history, institutions, and economic performance: the ... media...tenure system under british rule....

History, Institutions, and Economic Performance: The Legacyof Colonial Land Tenure Systems in India

By ABHIJIT BANERJEE AND LAKSHMI IYER*

We analyze the colonial land revenue institutions set up by the British in India, andshow that differences in historical property rights institutions lead to sustaineddifferences in economic outcomes. Areas in which proprietary rights in land werehistorically given to landlords have significantly lower agricultural investments andproductivity in the post-independence period than areas in which these rights weregiven to the cultivators. These areas also have significantly lower investments inhealth and education. These differences are not driven by omitted variables orendogeneity problems; they probably arise because differences in historical insti-tutions lead to very different policy choices. (JEL O11, P16, P51)

There is renewed interest among economistsin the question of whether history, through itseffect on the pattern of institutional develop-ment, has a persistent effect on economic per-formance. In a recent series of papers, Rafael LaPorta et al. (1998, 1999, 2000) have argued thatthe historical fact of being colonized by theBritish, rather than any of the other colonialpowers, has a strong effect on the legal systemof the country and, through that, on economicperformance. The role of history in determiningthe shape of present-day institutions is also atthe heart of two recent sets of papers, one byDaron Acemoglu et al. (2001, 2002) and theother by Stanley Engerman and KennethSokoloff (1997, 2000, 2002). Acemoglu et al.show that mortality rates among early Europeansettlers is a strong predictor of whether thesecountries end up with what economists todaycall “good” institutions (which protect privateproperty rights) and whether their economiesare doing well today. Engerman and Sokoloffargue that the reason why Brazil is where it is

today, and the United States is where it is, has alot to do with the fact that in the early years afterEuropean conquest Brazil was deemed to besuitable for growing sugar and the United Stateswas not. Since sugar cultivation demanded theuse of slave labor, Brazil ended up with a muchlarger slave population, and this, they argue,meant that Brazilian society was much morehierarchical than American society, causing adivergence in the types of institutions thatevolved in these two countries and eventually adivergence in the rates of growth.

This paper is a part of the same broad re-search agenda. Where it differs is in focusing onone very specific historical institution—the sys-tem for collecting land revenue—in one specificcountry—India. We compare the present-dayeconomic performance of different districts ofIndia, which were placed under different landrevenue systems by British colonial rulers as aresult of certain historical accidents. We showthat districts in India where the collection ofland revenue from the cultivators was assignedto a class of landlords systematically underper-form the districts where this type of intermedi-ation was avoided, after controlling for a widerange of geographical differences. The differ-ences show up in agricultural investment andyields, in various measures of public investmentin education and health, as well as in health andeducational outcomes. For example, the averageyield of wheat is 23 percent higher and infantmortality is 40 percent lower in non-landlorddistricts. The non-landlord effect remains sig-

* Banerjee: Department of Economics, MassachusettsInstitute of Technology, 50 Memorial Drive, Cambridge,MA 02139 (e-mail: [email protected]); Iyer: Harvard Busi-ness School, Soldiers Field, Boston, MA 02163 (e-mail:[email protected]). We thank Daron Acemoglu, Sam Bowles,Esther Duflo, Maitreesh Ghatak, Karla Hoff, Kaivan Mun-shi, Raghuram Rajan, Andrei Shleifer, two anonymous ref-erees, and numerous seminar participants for helpfulcomments. We also thank Nabeela Alam and TheresaCheng for research assistance and Michael Kremer for helpin accessing historical land tenure data.

1190

nificant even when we restrict our data analysisto a set of 35 districts, chosen so that a landlorddistrict always borders a non-landlord district.Finally, in all the data we have from the earlierperiod, i.e., from the nineteenth and early twen-tieth centuries, there is no evidence of landlorddistricts being at a disadvantage.

An obvious advantage of focusing on one spe-cific institution in one particular country is that itmakes it easy to locate the source of the differ-ence, relative to the case where there is a complexof institutions that are all different. Another ad-vantage is that we have access to a very detailedhistory of how the institutional variation cameabout, which makes it easier to argue for exoge-neity of specific pieces of the variation. In partic-ular, we will argue, based on historical facts, thatareas where the land revenue collection was takenover by the British between 1820 and 1856 (butnot before or after) are much more likely to havea non-landlord system, for reasons that have noth-ing to do with factors that directly influence agri-cultural investment and yields. We will thereforeuse the fact of being conquered in this period as aninstrument for having a non-landlord system. Weallow for the possibility that areas that were con-quered in this period may have had a differentexperience simply because, for example, theywere conquered later than most other areas, byincluding controls for the length of British rule.The instrumental variable estimates confirm theOLS results.

A third advantage of this particular experi-ment is that the land revenue systems intro-duced by the British departed with the British:there are no direct taxes on agricultural incomesin independent India. Our results therefore tellus that the system for land revenue collectionestablished by the British 150 years ago or morecontinues to have an effect, long after it wasabolished. We therefore have a pure example ofinstitutional overhang, underscoring how hard itis to reform the institutional environment.1

The one disadvantage of a very specific ex-periment like ours is the suspicion that it reflectsthe peculiarity of the Indian experience. In otherwords, our results would be more interesting if

we could identify the reasons for this extremepersistence. While our data do not allow us toidentify exactly the channel through which thehistorical land revenue system continues tohave an effect, there are a number of clues.When the British left, areas where landlordscollected the revenue had an elite class that hadenjoyed a great deal of economic and politicalpower for over a century; there was no coun-terpart to this class in the non-landlord areas.This meant that these areas inherited a moreunequal land distribution at the time of indepen-dence, and a very specific set of social cleav-ages, absent elsewhere.

Our data suggest, however, that in the post-independence period there is substantial conver-gence in inequality between the landlord andnon-landlord areas, probably because stateswith landlord-dominated areas tend to enact agreater number of land reforms. This makes itunlikely that the persistence of the landlordeffect is mainly through its effect on the con-temporaneous land distribution.

On the other hand, it seems that, despite theabolition of the formal structure of landlordism,the class-based antagonism that it created withinthe communities in these areas persisted well intothe post-independence period. The conflictual en-vironment this created is likely to have limited thepossibility of collective action in these areas. Thiscollective action–based view is consistent withthe fact that the gap between the non-landlord andlandlord districts grows particularly fast in theperiod 1965–1980 when there is extensive publicinvestment in rural areas. We find that states witha higher proportion of landlord districts havemuch lower levels of public development expen-ditures and that a substantial part of the gap be-tween landlord and non-landlord districts inhealth, education, and agricultural technology in-vestments can be explained by this difference inpublic spending. This suggests that the key towhat happened may lie in the relative inability ofthe landlord districts to claim their fair share ofpublic investment.

The paper is structured as follows: Section Idescribes the historical background and the landtenure system under British rule. We discuss thereasons why the tenure system varies from dis-trict to district, and argue that the choice oftenure system can be reasonably regarded as asource of exogenous variation. Section IIoutlines different mechanisms through which

1 This distinguishes this work from the recent empiricalliterature on the effects of current land reform on currenteconomic outcomes (see Banerjee et al., 2002; TimothyBesley and Robin Burgess, 2000; Justin Y. Lin, 1992,among others).

1191VOL. 95 NO. 4 BANERJEE AND IYER: LEGACY OF COLONIAL LAND TENURE SYSTEMS IN INDIA

historical land tenure might affect long-termoutcomes. Sections III and IV describe our dataand empirical strategy. Our main empirical re-sults are described in Section V. Section VIconcludes by discussing potential mechanismsthat might explain the persistence of the effectof British land tenure systems.

I. Historical Background

A. British Political Control

The British Empire in India lasted for nearlytwo hundred years. The British first arrived astraders: the English East India Company re-ceived a permit in 1613 from the Mughal em-peror, Jahangir, to build a factory at Surat. Theirempire building began with their victories inthe battle of Plassey in 1757 and the battle ofBuxar in 1764, as a result of which they ob-tained political control of the modern statesof Bengal and Bihar (formerly Bengal Presi-dency). The British were formally grantedrevenue-collection rights in these areas in 1765.After 1818, the British were the major politicalpower in India and by 1860 a large part of theterritories of modern India, Pakistan, and Ban-gladesh were part of the British Empire. Therewere also a large number of princely states indifferent parts of the country, all of which wereunder British political control but had autonomyin administrative matters.

Different parts of the country came under Brit-ish rule in different periods. While the BengalPresidency came into British hands in 1765, therest of eastern India was conquered much later.Some parts of the modern state of Orissa wereconquered in 1803 and Assam was conqueredbetween 1824 and 1826. Meanwhile, in southIndia, the British obtained four districts (the“Northern Circars”) as a grant from the Mughalemperor in 1765. These and other areas conqueredbetween 1792 and 1801 came to form the MadrasPresidency. Parts of the western state of Gujaratwere conquered in 1803 and the rest, along withlarge parts of Bombay Presidency, were obtainedafter conquering the Marathas in 1817–1818.Some of these areas formed part of the CentralProvinces, to which other parts were added over along period until 1860. In the north, large parts ofthe North-West Provinces were obtained from theNawab of Oudh in 1801–1803, but Oudh itselfwas not annexed by the British until 1856. The

northwestern state of Punjab was annexed afterthe Sikh wars in 1846 and 1849. Table 1 in theWebAppendix(http://www.e-aer.org/data/sept05_app_banerjee.pdf) provides district-wide detailson the date and mode of acquisition by theBritish.

The rule of the East India Company came toan end after the Mutiny of 1857, when Indiantroops revolted against their British officers.The revolt was soon suppressed, but it forcedthe British government to bring India under itsdirect control. The British left India in 1947,when the Indian Empire was partitioned intoIndia and Pakistan.2 Large parts of former Ben-gal Presidency and Panjab Province are now inBangladesh and Pakistan, respectively.

B. Pre-British and British Systemsof Land Revenue

Land revenue, or land tax, was the majorsource of revenue for all governments of India,including the British. During the period ofMughal rule in the sixteenth and seventeenthcenturies, land revenue was collected by non-hereditary, transferable state officials (the mans-abdari system introduced by Emperor Akbar).After the decline of Mughal power in the earlyeighteenth century, these officials and othersgrabbed power where they could and became defacto hereditary landlords and petty chiefs intheir local areas. As a result, by the time Britishrule was firmly established in India (toward theend of the eighteenth century), it was very hardto tell what the “original land revenue systems”of India had been, and different British admin-istrators could come to very different conclu-sions about it.

Land revenue, or land tax, continued to be themajor source of government revenue duringBritish times as well. In 1841, it constituted 60percent of total British government revenue,although this proportion decreased over time asthe British developed additional tax resources.Not surprisingly, land revenue and its collectionwere the most important issues in policy debatesduring this period. We use the terms “land rev-enue systems” or “land tenure systems” to refer

2 Bangladesh, formerly East Pakistan, became an inde-pendent nation in 1975.

1192 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 2005

to the arrangements made by the British admin-istration to collect the land revenue from thecultivators of the land. These systems definedwho had the liability to pay the land tax to theBritish. Up to a first approximation, all cultiva-ble land in British India fell under one of threealternative systems: (a) a landlord-based system(also known as zamindari or malguzari), (b) anindividual cultivator-based system (raiyatwari),and (c) a village-based system (mahalwari). Ta-ble 1 gives the number of districts in eachcategory for the states in our data. The map inFigure 1 illustrates the geographic distributionof these areas.

In the landlord areas, the revenue liability fora village or a group of villages lay with a singlelandlord. The landlord was free to set the reve-nue terms for the peasants under his jurisdictionand to dispossess any peasants who did not paythe landlord what they owed him.3 Whateverremained after paying the British revenue de-mand was for the landlord to keep. These rev-enue-collecting rights could be bequeathed, aswell as bought and sold (Kumar, 1982). In thissense, the landlord effectively had propertyrights on the land. Landlord systems were es-

tablished mainly in Bengal, Bihar, Orissa, theCentral Provinces (modern Madhya Pradeshstate), and some parts of Madras Presidency(modern Tamil Nadu and Andhra Pradeshstates). In some of these areas, the British de-clared the landlords’ revenue commitments tothe government to be fixed in perpetuity (the“Permanent Settlement” of 1793). In other ar-eas, a “temporary” settlement was implementedwhereby the revenue was fixed for a certainnumber of years, after which it was subject torevision.

In most areas of Madras and Bombay Presi-dencies, and in Assam, the raiyatwari systemwas adopted under which the revenue settle-ment was made directly with the individualraiyat or cultivator. In these areas, an extensivecadastral survey of the land was done and adetailed record-of-rights was prepared, whichserved as the legal title to the land for thecultivator. Unlike the Permanent Settlement ar-eas, the revenue commitment was not fixed; itwas usually calculated as the money value of ashare of the estimated average annual output.This share typically varied from place to place,was different for different soil types, and wasadjusted periodically in response to changes inthe productivity of the land.

In the North-West Provinces and Panjab, the3 Some measures for protecting the rights of tenants and

subproprietors were introduced in later years.

TABLE 1—STATE-WISE DISTRIBUTION OF LANDLORD AND NON-LANDLORD DISTRICTS

State

Meannon-landlordproportion

Classification of revenue systems

Total districtsLandlord

basedIndividual

based

Village bodies

Landlord Non-landlord

Andhra Pradesh 0.66 2 8 0 0 10Bihar 0.00 12 0 0 0 12Gujarat 1.00 0 7 0 0 7Haryana 0.85 0 0 0 5 5Karnataka 1.00 0 15 0 0 15Madhya Pradesh 0.10 14 1 0 0 15Maharashtra 0.78 4 14 0 0 18Orissa 0.32 6 2 0 0 8Punjab 0.87 0 0 0 6 6Rajasthan 0.00 1 0 0 0 1Tamil Nadu 0.75 2 9 0 0 11Uttar Pradesh 0.42 0 0 12 35 47West Bengal 0.00 11 0 0 0 11

Total 0.51 52 56 12 46 166

Notes: This table lists only districts that used to be part of British India. Areas where the British did not set up the land revenuesystem are excluded. Districts of British India currently in Pakistan, Bangladesh, or Burma are excluded. The table alsoexcludes the states of Assam and Kerala, for which agricultural data are not available in the World Bank dataset. The tablelists 1960 districts, some of which were split into two or more districts over time. We use unsplit districts in all our analyses.


village-based (mahalwari) system was adoptedin which village bodies which jointly owned thevillage were responsible for the land revenue.Village bodies could be in charge of varyingareas, from part of a village to several villages.The composition of the village body also variedfrom place to place. In some areas it was asingle person or family that made up the villagebody and hence was very much like the Bengallandlord system (zamindari), while in other ar-eas the village body had a large number ofmembers with each person being responsible fora fixed share of the revenue. This share waseither determined by ancestry (the pattidari sys-tem), or based on actual possession of the land(the bhaiachara system), the latter being verymuch like the individual-based raiyatwari sys-

tem. The revenue rates in these areas were de-termined on fairly ad hoc grounds, based on adiverse set of factors, including: “an examina-tion of rents recorded in the jamabandis, therates which were actually paid by the variousclasses of tenants and the rates which wereconsidered fair on each class of soil. ... Theseestimates are based primarily on soils, and sec-ondly on consideration of the caste of the ten-ant, capabilities of irrigation, command ofmanure &c, all of which points received atten-tion” (F. W. Porter, 1878, p. 108).4

4 Except in the areas under the Permanent Settlement, theamount of revenue actually paid was often less than thestated revenue liability, due to remissions being granted in

FIGURE 1. MAP OF INDIA


http://www.atypon-link.com/action/showImage?doi=10.1257/0002828054825574&iName=master.img-000.jpg&w=299&h=348

C. Choice of Land Revenue System

Why did the British choose different systemsin different areas? It is broadly agreed that theirmajor motivation was to ensure a large andsteady source of revenue for the government,while maintaining a certain political equilib-rium. It is also clear, however, that they oftenfaced a lack of hard information and based theirdecision on a priori arguments. For instance, SirThomas Munro argued for the establishment ofan individual cultivator system in Madras on thegrounds that it would raise agricultural produc-tivity by improving incentives; that the cultiva-tors would be less subject to arbitraryexpropriation than under a landlord; that theywould have a measure of insurance (via gov-ernment revenue remissions in bad times); thatthe government would be assured of its revenue(since small peasants are less able to resist pay-ing their dues); and that this was the mode ofland tenure prevailing in South India from an-cient times. The Madras Board of Revenue, inits turn, used more or less the same arguments(in reverse, of course) for favoring landlords.Large landlords would have the capacity to in-vest more and therefore productivity would behigher; the peasants’ long-term relationshipwith the landlord would result in less expropri-ation than the short-term one with a governmentofficial; a big landlord would provide insurancefor small farmers; a steady revenue would beassured because the landlords would be wealthyand could make up an occasional shortfall fromtheir own resources; and this was the mode oftenure prevailing from ancient times (NilmaniMukherjee, 1962)! While the British often in-voked history to justify the choices they made,they frequently misread history. For example,one reason they favored landlords in Bengal isbecause they found landlords in Bengal whenthey arrived. As has been pointed out by anumber of scholars,5 however, these landlordswere really local chieftains and not the largefarmers that the British had thought them to be.

Decisions were therefore often taken on the

basis of some general principle, and the ideol-ogy of the individual decision maker and con-temporary economic doctrines played animportant role in combination with the exigen-cies of the moment. Table 2 of the Web Appen-dix provides details of how different landrevenue systems came to be established in dif-ferent provinces of British India. Here, we sum-marize the main channels of influence.

Influence of Individual Administrators.—Theideas and political influence of particular ad-ministrators sometimes determined land reve-nue systems in whole provinces. For instance, inthe Madras land tenure debate cited above, theBoard of Revenue initially prevailed over SirThomas Munro, and all the villages were putunder village-level landlords with renewableleases. Munro, traveled to London, however,and managed to convince the Court of Directorsof the East India Company of the merits of theindividual-based raiyatwari system; they thenordered the Madras Board of Revenue to imple-ment this policy all over the province after1820, on the expiration of the landlord leases.Similarly, the individual system was tried out inBombay Presidency quite early, mainly becausethe governor, Lord Elphinstone, was in favor ofit and had been a supporter of Munro during thedebate in Madras.

Another instance of individual influence oc-curred in the North-West Provinces. Landlordsystems with short-term leases were imple-mented there initially, and there was consider-able debate as to whether or not there should bea Permanent Settlement along the lines of thatprevailing in Bengal. In 1819, however, HoltMackenzie, the Secretary of the Board of Rev-enue, wrote a famous Minute, which claimedthat historically every village had had a propri-etary village body and felt that no settlementthat did not give proper recognition to suchcustomary rights should be declared in perpetu-ity. This became the basis for Regulation VII of1822, which laid the basis for village-level set-tlements (B. R. Misra, 1942). The previous ac-tions, however, could not always be undone andin several places the previously appointed largelandlords (talukdars) retained their positions.6

times of bad harvests and other hardships. Our focus here isnot on the actual revenue paid or the revenue rates, whichprevailed at various points of time, but on the allocation ofrevenue and control rights in land.

5 See Tirthankar Roy (2000) and Ratnalekha Ray (1979).

6 For instance, the Aligarh settlement officer writes, “Sofar indeed had the action of our first officials sanctioned the


Political Events.—The most notable exampleof this occurred in Oudh province. This regionwas annexed by the British in 1856 and mergedwith the North-West Provinces to form theUnited Provinces (state of Uttar Pradesh today).Since the North-West Provinces had a village-based revenue system, it was proposed to ex-tend the same to Oudh, and a cadastral surveythat would form the basis of this settlement wasunder way when the mutiny broke out in 1857.After it was successfully subdued, the Britishfelt that having the large landlords (talukdars)on their side would be politically advantageous.Thus, there was a reversal of policy and severallandlords whose land had been taken away un-der the village-based settlement had the landgiven back to them, and in 1859 they weredeclared to have a permanent, hereditary, andtransferable proprietary right. Districts that usedto be a part of Oudh thus came to have a largerarea under landlord control than the other dis-tricts of Uttar Pradesh.

Date of Conquest.—There are at least threereasons why areas that came under British rev-enue administration at later dates were in gen-eral more likely to have non-landlord systems.First, areas conquered later had some non-landlord precedents to follow and these made iteasier to make the case for the non-landlordsystem. For instance, Berar was put under anindividual-based system because neighboringBombay had been; and similarly Panjab adoptedthe village-based system already in place in theNorth-West Provinces. In fact, once Munro’svictory over the Board of Revenue in Madraswas sealed by a widespread conversion of land-lord areas into raiyatwari areas, and Holt Mack-enzie had succeeded in making the case forvillage bodies, there were to be no new landlordareas until the reversal in Oudh. Second,landlord-based systems required much less ad-ministrative machinery to be set up by the Brit-ish, and so areas conquered in the early periods ofBritish rule were likely to have landlord-basedsystems. Once a landlord-based system was es-tablished, however, it was costly to change the

system (this was most obviously true wherethere was a Permanent Settlement) and hencethe landlord system survived. Finally, the in-creasing popularity of dealing directly with thepeasant mirrored shifts in the views of econo-mists and others in Britain. In the 1790s, underthe shadow of the French Revolution across theChannel, the British elites were inclined to sidewith the landlords. In the 1820s, with peasantpower long defeated and half forgotten, theywere more inclined to be sympathetic to theutilitarians and others who were arguing fordealing directly with peasants.7,8

Presence of a Landlord Class before the Brit-ish Took Over.—This was probably one of thefactors leading to the landlord system beingfavored, at least in Bengal. As the historianTapan Ray Chaudhuri says, “... in terms ofrights and obligations, there was a clear line ofcontinuity in the zamindari system of Bengalbetween the pre- and the post-Permanent Set-tlement era” (Dharma Kumar, 1982). This wasnot, however, always the case. For instance, itwas decided to have a landlord-based system inthe Central Provinces, even though there was noexisting landlord class.9

D. Post-Independence Developmentsin Land Policy

Under the constitution of independent India,states were granted the power to enact landreforms. Several states passed legislation in theearly 1950s, formally abolishing landlords andother intermediaries between the governmentand the cultivator. Other laws have also beenpassed by different states at different times re-garding tenancy reform, ceiling on land hold-ings, and land consolidation measures. Besleyand Burgess (2000) provide a good review of

usurpations of the Talukdars, that among other cases theygranted to Raja Bhagwant Singh a lease for life of the wholeof the pargana Mursan for Rs.80,000 leaving the old com-munities entirely at his mercy ...” (W. H. Smith, 1882).

7 James Mill actually worked for the East IndiaCompany, and George Wingate, who helped set up theindividual-cultivator system in Bombay, was heavily in-fluenced by him.

8 For a discussion of the role of ideology and economicdoctrines in the formation of the land revenue systems, seeRanajit Guha (1963) and Eric Stokes (1959, 1978a).

9 B. H. Baden-Powell (1892) states: “In the CentralProvinces we find an almost wholly artificial tenure, createdby our revenue-system and by the policy of the Governmentof the day.”


these laws and their impact on state-level pov-erty rates.

II. Why Should the Historical LandSystem Matter?

Why would we expect productivity and in-vestment (including public investment in infra-structure) to differ between areas having agreater or lesser extent of landlord control?Why would these differences persist and not bewiped out as soon as the landlord class is for-mally abolished? In this section, we list somepotential answers to these questions, postponingto Section VI any discussion of the empiricalplausibility of these answers.

A. Differences in the Distribution of Wealth

Under landlord-based systems, the landlordswere given a more or less free hand to set theterms for the tenants10 and, as a result, theywere in a position to appropriate most of thegains in productivity. Moreover, landlord areaswere also the only areas subject to the Perma-nent Settlement of 1793 (which fixed the land-lord’s dues permanently in nominal terms), andeven where the settlement was not permanent,the political power of the landlord class made itless likely that their rates would be raised whentheir surplus grew. As the nineteenth centurywas a period of significant productivity growthand inflation, the landlord class grew rich overthis period and inequality went up. By contrast,in the individual cultivator areas, rents wereraised frequently by the British in an attempt toextract as much as possible from the tenant.There was, as a result, comparatively little dif-ferentiation within the rural population of theseareas until, in the latter years of the nineteenthcentury, the focus of the British moved awayfrom extracting as much as they could from thepeasants. At this point, there was indeed in-creasing differentiation within the peasant class,but overall one would expect less inequality inthe non-landlord areas.

In fact, this is what the limited historical datawe have suggest. The provinces with a highernon-landlord proportion have lower Gini mea-

sures of land inequality in 1885 (Figure2A). Further, the differences in inequality per-sist until the end of the colonial period. In 1948,the districts of Uttar Pradesh that had a higherlandlord proportion had a much higher propor-tion of land revenue being paid by very largelandlords and a correspondingly higher measureof inequality (Figure 2B).

The distribution of wealth is important forthree reasons: first, because it determines thesize of the group within the peasantry that hasenough land and other wealth to be able tomake the many somewhat lumpy and/or riskyinvestments necessary to raise productivity;11

second, because it affects the balance betweenthose who cultivate mainly their own land andthose who cultivate other people’s land (as iswell-known, cultivating other people’s landgenerates incentive problems, which reduces in-vestment and productivity); finally, because itmade it likely that the political interests of therural masses would diverge substantially fromthat of the elite. In particular, it made it very

10 Under the Haftam regulation of 1799 and the Panjamregulation of 1812.

11 See Banerjee and Andrew F. Newman (1993) andOded Galor and Joseph Zeira (1993) for theoretical mod-els of the link between income distribution and long-rundevelopment.

FIGURE 2. LAND TENURE AND LAND INEQUALITY


tempting for the peasants to support politicalprograms that advocate expropriating the assetsof the rich. To the extent that the differences inthe land distribution still persist, this would beone mechanism through which historical differ-ences in the land tenure system could continueto affect productivity today.

B. Differences in the Political Environment

The right to set the land revenue rates and topenalize those who did not pay gave the land-lords a substantial degree of political powerover the rest of the population in their domain.One possible consequence of this may be thatpeasant property was relatively insecure in thelandlord areas. Investments that made the landmore productive were discouraged because ofthe risk of expropriation by the landlord. Incontrast, in the raiyatwari areas, the proprietaryrights of peasants were based on an explicit,typically written, contract with the colonialstate, which the colonial state was broadly com-mitted to honor. This may have resulted inbetter incentives for the peasants in the non-landlord areas in the colonial period.

The exercise of this type of more or lessarbitrary power by the landlord over the prop-erty and not, infrequently, the body of the peas-ant, created a political ethos of class-basedresentment in these areas, which persisted wellinto the post-independence period.12 Those fa-miliar with post-independence India will recog-nize, for example, that the areas most associatedwith Maoist peasant uprisings (known as“Naxalite” movements)—clearly the most ex-treme form of the politics of class conflict inIndia—are West Bengal, Bihar, and theSrikakulam district of Andhra Pradesh, all land-lord areas. Paul R. Brass (1994, pp. 326–27)argues explicitly that these peasant movementshad their roots in the history of exploitation andoppression of peasants by landlords. Moreover,these class-based conflicts go back to the colo-nial period. Kathleen Gough (1974) studies 77peasant struggles from the end of the Mughalera until today and suggests that at least a thirdof these originated in Bengal, the oldest and

best established of the landlord areas. Along thesame lines, Partha Chatterjee (1984) has argued,based on the pattern of voting on the TenancyAct Amendment in the Bengal LegislativeCouncil in 1928, that the representatives of thepeasants voted largely in a block against thelandlords, and vice versa.

Given this history, it is no surprise that theelites and the masses in these areas rarely sharedthe trust that is essential for being able to acttogether in the collective interest.13 It is quiteplausible that, in the post-independence period,the political energies of the masses were di-rected more toward expropriating from the rich(via land reforms, for example) than towardtrying to get more public goods (schools, tapwater, electricity) from the state, while the po-litical energies of the rich were aimed at tryingto ensure that the poor did not get their way.14

Moreover, it was not uncommon for the ruralelites in the landlord areas to be quite disasso-ciated from the actual business of agriculture,since they typically were more likely to be rentcollectors than farmers, and even the rent col-lection rights were often leased out. This wouldtend to weaken the political pressure on the stateto deliver public goods that were important tofarmers. Moreover, they were often physicallyabsent, preferring to live in the city and simplycollect their rents, and as a result had only ratherlimited stakes in improving the living condi-tions in rural areas.

C. Differences in the Relationship with theColonial State

Since it was easier for the colonial govern-ment to raise rents in non-landlord areas, itmeant that the state could capture some of theproductivity gains from these areas, and hencehad more reason to invest in irrigation, railways,schools, and other infrastructure in these areasduring the colonial period.15 In this context, we

12 See Sugato Bose (1993) for an account of the rise ofclass-based agrarian politics in colonial Bengal (a landlordarea) and its subsequent influence on the politics of inde-pendent West Bengal.

13 See Alberto Alesina and Dani Rodrik (1994) andTorsten Persson and Guido Tabellini (1994) for modelswhere collective action fails in the presence of groups withmisaligned interests.

14 For instance, the rich could undercut democratic pro-cesses and resist public policies that would empower thepoor, very much along the lines taken by the Latin Ameri-can elites (see Engerman and Sokoloff, 2002).

15 Amiya K. Bagchi (1976) also makes this point.


should note that almost all canals constructed bythe British were in non-landlord areas. If, in-deed, these areas had better public goods whenthe British left, it is plausible that they couldcontinue to have some advantage even now.

III. Data

We use a combination of historical and recentdata for our analysis. All data are at the districtlevel, a district in India being an administrativeunit within a state. In 1991, India had 415districts in 17 major states, a district on averagehaving an area of 7,500 square kilometers and apopulation of 1.5 million.

We chose to use district-level rather thanstate-level data for three major reasons. First,modern Indian state boundaries are completelydifferent from older British province boundariesdue to the linguistic reorganization of states in1956. Although district boundaries have alsochanged a little over time, it is still possibleunambiguously to match current districts toolder districts—the main source of change isthat some of the older districts have been splitinto two or more districts over time. Second,because of the integration of several princelystates in 1947, nearly all the states in our dataare composed of both British-ruled districts anddistricts that were ruled by Indian kings in thecolonial period. Since we have historical dataon land tenure only for British districts, it ishard to compute a good state-level measure ofhistorical institutions. Third, using district-leveldata gives us a larger sample size. The draw-back is that we are limited in the kind of datathat we can get. For instance, we do not havemeasures of GDP or average income per capitaat the district level. We will thus be using othercorrelates or proxies of economic prosperity forwhich we have data at the district level: agri-cultural investment outcomes (the proportion ofirrigated gross cropped area, quantity of fertil-izer used per hectare of gross cropped area, andthe proportion of area sown with high-yieldingvarieties (HYV) of rice, wheat, and other cere-als); agricultural productivity (crop yields); andthe stock of health and education infrastructure(schools and health centers).

The district-level data on agricultural invest-ments and productivity come from the IndiaAgriculture and Climate Data Set assembled bythe World Bank and cover the period 1956–

1987. This dataset has information on 271 dis-tricts in 13 major states.16 All data are at the1961 district level, aggregating over subsequentsplits in districts. We also have data for healthand education infrastructure from the 1981 Cen-sus. We matched each modern district to anolder British district using old and new maps,and retained only the districts where the landtenure system was established by the British,because we do not have detailed information onthe land systems in districts that were undernative princes or tribal chiefs.17 For each districtof British India,18 we then proceed to computea measure of non-landlord control in the colo-nial period as follows: for many areas (the statesof Andhra Pradesh, Madhya Pradesh, Panjab,Tamil Nadu, and Uttar Pradesh), we havedistrict-level information on the proportion ofvillages, estates, or land area, not under therevenue liability of landlords; for other areaswhere we do not have the exact proportion(Bihar, Karnataka, Maharashtra, Orissa, WestBengal), we assign the non-landlord measure asbeing either zero or one, depending on the dom-inant land revenue system. In all cases, themeasure of non-landlord control is computedbased on data from the 1870s or 1880s. Thedetails of the data sources and the constructionof this variable are in Table 3 of the WebAppendix.

IV. Empirical Approach

We will compare agricultural investmentsand productivity between landlord and non-landlord areas by running regressions of theform

16 The states included in the dataset are Andhra Pradesh,Bihar, Gujarat, Haryana, Karnataka, Madhya Pradesh, Ma-harashtra, Orissa, Punjab, Rajasthan, Tamil Nadu, UttarPradesh, and West Bengal. Assam, Himachal Pradesh,Jammu and Kashmir, and Kerala are the large states notcovered.

17 This usually corresponds to the areas under directBritish administrative control, with one exception. In theprincely state of Mysore (part of modern Karnataka state),the British took over the administration in 1831 and ruledfor 50 years, before reinstating the royal family in 1881.During this time, the British instituted an individual-basedland revenue system, which the ruler was obliged to con-tinue after his reinstatement.

18 We dropped districts currently in Pakistan andBangladesh.


(1) yit � constant � �t � �NLi � Xit� � �it

where yit is our outcome variable of interest(investment, productivity, etc.) in district i andyear t, �t is a year-fixed effect, NLi is thehistorical measure of the non-landlord controlin district i, and Xit are other control variables.Our coefficient of interest is �, which capturesthe average difference between a non-landlorddistrict and a landlord district in the post-independence period.

In all our regressions, we control for suchgeographic variables as latitude, altitude, soiltype, mean annual rainfall, and a dummy forwhether the district is on the coast or not. Inaddition, we also control for the length of timeunder British rule (or, equivalently, the date ofBritish conquest), which may have independenteffects, because early British rule was particu-larly rapacious or because the best (or the worst)districts fell to the British first. Note that we donot include district fixed effects in this regres-sion, since NLi is fixed for district i over time (itis the historical land arrangement). We do ad-just our standard errors for within-district cor-relation, however, since our data consist ofrepeated observations for each district overtime. We also do not use state fixed effects inour base specification because the within-statevariation in non-landlord proportion is much lessthan the cross-state variation.19 More importantly,the modern states were formed at a later date thanour non-landlord proportion and we would like tosee how far historical factors can account for thewidely varied performance of Indian states in thepost-independence period.

As mentioned in the introduction, we will tryto deal with concerns about exogeneity, first bylooking only at the difference between neigh-boring districts, and second by adopting aninstrumental variables approach. After estab-lishing the robustness of the differences in in-vestment and productivity between landlord andnon-landlord areas, we estimate some additionalspecifications. First we reestimate the yieldequations after controlling for various measuresof investment in agriculture (fertilizer use, irri-gation, etc.) to check whether there is a non-landlord effect over and above the effect on

investment. Then we allow the non-landlordcoefficient to vary over time to see whether wecan demonstrate how the gap between landlordand non-landlord areas has evolved over time.

V. The Impact on Agricultural Outcomes

A. Differences in Geography andOther Differences

There are significant geographical differencesbetween landlord areas and non-landlord areas(Table 2). Landlord areas have somewhat loweraltitudes, higher rainfall, and fewer areas withblack soil as compared to non-landlord areas. Inparticular, we note that landlord areas have agreater depth of topsoil, which together with thegreater rainfall and lower altitudes seems to indi-cate that these areas might be inherently morefertile and productive. Landlord areas have aslightly higher total population and a significantlyhigher population density than non-landlord areas.This is consistent with the fact they seem to bemore fertile areas. They have a greater proportionof minorities, such as castes that were discrimi-nated against historically and are formally listed as“Scheduled Castes” in the Indian Constitution,and more people living in rural areas. Further,landlord areas have a greater proportion of theworkforce engaged in farming, and devote morearea to food crops like rice and wheat and less tocash crops like cotton, oilseeds, tobacco, andsugarcane. This could be due simply to differentclimatic conditions or could reflect an endoge-nous shift toward commercial agriculture innon-landlord areas.

We have very limited historical data onyields. Looking at data for rice yields in tendistricts of Madras Presidency, and rice andwheat yields for 17 districts of Uttar Pradeshduring the colonial period, we see in Figure3 that yields were in fact lower in non-landlordareas during this period.20 Given the size of thesample, we cannot hope to control for geo-graphical differences between the districts.These yield differences may therefore reflectdifferences in geography. The only point we are

19 In our later regressions with state fixed effects, we arein effect dropping the states of Bihar, Gujarat, Karnataka,Rajasthan, and West Bengal.

20 The yield data for Uttar Pradesh come from the samesettlement reports of the 1870s and 1880s that we use tocalculate our non-landlord proportion. Very few of thereports contain data on yields, resulting in a very smallsample. We also have data for ten Tamil Nadu districts fromHaruka Yanagisawa (1996).


making here is that the landlord districts did notstart with a disadvantage.

B. Differences in Agricultural Investmentsand Productivity

We mainly investigate investment and pro-ductivity differences in the 1956 –1985 pe-riod. Table 3 documents large and significantdifferences in measures of agricultural invest-ments and productivity between landlord andnon-landlord areas in the post-independenceperiod. Each entry in this table represents theregression coefficient from a regression of thedependent variable on the non-landlord pro-

portion, controlling for year fixed effects,geographical variables (latitude, altitude,mean annual rainfall, and soil types), lengthof British rule, and within-district clusteringof errors. We show the detailed regressionspecification for one of the dependent vari-ables (log agricultural yield) in Table 4 in theWeb Appendix, listing the coefficients on allour control variables. Our base specificationin column (1) shows that non-landlord dis-tricts have a 24-percent-higher proportion ofirrigated area and 43-percent-higher levels offertilizer use. They have a 27-percent-higherproportion of rice area and 18 percent morewheat area under high-yielding varieties.

TABLE 2—DIFFERENCES IN GEOGRAPHY AND DEMOGRAPHICS

MeanStandarddeviation Differencea

Standard errorof difference

GeographyLatitude 22.19 5.60 �4.35*** (0.961)Altitude 366.41 148.14 93.64*** (25.98)

—Mean annual rainfall (mm) 1263.09 471.64 373.99*** (80.83)Coastal dummy 0.1497 0.3579 0.084 (0.065)Top 2 soil types Black soil 0.2096 0.4082 0.244*** (0.072)

Alluvial soil 0.1677 0.3747 �0.135** (0.067)Red soil 0.5689 0.4967 0.075 (0.090)

Top-soil depth �25 cm 0.0181 0.1336 0.016 (0.024)25–50 cm 0.1145 0.3193 �0.076 (0.058)50–100 cm 0.2289 0.4214 0.193 (0.075)100–300 cm 0.0904 0.2876 0.135*** (0.051)�300 cm 0.5482 0.4991 �0.268*** (0.088)

Area share of various crops: 1956–1987Area share of rice 0.366 0.298 �0.194*** (0.054)Area share of wheat 0.149 0.157 �0.058** (0.026)Area share of other cereals 0.205 0.172 0.128*** (0.031)Area share of oilseeds 0.067 0.088 0.065*** (0.013)Area share of cotton 0.041 0.096 0.066*** (0.018)Area share of tobacco 0.003 0.015 0.005** (0.002)Area share of sugarcane 0.031 0.053 0.005 (0.008)Cash crops-to-cereals ratio 0.149 0.257 0.152*** (0.048)

Demographics: 1961, 1971, 1981, 1991Log (Population) 14.26 0.634 �0.088 (0.109)Population density 36.44 85.92 �11.22** (4.02)Proportion of scheduled castes 0.1598 0.0733 �0.034** (0.014)Proportion of scheduled tribes 0.0980 0.1630 �0.010 (0.031)Proportion rural 0.8102 0.1237 �0.066*** (0.023)Proportion of working

population in farming0.7119 0.1352 �0.050* (0.027)

Notes: Standard errors in parentheses, corrected for district-level clustering. * Significant at 10-percent level; ** significantat 5-percent level; *** significant at 1-percent level. For the area under different crops and demographics, the difference iscalculated after controlling for year fixed effects.

a Difference represents the average difference between non-landlord and landlord districts, computed as the regressioncoefficient on the non-landlord proportion.


Overall agricultural yields are 16 percenthigher, rice yields are 17 percent higher, andwheat yields are 23 percent higher. Further,column 2 shows that these differences areslightly bigger if we exclude the states ofWest Bengal and Bihar, the two states thathave the highest proportion of landlord dis-tricts and the first to be conquered by theBritish. (We wanted to be sure that somethingidiosyncratic about these states was not driv-ing our results.)

It is worth noting that these differences aredriven neither by substitution away from agri-culture in landlord districts nor by a greater shifttoward crops other than rice or wheat. As wesee in Table 2, landlord areas have a higherproportion of their working population engagedin farming, and they also devote a lower pro-portion of area to growing cash crops.

C. Results Using Binary Measures ofNon-Landlord Control

Our results are robust to using a binary land-lord/non-landlord classification rather than thecontinuous measure. We construct this classifi-cation as follows: a district is classified as“landlord” if it was under a landlord-based sys-tem, if it was under a landlord-based system andonly partly converted to a different system (sev-eral districts of Madras), or if it was in Oudh,which we have argued had a higher proportionof landlords due to the reversal of policy after1856. All other individual-based or village-based systems are classified as “non-landlord.”Column 3 shows that our results are relativelyrobust to using a binary classification. A fewcoefficients are no longer significant here, prob-ably because we are deliberately mismeasuringour regressor—some of the “non-landlord” dis-tricts in our binary classification neverthelesshave large areas under landlords.21 We alsocompute results using a more restricted sample:since we might not be fully sure of the classi-fication of village-based districts, we excludethem and do a comparison of only landlorddistricts with individual-based districts. Someof the coefficients in this specification are largerthan our base specification (column 4). This isprobably because when we leave out the vil-lage-based districts, we are comparing almostwholly landlord areas with the other extreme,the individual-cultivator areas.

D. Results Using Neighboring Districts

Obviously, our interpretation of these re-sults has to be tempered by the possibility thatthe non-landlord gap might reflect omittedvariables. One strategy to control for possibleomitted variables is to consider an extremelyrestricted sample: we consider only those dis-tricts that happen to be geographical neigh-bors (i.e., share a common border), but that

21 In this classification, the “landlord” districts have atmost 40 percent of land under non-landlord control, whilesome of our so-called “non-landlord” districts in fact haveless than 20 percent of their land under non-landlord con-trol. We have also tried an alternative specification wherethe binary variable takes the value one if the non-landlordproportion is greater than 0.5, and zero otherwise. Ourresults are robust to this specification as well (results notshown).

FIGURE 3. AGRICULTURAL YIELDS IN COLONIAL PERIOD


happened to have different historical land sys-tems. (These districts and the historical rea-sons for their land system differences arelisted in Table 5 of the Web Appendix.) Weexpect that there would be fewer differencesin omitted variables, if any, in this sample ofgeographic neighbors than in our overall sam-ple, and we verify that there are no significantdifferences in our observed geographic anddemographic variables between these districts(results available upon request).

Even when we restrict our sample to thissmall set of 35 geographically neighboringdistricts, we still see large and significantdifferences between landlord and non-land-lord districts in agricultural investments andoutcomes (Table 4, panel A, column 1). Inparticular, total yields are 15 percent higher

and wheat yields 25 percent higher in non-landlord areas than in landlord areas. Theseestimates are very close to the estimates inour base specification. The differences in fer-tilizer use and HYV adoption for wheat arealso fairly close to the magnitudes obtained inour base specification. These results serve toconfirm that our original results were notcaused primarily by some unobserved districtcharacteristics.

E. Results Using Instrumental Variables

As discussed above, our results might also bebiased if the British decision regarding whichland tenure system to adopt depended on othercharacteristics of the area in systematic ways.We would like to highlight a few facts in this

TABLE 3—DIFFERENCES IN AGRICULTURAL INVESTMENTS AND YIELDS

(Mean non-landlord proportion � 0.5051 (s.d. � 0.4274))

Dependent variable

Mean ofdependentvariable

Coefficient on non-landlordproportion

Coefficient on non-landlorddummy

OLSFull sample

(1)

OLSExcluding Bengal

and Bihar(2)

OLSFull sample

(3)

OLSExcluding

village-baseddistricts (4)

Agricultural investmentsProportion of gross cropped area irrigated 0.276 0.065* 0.066* 0.077*** 0.005

(0.034) (0.035) (0.027) (0.032)Fertilizer use (kg/ha) 24.64 10.708*** 10.992*** 9.988*** 10.695***

(3.345) (3.406) (2.301) (3.040)Proportion of rice area under HYV 0.298 0.079* 0.094** 0.016 0.074*

(0.044) (0.043) (0.032) (0.038)Proportion of wheat area under HYV 0.518 0.092** 0.119*** 0.031 0.107**

(0.046) (0.045) (0.036) (0.052)Proportion of other cereals area under HYV 0.196 0.057* 0.084*** �0.035 0.109***

(0.031) (0.024) (0.025) (0.041)Agricultural productivitylog (yield of 15 major crops) 0.157** 0.152** 0.173*** 0.089

(0.071) (0.074) (0.053) (0.085)log (rice yield) 0.171** 0.195** 0.099 0.173**

(0.081) (0.081) (0.062) (0.079)log (wheat yield) 0.229*** 0.228*** 0.188*** 0.143

(0.067) (0.070) (0.054) (0.098)No. of districts 166 143 166 109Year fixed effects YES YES YES YESGeographic controls YES YES YES YESDate of British land revenue control YES YES YES YES

Notes: Standard errors in parentheses, corrected for district-level clustering. * Significant at 10-percent level; ** significant at5-percent level; *** significant at 1-percent level. Each cell represents the coefficient from a regression of the dependent variableon the measure of non-landlord control. Data are from 1956 to 1987. Data for area under high-yielding varieties (HYV) is after1965. Geographic controls are altitude, latitude, mean annual rainfall, and dummies for soil type and coastal regions. Thenon-landlord dummy is assigned as follows: the dummy equals one for all individual-based districts and all village-based districtsexcept those in Oudh. For landlord-based districts and the village-based districts of Oudh, the dummy is zero.


regard. First, we do not expect the choice ofland tenure system to be very highly correlatedwith local district characteristics, since thechoice of land tenure system was made for largecontiguous areas at the same time and was oftenbased on very little information regarding localconditions. Second, as explained in Section I C,places that were conquered earlier tended tohave landlord-based systems. If British an-nexation policy was selectively directed toward

the more productive places,22 then landlord-controlled areas are likely to be inherently moreproductive. Third, zamindari areas were usuallyhighly fertile areas which created enough rentto support a landlord-tenant-laborer hierarchy(Roy, 2000). In some areas, where landlord

22 See Iyer (2005) for some evidence in support of thishypothesis.

TABLE 4—ROBUSTNESS OF OLS RESULTS

Panel A: Robustness checks

Dependent variable

Coefficient on non-landlord proportion

OLS Neighbors only (1) IV Full sample (2)

Agricultural investmentsProportion of gross cropped area irrigated 0.101** 0.216

(0.041) (0.137)Fertilizer use (kg/ha) 10.589** 26.198**

(4.979) (13.244)Proportion of rice area under HYV �0.015 0.411**

(0.083) (0.163)Proportion of wheat area under HYV 0.078** 0.584***

(0.034) (0.163)Proportion of other cereals area under HYV �0.025 0.526***

(0.024) (0.129)Agricultural productivitylog (yield of 15 major crops) 0.145** 0.409

(0.061) (0.261)log (rice yield) 0.126 0.554*

(0.098) (0.285)log (wheat yield) 0.253*** 0.706***

(0.084) (0.214)No. of districts 35 166Year fixed effects YES YESGeographic controls YES YESDate of British land revenue control YES YES

Panel B: First-stage regressions for IVDependent variable: Non-landlord proportion

Coefficient on (1) (2) (3)

Instrument (�1 if date of British revenue control is between 1820 and 1856) 0.331*** 0.430*** 0.419***(0.086) (0.092) (0.087)

R-squared 0.40 0.43 0.63No. of observations 166 166 166Geographic controls YES YES YESDate of British land revenue control YES YES YESDate of British land revenue control squared NO YES NOState fixed effects NO NO YES

Notes: Standard errors in parentheses, corrected for district-level clustering. * Significant at 10-percent level; ** significantat 5-percent level; *** significant at 1-percent level. Each cell in Panel A represents the coefficient from a regression of thedependent variable on the non-landlord proportion. Data are from 1956–1987. Data for area under high-yielding varieties(HYV) is after 1965. Geographic controls are altitude, latitude, mean annual rainfall, and dummies for soil type and coastalregions. Instrument is a dummy that equals one if the date of British revenue control is after 1820 and before 1856.


defaults were excessive, these were laterchanged to different forms of settlement. There-fore, areas that ended up with non-landlord sys-tems are more likely to be inherently lessproductive, or at least were less productive incolonial times. Another way to deal with thispotential problem of omitted variables is to usean instrumental variables strategy. This has theadditional advantage of helping us deal with theproblem of measurement error in our non-land-lord proportion variable, caused by districtboundary changes and the fact that the historicalrecord tends to be impressionistic (in any case,reflects the impression of one observer at onepoint of time).

Our instrumental variables strategy is basedon the observation, mentioned in Section I, thatareas that came under British revenue adminis-tration after 1820 have predominantly non-land-lord systems, except for the policy reversalwhich occurred in Oudh (taken over in 1856)after the revolt of 1857. We believe that thesource of this variation is in part due to thesuccess of Munro and Mackenzie in establish-ing non-landlord systems in Madras and theNorth-West Provinces (starting around 1820),which created the all-important precedents thatwere followed in the districts conquered after1820, as well as a broader shift in ideology inEngland. Therefore, the fact that areas con-quered between 1820 and 1856 got non-land-lord systems does not depend on thecharacteristics of the district, and a dummy forthe date of conquest being between 1820 and1856 is a valid instrument for the non-landlordproportion, especially after we control for thedate of British conquest to take into account anydirect effects of a longer period of British rule.23

Figure 4 demonstrates the basis for our in-strumental variable strategy. In this figure, weplot the kernel regression of the non-landlordproportion and the mean log agricultural yieldagainst the date of conquest. It is clear that there

is a good fit in the shape of the two graphs andthat both curves are highly nonlinear. There-fore, the co-movement in the two graphs is notdriven by the fact that both are trending up ordown, making it less likely that the relationbetween the two reflects the direct effect of thedate of conquest. The figure also demonstratesthat the non-landlord proportion is significantlyhigher for areas conquered between 1820 and1856 compared to areas conquered earlier orlater. This is exactly what we would have ex-pected given the discussion above.24 Panel B inTable 4 shows the first-stage coefficients of ourIV strategy; we should note that the first-stagerelationship remains significant even when weinclude a quadratic control for the length ofBritish rule, as well as when we include statefixed effects.

Our IV results confirm that non-landlord sys-tems indeed have a large and significant impacton current outcomes (Table 4, panel A, column2). In fact, all the IV coefficients are larger thantheir OLS counterparts, although the differencebetween the two estimates is not statisticallysignificant.25 The standard errors for the IVestimates are also larger than the OLS standarderrors, but the non-landlord effect remains sta-tistically significant in the case of HYV adop-tion, as well as in fertilizer usage and wheatyields. Rice yields are significantly greater atthe 10-percent level. Specifications involving aquadratic control for the length of British ruletypically give coefficients that are smaller inmagnitude, but generally of the same level ofsignificance (results not shown).

The fact that the IV results are larger thanthe OLS results suggests that the OLS resultsare biased downward. This is the directionof bias we would have expected, given ourdiscussion above, especially the fact thatlandlord areas, which were not productiveenough to sustain a landlord class, tended tobecome non-landlord. It is also the directionof bias suggested by the presence of classicalmeasurement error. Since our non-landlordvariable is limited to being between 0 and23 By “date of conquest,” we mean the date when the

district came under British land revenue administration. Thetwo dates are usually the same, with two exceptions. Thefirst is the kingdom of Mysore, which was under Britishadministration for the period when the land revenue systemswere put in place, but was never part of the British Empire.The second is the kingdom of Nagpur, which was formallyannexed in 1854, but had been under British revenue controlin 1818.

24 The other “hump” (or mode) on the left is mainly dueto the districts of Madras Presidency, which were conqueredfairly early, but which switched over to a non-landlordsystem after 1820.

25 A Hausman test does not reject the null hypothesis thatthe OLS and IV coefficients are equal.


1, however, we have nonclassical measure-ment error. Even then, for the special case ofa binary regressor and no covariates, (non-classical) measurement error will still biasthe OLS coefficient downward, but will alsobias the IV coefficient upward (Kane et al.,1999). For this special case, we verify thatmeasurement error is not the only source ofthe difference between the OLS and IVestimates.26

Our IV results, together with the results onneighboring districts and the historical data,lead us to conclude that our OLS results are notbiased upward due to omitted district character-istics. Because of the possibility of upward biasin the IV estimates, however, we will continueto treat the OLS results as benchmark estimatesof the difference between landlord and non-landlord districts.

F. Does Land Tenure Have an IndependentEffect on Productivity?

We have established large and robust differ-ences between landlord and non-landlord dis-tricts in terms of agricultural investments andproductivity, with the non-landlord districtsshowing better performance in all of these mea-sures. In Table 5, we argue that the differencesin productivity are due largely to differences ininvestments. We do this by regressing pro-ductivity measures on the proportion of non-landlord control, as well as on the measures ofinvestment. All the measures of investment (ir-rigation, fertilizer use, and adoption of HYV)are positive and strongly significant, as wewould expect. The addition of these measuresreduces the coefficient on the non-landlord pro-portion by 78 percent for total yields, 59 percentfor rice yields, and 52 percent for wheat yields.The non-landlord variable is also no longer sta-tistically significant.

G. When Do the Differences Arise?

As shown before, non-landlord districtswere not more productive than landlord-baseddistricts in the colonial period. Figure 5 indi-cates that the differences in investments (ir-rigation, fertilizer) and yields widen in themid-1960s. Table 6 (panel A) formally estab-

26 We run the regressions with the binary regressor (de-fined in Section V C) and no covariates. If there were onlymeasurement error, the OLS would be biased downward,the IV would be biased upward but have the same sign asthe OLS coefficient, and the ratio of the two would be thesame for all the outcome variables. We find that, of the eightoutcome variables, the IV coefficient is larger than the OLSfor five, the IV is smaller in magnitude than the OLS forone, and for the remaining two outcomes, the OLS coeffi-cient is negative while the IV is positive. This suggests thatmeasurement error is not the only problem.

FIGURE 4. INSTRUMENTAL VARIABLES STRATEGY


lishes that the gap between landlord and non-landlord areas is larger after 1965 than in the1956 –1965 period. We also have data on riceyields for a limited sample of ten districts ofTamil Nadu from the colonial period onward.Figure 5D indicates that the non-landlord ar-eas overtake the landlord areas during themid-1960s. Table 6 (panel B) also checks thisformally by computing the gap in the pre-1965 and post-1965 period.

VI. Why do the Landlord Districts Fall Behind?

The period after 1965 saw the state in Indiabecoming much more active in rural areas,through the Intensive Rural DevelopmentPrograms, the efforts to disseminate newhigh-yielding varieties of crops (resulting inthe “Green Revolution”), and the building ofpublic infrastructure (including fertilizer de-livery systems) in rural areas under the 1971

TABLE 5—ARE YIELDS EXPLAINED BY INVESTMENTS?

Dependent variables

Log total yieldOLS (1)

Log rice yieldOLS (2)

Log wheat yieldOLS (3)

Proportion non-landlord 0.035 0.070 0.109(0.053) (0.063) (0.063)

Proportion of gross cropped areairrigated

0.693** 0.439** 0.435**(0.112) (0.096) (0.117)

Fertilizer use (kg/ha) 0.007** 0.004** 0.001(0.001) (0.001) (0.001)

Percent area under HYV 4.274** 0.580** 0.618**(1.122) (0.063) (0.070)

Adjusted R-squared 0.60 0.52 0.56No. of districts 166 166 166Year fixed effects YES YES YESGeographic controls YES YES YESDate of British land revenue

controlYES YES YES

Notes: Standard errors in parentheses, corrected for district-level clustering. * Significant at10-percent level; ** significant at 5-percent level; *** significant at 1-percent level. Data are from1956–1987. Data for area under high-yielding varieties (HYV) is after 1965. Geographic controlsare altitude, latitude, mean annual rainfall, and dummies for soil type and coastal regions.

A. Proportion of irrigated areayear

Landlord districts Non−landlord districts

1960 1965 1970 1975 1980

0.176707

0.466565

B. Fertilizer usageyear


1960 1965 1970 1975 1980 1985

0.346324

85.713

C. Mean log yieldyear


1960 1965 1970 1975 1980 1985

−0.488489

0.418232

D. Log rice yield: Tamil Nadu districtsyear


1901 1911 1917 1927 1937 1947 1957 1967 1977

−0.148565

0.773232

FIGURE 5. INVESTMENT AND PRODUCTIVITY TIME SERIES


Garibi Hatao (poverty alleviation) program.As we have seen, the landlord areas wereslower in the adoption of high-yielding vari-eties. They also seem to have benefited lessfrom the growth in public investment in irri-gation, though our numbers do not distinguishbetween public and private irrigation facili-ties. Why were landlord areas unable to takeadvantage of the new opportunities that pre-sented themselves after the mid-1960s? Wediscussed some potential answers in SectionII, and we assess their empirical relevance here.

Of the three alternative classes of explana-tions discussed earlier, the explanation based

on differential investment by the colonialstate is probably the least compelling, giventhat the major differences between the land-lord and non-landlord areas arose after 1965(Table 6). In principle, one could still arguethat the advantage they got from these earlypublic investments continues to help them inthe post-independence period.27 The fact that

27 Tirthankar Roy (2002) makes the argument that theareas that gained from the Green Revolution were those thatshowed improvements during the colonial period as well.

TABLE 6—WHEN DO THE DIFFERENCES APPEAR?

Panel A: Full sample

Dependent variable

Coefficient on non-landlordproportion

Difference(3)

1956–1965(1)

After 1965(2)

Agricultural investmentsProportion of gross cropped area irrigated 0.046 0.079** 0.033**

(0.033) (0.036) (0.016)Fertilizer use (kg/ha) 1.026** 15.581*** 14.55***

(0.425) (4.763) (4.44)Agricultural productivitylog (yield of 15 major crops) 0.066 0.201*** 0.135***

(0.065) (0.076) (0.033)log (rice yield) 0.108 0.196** 0.088**

(0.069) (0.089) (0.044)log (wheat yield) 0.146** 0.268*** 0.122*

(0.058) (0.079) (0.063)No. of districts 166 166 166Year fixed effects YES YES YESGeographic controls YES YES YESDate of British land revenue control YES YES YES

Panel B: Rice yields for Tamil Nadu districtsSample: 10 districts of Tamil Nadu. Data are for 1870, 1901, 1911, 1917, 1919, and five-yearly intervals from 1922 to 1982.


Dependent variable Before 1965 After 1965 Difference

Log rice yield �0.099 0.415 0.514**(0.172) (0.366) (0.217)

No. of districts 10 10 10Year fixed effects YES YES YES

Notes: Standard errors in parentheses, corrected for district-level clustering. * Significant at10-percent level; ** significant at 5-percent level; *** significant at 1-percent level. Data arefrom 1956–1987. Geographic controls are altitude, latitude, mean annual rainfall, and dum-mies for soil type and coastal regions. Estimates in column (3) are computed from a regressionof the dependent variable on the interaction of the non-landlord proportion and a dummy foryear �1965, after controlling for the main effects of these variables, as well as geographiccontrols.


the main source of the non-landlord advan-tage does not come from the mahalwari dis-tricts of northern India,28 which were themain beneficiaries of the canal constructionduring the colonial period, makes it harder tobelieve that this is the source of the entiredifference.

We noted in Section II that the landlord-controlled areas had higher levels of land in-equality in the colonial period. It thereforecomes as no surprise that the major landlord-dominated states enacted an average of 6.5 landreform measures in the period between 1957and 1992, while non-landlord states had an av-erage of 3.5.29 Besley and Burgess (2000) re-port that states that enacted a larger number ofland reforms had a somewhat greater decline inthe Gini coefficient of land inequality. This doesnot mean that there has been complete conver-gence in the land distribution in the two areas.As late as 1990, 64 percent of all land holdingsin landlord areas were classified as “marginal”(less than one hectare), which is about eightpercentage points higher than the correspondingfigure in non-landlord areas.30 Further, 48 per-cent of all holdings are small to medium sized(one to ten hectares) in individual-based areas,but only 35 percent in landlord areas. There isno significant difference in the proportion ofextremely large holdings, which is probably dueto the impact of land ceiling laws passed afterindependence.

These differences in the land distribution,however, cannot explain our results. For in-stance, if we were to ascribe the entire differ-ence of 16 percent in agricultural yields to thefact that there are more marginal holdings inlandlord areas, on the grounds that these hold-ings are less productive because they underin-

vest, we would have to accept that the smallholdings are only about 12 percent as produc-tive as larger holdings, which seems implausi-bly low.31 This also contradicts the evidencefrom developing countries, which suggests thatsmall farms are, if anything, more productivethan large farms (Binswanger and Rosenzweig,1986). Further, our results do not change whenwe control directly for the Gini coefficient ofland holdings in 1971 or the number of landreforms passed by the state. If we use consump-tion inequality as a better measure of wealthinequality, we find that landlord areas showsignificantly larger declines in consumption in-equality between 1972 and 1987 than non-landlord areas (Table 6 in the Web Appendix).In fact, by 1987 the landlord districts showsignificantly lower consumption inequality.32

We therefore feel that the biggest piece of thestory is probably the differences in the politicalenvironment. If the effect of the political envi-ronment operated mainly through the insecurityof peasant property in the landlord areas, how-ever, we would have observed convergencerather than divergence after independence, sincepeasant property clearly became less insecureonce the landlords lost their formal authority.This suggests that the important difference inthe political environment probably has to dowith the nature of collective action in the twoareas. We find that in addition to placing agreater emphasis on land reform measures,states with a higher proportion of landlord areasspent less on development expenditure. Be-tween 1960 and 1965, the landlord states spent13 rupees per capita on development expendi-ture, compared to 19 rupees in the non-landlordstates. This spending gap is higher in the post-

28 Table 3, column 4, shows that leaving them outmakes the non-landlord coefficient larger for some of theoutcomes.

29 Data on state-level land reforms comes from Besleyand Burgess (2000). We classify Bihar, Madhya Pradesh,Orissa, Rajasthan, Uttar Pradesh, and West Bengal as “land-lord” states, and Andhra Pradesh, Assam, Gujarat, Karna-taka, Kerala, Maharashtra, Punjab, and Tamil Nadu as “non-landlord” states.

30 The difference of eight percentage points is obtainedby regressing the proportion of marginal (less than onehectare) holdings on the non-landlord proportion, after con-trolling for geographic variables.

31 Suppose small farms are � times as productive as largefarms, z is the share of small farms and total productivity issimply the sum of large farm and small farm productivity.Then the percentage productivity difference between non-landlord and landlord areas equals {[(1 � �)�z]/[1 � (1 ��)zlandlord]}. Using productivity difference � 0.16, �z �0.08 and zlandlord � 0.64, we obtain � � 0.12.

32 These measures are computed using household surveydata from the National Sample Surveys (NSS). We shouldkeep in mind that these data are not at the district level butat the NSS region level, usually consisting of three to tendistricts. Our standard errors for these regressions are clus-tered at the NSS region level to take care of this aspect ofour data.


1965 period, just when new technologies wereappearing in the agricultural sector: landlordstates spent 29 rupees per capita, while thenon-landlord states spent a much higher 49 ru-pees per capita (Table 7 in the Appendix). Thisis not simply because of lack of resources: de-velopment expenditure as a proportion of statedomestic product is also lower in the landlordstates, and the difference in per capita spendingpersists even after controlling for state domesticproduct per capita (Appendix Table 7, column3). Given that the difference in the number ofland reforms is also mainly from the post-1965period, one way to characterize the difference inthe nature of public action is to say that land-lord-dominated states were busy carrying outland reform exactly when the non-landlordstates started focusing on development.

This difference in public spending turns outto be important in explaining our results. Whenwe add development expenditure per capita asan explanatory variable in our base regressions,we find that it sharply reduces the magnitude ofthe non-landlord coefficient for the measures ofHYV adoption (Table 7, column 2). The idea ofstate policy priorities as the major channel ofinfluence is consistent with what we find whenwe estimate the investment and yield equationsafter including a fixed effect for each state. Thisreduces the estimated coefficient on the non-landlord share substantially (by 50 percent orso), though the signs are unaltered and severalremain significant (Table 7, column 3).33 Thedifferences in state policies are also reflected inthe substantial difference between landlord andnon-landlord areas in the provision of educa-tional and health facilities: landlord areas had21 percent fewer villages (15 percentage points)equipped with primary schools, while the gap inmiddle school and high school availability are61 percent and 63 percent, respectively. Giventhese differences in investments, it is not sur-prising that literacy rates are 5 percentage pointshigher in non-landlord areas, while infant mor-

tality rates are 40 percent lower; both thesedifferences are statistically significant (Table7, panel D).34 A large part of these differencescan be attributed to the difference in state de-velopment expenditure (column 2).

Why are the political priorities so different inthese two areas? As already suggested in Sec-tion II, the masses in the landlord areas, withtheir memories of an oppressive and often ab-sentee landlord class, may perceive their inter-ests as being opposed to that of the local elite,while those in the non-landlord areas may bemore interested in working with that elite. Theexistence of a highly conflictual environment isconsistent with our results on crime rates (Ap-pendix, Table 8). Landlord districts have signif-icantly higher levels of violent crime (such asmurder, rape, kidnap, armed robbery, and riots),but not of economic crimes like cheating orcounterfeiting.

The perception of a large divergence of in-terests between the masses and the elite in land-lord areas may not, however, be necessarilycorrect. The final empirical exercise in this pa-per compares poverty reduction in the landlordand non-landlord areas. While the head countratio falls in both areas between 1972 and 1987(the mean reduction is about 11 percentagepoints), the decline in poverty according to ourOLS estimates is about seven percentage pointshigher in non-landlord areas (Appendix, Table 6).The difference in poverty reduction is five per-centage points for the sample of neighboringdistricts and is robust to the inclusion of a statefixed effect. The IV estimate, however, is com-pletely insignificant and has the opposite sign.In sum, there is no evidence that the masses farebetter in the landlord areas, and there is someevidence that they do worse. If we were pre-pared to attribute the change in poverty to thedifferences in political priorities and the result-ing differences in policies, these results wouldsuggest that the masses could perhaps havedone a little better, or at least no worse, byfocusing on what they had in common with theelites.

33 We need to be a little cautious when interpreting theseresults. Adding state fixed effects effectively drops thestates that have no within-state variation in non-landlordproportion. These states (Bihar, Gujarat, Karnataka, Rajas-than, and West Bengal) account for about one-fourth of oursample, so putting in state fixed effects results in a lack ofpower in our estimation.

34 IV estimates of these differences are larger in magni-tude than the OLS estimates for literacy, infant mortality,and primary school provision (results not shown).


REFERENCES

Acemoglu, Daron; Johnson, Simon and Robinson,James A. “The Colonial Origins of Compar-ative Development: An Empirical Investiga-tion.” American Economic Review, 2001,91(5), pp. 1369–1401.

Acemoglu, Daron; Johnson, Simon and Robin-son, James A. “Reversal of Fortune: Geog-

raphy and Institutions in the Making of theModern World Income Distribution.”Quarterly Journal of Economics, 2002,117(4), pp. 1231–94.

Alesina, Alberto and Rodrik, Dani. “DistributivePolitics and Economic Growth.” QuarterlyJournal of Economics, 1994, 109(2), pp.465–90.

Baden-Powell, Baden H. The land-systems of

TABLE 7—IMPACT OF STATE POLICY

Dependent variables

Mean ofdependentvariable


OLS Basespecification

(1)

OLS Control forstate dev expper capita (2)

OLSState FE

(3)

Panel A: Agricultural investmentsProportion of gross cropped area irrigated 0.276 0.065* 0.074** 0.028

(0.034) (0.035) (0.036)Fertilizer use (kg/ha) 24.64 10.708*** 10.805*** 4.297

(3.345) (3.717) (3.308)Proportion of rice area under HYV 0.298 0.079* 0.007 0.000

(0.044) (0.040) (0.042)Proportion of wheat area under HYV 0.518 0.092** 0.061 0.028

(0.046) (0.047) (0.039)Proportion of other cereals area under HYV 0.196 0.057* 0.025 0.043*

(0.031) (0.030) (0.026)Panel B: Agricultural productivitylog (yield of 15 major crops) 0.157** 0.174** 0.059

(0.071) (0.076) (0.072)log (rice yield) 0.171** 0.083 0.016

(0.081) (0.082) (0.078)log (wheat yield) 0.229*** 0.243*** 0.150***

(0.067) (0.072) (0.045)Panel C: Education and health investments, 1981Proportion of villages having:

Primary school 0.745 0.154*** 0.062* 0.102***(0.036) (0.037) (0.039)

Middle school 0.204 0.125*** 0.093*** 0.064***(0.023) (0.021) (0.018)

High school 0.082 0.052*** 0.019 0.030**(0.018) (0.014) (0.013)

Primary health center 0.023 0.011*** 0.002 0.012***(0.004) (0.004) (0.004)

Primary health subcenter 0.031 0.033*** 0.011 0.006(0.011) (0.009) (0.006)

Panel D: Education and health outcomesLiteracy rate (1961, 1971, 1981, 1991) 0.2945 0.0524** 0.0290* 0.0241

(0.0190) (0.0171) (0.0176)Infant mortality rate (1991) 82.17 �32.71*** �25.43*** �15.81***

(5.38) (5.28) (5.40)State fixed effects NO NO YESYear fixed effects YES YES YESGeographic controls YES YES YESDate of British land revenue control YES YES YES

Notes: Standard errors in parentheses, corrected for district-level clustering. * Significant at 10-percent level; ** significantat 5-percent level; *** significant at 1-percent level. Geographic controls are altitude, latitude, mean annual rainfall, anddummies for soil type and coastal regions.


British India, 3 Volumes. Oxford: ClarendonPress, 1892.

Bagchi, Amiya K. “Reflections on Patterns ofRegional Growth in India under BritishRule.” Bengal Past and Present, 1976, 95(1),pp. 247–89.

Banerjee, Abhijit V.; Gertler, Paul J. and Ghatak,Maitreesh. “Empowerment and Efficiency:Tenancy Reform in West Bengal.” Journal ofPolitical Economy, 2002, 110(2), pp. 239–80.

Banerjee, Abhijit V. and Newman, Andrew F.“Occupational Choice and the Process of De-velopment.” Journal of Political Economy,1993, 101(2), pp. 274–98.

Besley, Timothy and Burgess, Robin. “Land Re-form, Poverty Reduction, and Growth: Evi-dence from India.” Quarterly Journal ofEconomics, 2000, 115(2), pp. 389–430.

Binswanger, Hans P. and Rosenzweig, MarkR. “Behavioural and Material Determinantsof Production Relations in Agriculture.”Journal of Development Studies, 1986, 22(3),pp. 503–39.

Bose, Sugato. Peasant labour and colonial cap-ital: Rural Bengal since 1770. Cambridge:Cambridge University Press, 1993.

Brass, Paul R. The Politics of India since inde-pendence. Cambridge: Cambridge UniversityPress, 1994.

Chatterjee, Partha. Bengal 1920–1947, Vol. 1:The land question. Calcutta: K. P. Bagchi andSouth Asia Books, 1984.

Engerman, Stanley L. and Sokoloff, Kenneth L.“Factor Endowments, Institutions, and Dif-ferential Paths of Growth among New WorldEconomies: A View from Economic Histori-ans of the United States,” in Steven Haber,ed., How Latin America fell behind: Essayson the economic histories of Brazil and Mex-ico, 1800–1914. Stanford: Stanford Univer-sity Press, 1997, pp. 260–304.

Engerman, Stanley L. and Sokoloff, Kenneth L.“Factor Endowments, Inequality, and Pathsof Development among New World Econo-mies.” Economia: Journal of the Latin Amer-ican and Caribbean Economic Association,2002, 3(1), pp. 41–88.

Galor, Oded and Zeira, Joseph. “Income Distri-bution and Macroeconomics.” Review ofEconomic Studies, 1993, 60(1), pp. 35–52.

Gough, Kathleen. “Indian Peasant Uprisings.”Economic and Political Weekly, 1974, 9(13),pp. 1391–1412.

Guha, Ranajit. A rule of property for Bengal: Anessay on the idea of permanent settlement.Paris: Mouton & Co., 1963.

Gupta, Rai M. N. Land system of Bengal. Cal-cutta: University of Calcutta, 1940.

Iyer, Lakshmi. “The Long-Term Impact of Co-lonial Rule: Evidence from India.” HarvardUniversity, Harvard Business School Work-ing Papers: No 05-041, 2005.

Kane, Thomas J.; Rouse, Cecelia Elena andStaiger, Douglas. “Estimating Returns toSchooling When Schooling Is Misreported.”National Bureau of Economic Research, Inc.,NBER Working Papers: No. 7235, 1999.

Kumar, Dharma. The Cambridge economic his-tory of India, Vol. 2:c. 1757-c. 1970. Cam-bridge: Cambridge University Press, 1982.

La Porta, Rafael; Lopez de Silanes, Florencio;Shleifer, Andrei and Vishny, Robert. “Law andFinance.” Journal of Political Economy,1998, 106(6), pp. 1113–55.

La Porta, Rafael; Lopez de Silanes, Florencio;Shleifer, Andrei and Vishny, Robert. “TheQuality of Government.” Journal of Law,Economics, and Organization, 1999, 15(1),pp. 222–79.

La Porta, Rafael; Lopez de Silanes, Florencio;Shleifer, Andrei and Vishny, Robert. “InvestorProtection and Corporate Governance.” Jour-nal of Financial Economics, 2000, 58(1-2),pp. 3–27.

Lin, Justin Y. “Rural Reforms and AgriculturalGrowth in China.” American Economic Re-view, 1992, 82(1), pp. 34–51.

Misra, Babu Ram. Land revenue policy in theunited provinces under British rule. Benares:Nand Kishore & Brothers, 1942.

Mukherjee, Nilmani. The Ryotwari system inMadras 1792–1827. Calcutta: Firma K. L.Mukhopadhyay, 1962.

Patel, Govindlal Dalsukhbhai The land problemof re-organized Bombay State. Bombay:N.M. Tripathi Pvt. Ltd., 1957.

Persson, Torsten and Tabellini, Guido. “Is In-equality Harmful for Growth?” AmericanEconomic Review, 1994, 84(3), pp. 600–21.

Porter, F. W. Final settlement report of the Al-lahabad district. Allahabad: North-WesternProvinces and Oudh Government Press,1878.

Ray, Ratnalekha. Change in Bengal agrariansociety 1760–1850. New Delhi: Manohar,1979.


Roy, Tirthankar. The economic history of India,1857–1947. Oxford: Oxford UniversityPress, 2000.

Roy, Tirthankar. “Economic History and ModernIndia: Redefining the Link.” Journal of Eco-nomic Perspectives, 2002, 16(3), pp. 109–30.

Smith, W. H. Final report on the revision ofsettlement in the district of Aligarh. Alla-habad: North-Western Provinces and OudhGovernment Press, 1882.

Sokoloff, Kenneth L. and Engerman, Stanley L.“Institutions, Factor Endowments, and Pathsof Development in the New World.” Journalof Economic Perspectives, 2000, 14(3), pp.217–32.

Stokes, Eric. The English utilitarians and India.Oxford: Clarendon Press, 1959.

Stokes, Eric. “The Land Revenue Systems of theNorth-Western Provinces and Bombay Dec-can 1830–1948: Ideology and the OfficialMind.” in Eric Stokes, ed., The peasant andthe Raj: Studies in agrarian society and peas-ant rebellion in colonial India. Cambridge:Cambridge University Press, 1978a.

Stokes, Eric. “The Structure of Landholding inUttar Pradesh 1860–1948.” in Eric Stokes,ed., The peasant and the Raj: Studies inagrarian society and peasant rebellion in co-lonial India. Cambridge: Cambridge Univer-sity Press, 1978b, ch. 9.

Yanagisawa, Haruka. A century of change:Caste and irrigated lands in Tamil Nadu1860’s–1870’s. New Delhi: Manohar,1996.


history, institutions, and economic performance: the ... media...tenure system under british rule....

Documents