o1t e alo Lahye (1975)

ICe -.~~ ~ ~~~24 -.4 2 - uccb (I'Ccaas iU tratcuoieo ou.-Lhe19)Wlorld (.2-- .47) (2.,6) Quoted .-I-tliCitlr-t c_sod pt prCe Projoctloc. hew Y,rhk 1B1 (177

World 1955-65 (-13- .45) - urel aaS oiu oiao d,nand were aetie tatd .90-95 i.yiena (1960)-

aod Loc-e -aribleeV... ..ne otloatnecr ie.1 i os

Ue.lS. -.27 Th.se seePA -iti .t. Indiretly nited to Rlemquint end Itae.ee. II1972) PAO

A..-rI. .35

Demark .49 -

berld (wotghtod) -.374Ctee (henna) Xe. -.5o. -.6) Rnport.d Io, fillIOb (1966) PItee (1955)

Teem Worl~~I;"d -.48 Repored In Killick (191661 AC__ W.. -:4 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Reported In BIojoie and iionl(1)72) Cocoo (1970)

S9D J-S1 A-o 1 1r30 0t)1 01 ...3;. 00 1.1049qs /l :*3e"zA

109 913 1' 3000121091 19 9

011011o39

ok^IlI I- (u>3ul 6 1 3 w1s5_0 ... 10 ils9 I% 1,l29-10 All-13131 S - 0-011033-

*OolJO.n t.aos ptov odO.oo)ttoOlOON 0 3-O O..3300313 " 019 03 93 50300) 030 000921 Jp--31 99300 0*°9 U3.* 0991ua 01y .3.0 ol_________

(CC61) aLl91 - *"'° 9sn Aq -dpnf uo;loo(Old I pocc oul 0 1910033 p .oOb 01 OJJJ-J wO 901 (1 o-%() S - I1 _ Fl° p00..10 *03-'

(CC61) Ori O - 0S- P9I0 0...qI

(9161) 00096 0~-o oopoti ooiol o03qos oI1o3po(d - q pu - 3 0013 PS 0193 0...t C °? * o *Oz- t o - I Z- 5561 03d8(916!) 03009 1 990 1 0 o-P *tJ p-oao;o...Iso * apuo I OD -I CZ - * t0 C-5(9161) *..'99 t99 0 01 9000 j. 0I10 .d 0.P-19 101 I- 930I 0 *o,..t 0 I Z (031 91 OZ - I1SS61 '0 ml 1°

| ~~~~~~~~~~~~~VS1 Pu. (o,onj<oo1-o YY1 09 o 0109 '03l 90100.01000 000 00909Li) (9561) * 0PV C60 00j9d 0011I 02d) 329 13010,3 id.P: I00 00 33P90 '1 Z9 lt3( _.9 tSJ(61 021200 (9L61) PV9 Z6 0 1191t)09 3000031050- I I 00 90 109u p. -nql-1l3p 030100006 300l l ,, *7 - (Z tC-5561 *00

(916.) ''7V 9660 A2tfl 00 1 6o9

s 3n0. 090p 9ol p. ,l9 3.,.P I q01 39 V 0

1 dl -9 Z0- El-S%61 '31 00110

X9Ato -p- 90-0 11- '113 .0-10003)9093003 00331 002200 A0 III0 1.... I. I'd 0 p..n 01 00jI

(1161) 93dI9 - P080 u0ptto0 'I'0I3U0 013 loofoJd 111poaoOl 003191139 3o0b (Z9 _51) 0t* C9309(9169) .. P9 990 0)931 tP 00-II. 333 30 01.P 411.' P0100.3--00009303103 9. 6* &(3 08(9161) 0c0pY90 9( 091u p3u- qux 90910 0 91°0 10000033 10100 0309lO 90 '29U 10)d 9a C9 t _- - IC-SSol 03d9(9L61) ...vV °t' 0 .3da uo" Pald "eIP uZnp I}'dw ut oJ t ta91l-561(9!61) * 'PY 06 0 ".ap.. .1 p.11d,q ... ..n saod.J 3t ,Je3S ..LO'-I-poc Sol *Iq'°lL C 17 t0 _ _ IL-5961 *311

(IL61) 9V91 ... I o - - ,ji.d.1

(LC6l) CSSI - balJd 401@ 1"A ^aN a~q, lJaa Ap.. gIq] l .* .. d OwSo 1 °1 sn(Ct61) LDd31 - ls~~~~~~~Al u..14-f-d Ai;po7o . } . aX3rt .@tn ez oIz q>(IC61) 0199 - IL sSJ, t

(1161) '369Y 9C0 0oPoI. 34

(9616) 0 w09 rs o 001.09 J909 t '10031000 ... .... 3.9....dS..j.... fv002...) 0131 ' 0097 JO - - ZL-S-61 (69 038(16) 00099 96 000a3llOtt030 093309 01 3333)0910100 ' 9 It 19(1 099 v V610 .. 30 .30 I.) ttz~~~~~~~~~~~~~~~~~~~~~~~~00 00P16) 099 930 9300300 031011) 03 0 01 1.0 )109330 10 01"t39 oOl

t,* - 99! o- --: a3

12390 Y 30 OoOooto Jon}s]¢ 3301 10110 0919 01Ew u0 1'313 030rs 9993 33n (030 [ 9tt_(0L61) lu 90 sldt -d I'}n ^ qr5w^ aLldduz)two) I6+9-t -a C9tt6 Ml oav*|

~~~~~99 303b3.3 4AAk 00 01V4 01091J 0091) 9000010 9f<> l0O9 wJVl (6Z90 '7--61303 (1- ) 69-61 P I 00 (3. * -

9 zsinb lold CZd o 4D_v 1. q *-lJ Zrxzas ao u pw _ z_6-9t6 P..°

¶ snol 0033001 90300 11qcsw 0009 00003 30ar 00000s 0013i 9000.0 930J901 013)001 .nL Z(- ---- IZ-) 69-9969 90 0003 000033

(1161) 90001) 030001 103310) 3 1J 99 ZE

9 10)060039 9960 0990310_ _03 090 93... _390 00if13(p0 9 0 , 13-9 )30 0000

10139 9 0*.00~,fololt 01 t... 303--qi . *- -. 0 100-09000 -,:_0.

(OM) 1.0 zo d.) .43 1. ."I I" Z 0-061 .1 ik ..- a~~~~~~~~~~~~~~~~.03033 9.-P-', I. '.. .:.,)g,- -(~~~~~~~~~~~~~~~~~~~~~~~~~~~~~3303016

-86-

At4ailduiX III

Table 3

Numerical Ranges of Price Elasticitiesof Food Demand in LDCs and Other Aggregate Studies

CountriesFood Category Less than 1/3 1/3 - 2/3 2/3 - 1 More than 1

Total Food (exp) Ireland** Chile*** Panama I Sri Lanka***Korea (ufban) Greece*** Nigeria Dominicfn RepublicRhodesia HondurasI Thailand Ecuador13 OECD Jamaica 15 LDCs Israel***countries Korea (rural) within Peru***

Philippines Thailand***Puerto RicoS. AfricaTaiwanW. PakistanIsrael***Yugoslavia15 LDCs(between)

Foodgrains Bangladesh** India***World

Total Cereals India*** Ghana*

Rice W. Pakistan Argentina** India*** Ghana***Bangladesh Ivory CoastPhilippines**LDCs World DCs

Wheat IndiaI Taiwan***W. Pakistan LDCsArgentina DCs

Rice & Wheat Bangladesh***

Maize Ivory Coast GhanaIPhilippines

Roots Argentina** Ghana* Ghana*Israel**

Pulses W. Pakistan India***

Onion Argentina**

Vegetables Israel**

-87-

Appendix III

Table 3 (cont.)

CountriesFood Category Less than 1/3 1/3 - 2/3 2/3 - 1 More than 1

Meats

Beef Nigeria Ivory CoastArgentina World

Pork Argentina

Poultry Israel

Lamb Argentina

Fish W. PakistanArgentina

Oils

Edible W. Pakistan

Coconut Oil Philippines

Butter Israel

Dairy Products Argentina** W. PakistanIsrael** Israel

Sugar W. Pakistan Argentina**LDCs CPEsDCsWorld

Coffee & Tea Argentina CPEsW. PakistanLDCsDCs

Cocoa World

Tobacco W. PakistanIvory Coast

Fruits & Nuts Argentina** Argentina** Argentina**Ivory Coast Israel**

Notations: ***: Statistically significant at 1% level**: Statistically significant at 10% level*: Statistically significant at 30% levelI: Statistically insignificant

SOURCE: Extracted from Table 1 in Annex.

-88-

REFERENCES - APPENDIX III

Adams, F. Gerard and Jere R. Behrman (1976), Econometric Models of WorldAgricultural Commodity Markets, Ballinger Publishing Co., Mass.

Alamgir, M. and J.J.B. Berlage (1973a), "Foodgrain (Rice and Wheat) Demand,Import and Price Policy for Bangladesh," The Bangladesh Economic Review,1:1, 25-58.

(1973b), "Estimation of Income Elasticity of Demandfor Foodgrain in Bangladesh from Cross Section Data: A Skeptical View,"The Bangladesh Economic Review, 1:4,387-408.

Anderson, P.S. et al. (1969), Projections of Supply and Demand for Agricul-tural Products in Denmark (1970-19580), AARHUS UNIVERSITETS OKONOMISKEINSTITUT.

Bard, Y. (1967), Nonlinear Parameter Estimation and Programming, IBM Con-tributed Program Library, 360-D-13.6.003.

Blomqvist, A.G. and W. Haessel (1972), "The Price Elasticity of Demand forGhana's Cocoa: Some Estimates and Policy Implications," EconomicBulletin of Ghana, 2:3,15-29.

Bussink, W.C. (1970), "A Complete Set of Consumption Coefficients forWest Pakistan," The Pakistan Development Review, 10:2, 193-231.

Chen, Lincoln C. (1975), "An Analysis of Per Capital Foodgrain Availability,Consumption and Requirements in Bangladesh," The Bangladesh DevelopmentStudies, IV.No.2.

De Janvry, et al. (1972), "Estimation of Demand Parameters under ConsumerBudgeting: An Application to Argentine," Amer.J. of Ag. Econ., 54:3,422-30.

Deaton, Angus (1974), "A Reconsideration of the Empirical Implications ofAdditive Preferences," Economic Journal, June, 338-48.

Dillion, J.L. and A.A. Powell (1965), "Un Modelo Econometrico de la Demandaal Detalle en Santiage de Chile," Cuadernos de Economia, Sept., 2:7.

Farrel, M.J. (1954), "Demand for Motor Cars in the United States,"J. Royal Stat. Soc. Pt.2,117A,171-201.

Ferber, Robert (1966), "Research on Household Behavior," in AEA, Survey onEconomic Theory, 3:12, 114-154.

-89-

Frisch, R. (1959), "A Complete Scheme for Computing All Direct and Cross

Demand Elasticities in a Model of Many Sectors," Econometrica, 27,177-96.

Goldberger, A.S. and T. Gamaletsos (1970), "A Cross-Country Comparisonof Consumer Expenditure Patterns," European Economic Review, 1:357-400.

Goreux, Louis M. (1972), "Price Stabilization Policies in World Markets forPrimary Commodities: An Application to Cocoa," mimeo. World Bank.

Haessel, W. (1976), "The Demand for Agricultural Commodities in Ghana: AnApplication of Nonlinear Two-Stage Least Square with Prior Information,"Amer. J. of Ag. Econ., 58:2,341-5.

Holmes, R.A. (1968), "Combining Cross-Section and Time-Series Informationon Demand Relationships for Substitute Goods," Amer. J. of Ag. Econ.

50:1, 56-65.

Houthakker, H.S. (1965), "New Evidence on Demand Elasticities," Econometrica,33:2,277-288.

IBRD (1977), Price Prospects for Major Primary Commodities, June 1977,the World Bank Gray Cover Report No. 814/77.

Killick, Tony (1966), "The Economics of Cocoa," Chap. 15 in A Study ofContemporary Ghana, Vol. I. The Economy of Ghana, W. Birmingham,et al. (eds), Northwestern University Press.

Klein, L.R. and H. Rubin (1947-8), "A Constant-Utility Index of the Costof Living," Rev. of Econ. Studies, 15:840-87.

Labys, Walter C. (1975), Quantitative Models of Commodity Markets,Cambridge, Mass.

Li, Shou-Hua (1969), "An Econometric Analysis of Demand for Wheat inTaiwan," Master thesis, National Taiwan University.

Librero, A.R. (1971), "The International Demand for Philippine CoconutProducts: An Aggregate Analysis," The Philippine Econ. J., 10:1,1-22.

Lluch, Constantino, et al. (1977), Patterns in Household Demand and Saving,published for the World Bank, Oxford University Press.

Mangahas, M. et al. (1966), "Price and Market Relationship for Rice and Cornin the Philippines," J. of Farm Economics, 98 (Aug., 1966), 685-703.

Mears, L.A. et al. (1974), Rice Economy of the Philippines, University ofPhilippines Press.

-90-

Mellor, John (1975), Agricultural Price Policy and Income Distribution inLow Income Nations, World Bank Staff Working Paper, No. 214.

Mundlak, Yair (1964), Long-Term Projections of Supply and Demand forAgricultural Products in Israel, I. General View and Summary, FalkProjects for Economic Research in Israel, the Hebrew University,Jerusalem.

Nasol, R.L. (1971), "Demand Analysis for Rice in the Philippines,"J. of Agricultural Economics and Development, 1:1, 1-13.

NCAER (1970), Projections of Demand and Supply of Agricultural Commodities,Nat'l Council of Applied Econ. Research, New Delhi.

Olayide, S.O. and S.A. Oni (1969), "Short-Run Demand for Beef in WesternNigeria," The Nigeria J. of Economic and Social Studies, 2:2, 165-72.

Oni, S.A. (1970), "The Demand for Nigeria Cocoa: A Statistical Analysisof Regional Markets and Projections," The Nigeria J. of Econ. andSoc. Studies, 12:1, 61-71.

Pandey, R.K. (1973), "The Analysis of Demand for Foodgrains," Indian J. ofAgricultural Economics, 28:2, 49-55.

Parks, R.W. (1971), "Maximum Likelihood Estimation of the Linear ExpenditureSystem," J.A.S.A. 66, 900-3.

Parks, R.W. and Barten, A.P. (1973), "A Cross Country Comparison of theEffects of Prices, Income and Population Composition on ConsumptionPatterns," Econ. J. 83:331,834-52.

Pollak, R.A. and T.J. Wales (1969), "Estimation of the Linear ExpenditureSystem," Econometrica, 37, 629-750.

Samuelson, P.A. (1947-8), "Some Implications of Linearity," Review ofEcon. Studies, 15:88-90.

Stone, R. (1954), "Linear Expenditure Systems and Demand Analysis: AnApplication to the Patterns of British Demand," Econ. J. 64:511-527.

Stone, R. (1954), The Measurement of Consumers' Expenditure and Behaviorin the United Kingdom, 1920-1938. Vol.I., Cambridge: CambridgeUniversity Press.

Tobin, James (1950), "A Statistical Demand Function for Food in the U.S.A.,"J. Royal Stat. Soc. Ser.A.,113, Pt II,113-41.

Vaurs, R. et al. (1972), An Agricultural Model from the Ivory CoastProgramming Study, IBRD Econ. Staff Working Paper, No. 125.

-95-

Food and Agriculture Organization of the,United Nations. Production Yearbook,various years.

Trade Yearbook, various years.

Gotsch, C. and G.Brown, "Prices, Taxes and Subsidies in Pakistan Agriculture,1960-1976", World Bank Staff Working Paper-No. 387, April 1980.

Grubel, H.G. and Johnson, H.G. "Effective Tariff Protection". GATT,Geneva, 1971.

Hansen, John R., Guide to Practical-Project Appraisal - Social Benefit CostAnalysis in Developing Countries, Project Formulation and EvaluationSeries No. 2, UNIDO, New York, 1978.

Hillman, J.S. "Nontariff Barriers: Major Problems in Agricultural Trade".Amer. J. Agr. Econ. 60 (1978): 491-501.

Johnson, H.G. "The Cost of Protection and the Scientific Tariff".J. Pol. Econ. 48 (1960): 327-345.

Johnson, P.R. "The Social Cost of the Tobacco Program". J. Farm Econ.47 (1965): 242-255.

Kim, H.M. "Agricultural Prices and Subsidies--Portugal Case Study".Economics and Policy Division, Agriculture and Rural DevelopmentDepartment, World Bank, 1977, unpublished.

Little, I.M.D. and J.A. Mirrlees, raisal and Planning forDeveloping Countries, Heinemann (London) 1974 and Basic Books (NewYork 1974).

McLoughlin, P. Associates Ltd. "A Study of Prices and Subsidies in Kenya'sAgricultural Sector". Economics and Policy Division, Agriculture andRural Development Department, World Bank, 1977, unpublished.

Reca, Lucio G., "Argentina: Country Case Study of Agricultural Prices,Taxes and Subsidies", World Bank Staff Working Paper No. 386, April 1980.

Saleh, A.A. "Disincentives to Agricultural Production in DevelopingCountries: A Policy Survey". Foreign Agriculture. USDA 1975.

Schultz, Theodore, W., (ed.) Distortions of Agricultural Incentives, IndianaUniversity Press, Bloomington and London, 1978.

Snape, R.N. "Sugar: Costs of Protection and Taxation". Economica (1969):(29-41).

-96-

ULG Consultants Limited, in association with Economic Consultants Limited,"Yugoslavia: Agricultural Prices and Subsidies Case Study".Economics and Policy Division, Agriculture and Rural DevelopmentDepartment, World Bank, 1977, unpublished.

UNIDO, Guidelines for Project Evaluation, Project Formulation and EvaluationSeries No. 2, Vienna, 1972.

Valdes, A. "Trade Policy and Its Effect on the External Agricultural Tradeof Chile 1945-1965". Amer. J. Agr. Econ. 55 (1973): 154-164.

Wallace, T.D. "Measures of Social Cases of Agricultural Programs".J. Farm Econ. 44 (1962): 580-597.

Wipf, L.J. "Tariffs, Nontariff Distortions, and Effective Protection inU.S. Agriculture". Amer. J. Econ. 53 (1971): 423-430.

Wong, C.M. "Model for Evaluating the Effects of Rice Exports on Trade andWelfare". Amer. J. Agr. Econ. 60 (1978): 65-73.