-~~~NS
Policy Research
WVORKING PAPERS B
Agricultural Policies
Latin America and the Caribbean
Technical Department and the
Agriculture and Rural Development Department
The World Bank
July 1993
WPS 1164
Power, Distortions,
Revolt, and Reform
in Agricultural Land Relations
Hans P. Binswanger
Klaus Deininger
and
Gershon Feder
If the efficiency of the large commercial farm is a myth, why do
markets for the rental and sale of agricultural land rarely reallo-
cate land to the most efficient uses and users (family farmers)?
PoUcyRcarch Wosking Papczs disseninatothe findings of work in progress and encourage the exchange of ideas among Bank staff and
alodtcsintrestedindvelopmentisues Thesepapers,disuibutedbytheRescarchAdvisoyStaff,carry thenamesoftheauthors5reflect
ordytheirviews,andshouldbouseda ndcitedaccordingly.Thcfindings,interpreutions.andconclusionsamtheauthors'own.nTcyshould
na be attributed to the World Bank. its Board of Directon, its management, or any of its member countries.



l ~~Polloy Research
Agricukural Policese
WPS 1i64
This paper -a product of the Advisory Group, Latin America and the Caribbean Technical Departnent
and the Agricultural Policies Division, Agriculture and Rural Development Department - was prepared
for the Handbook of Development Economics, Volume II, edited by Jere Behrman and T. N. Srinivasan.
Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433.
Please contact Hans Binswanger, rooml4-021, extension 31871 (July 199^, 121 pages).
Most work on the relationship between farm size       Binswanger, Deininger, ard Feder examine
and productivity strongly suggests that farms that    how these power relaJons emerged and what
rely mostly on family labor are more productive   legal means enabled relatively few landowners to
than large farms operated primarily by hired      accumulate and ho'd on to large landholdings.
labor.                                            They discuss the successes and failures of reform
in market and socialist economies, and the
This study began as an inquiry into how       perversions of reforms in both systems, mani-
rental and sales markets for agricultural land in  fested in large commercial farms and collectives.
the developing world affect efficiency and
equity. What emerged was lie clear sense that        They survey the history of land relations and
great variations in land relations around the     the legacies that history leaves. They discuss the
world and over time cannot be understood in the   three analytical controversies surrounding
common paradigm of property rights and com-       economies of scale, and the efficiency of the
petitive markets. Under that paradigm, land       land sales and land rental market.
scarcity leads to better definition of rights, which
are then traded in sales and rental markets           Tley discuss the main policy issues and
accessible equally to all players. The outcome    implications of various distortions and successful
should be the allocation of land to the most      and unsuccessful reforms in the developing
efficient uses and users, yet this rarely happens.  world, including land registration and titling,
land taxation, regulations restricting land sales
Instead, land rights and ownership tend to    and rentals, fragmentation and consolidation of
grow out of power relationships. Landowning       land, redistributive land reform, and
groups have used coercion and distortions in      decollectivization.
land, labor, credit, and commodity markets to
extract economic rents from the land, from            In an epilogue on methodology, they exam-
peasants and workers, and most recently from      ine how various strands of economic theory have
urban consumer groups or taxpayers. Such rent-    contributed, or failcd to contribute, to the expla-
seeking activities reduce the efficiency of       nation of variations in policies, distortions, and
resource use, retard growth, and increase the     land relations over space and time.
poverty of the rural population.
The Policy Research Working PaperSeriesdisseminates the rndings of work under way in theBank. Anobjectiveofthe series
is to get these findings out quickly, cven if presentations are less than fully polished. The fimdings, interpretations, and
conclusions in these papers do not necessarily represent official Bank policy.
Produced by 'he Policy Research Disseminadon Center



POWER, DISTORTIONS, REVOLT AND REFORM IN AGRICULTURAL
LAND RELATIONS
Hans P. Binswanger, Klaus Deininger, and Gershon Feder
Prepared for the Handbook of Development Economics, Volume m, Jere
Behrman and T.N. Srinivasan, editors.
The authors of thJ paper have benefittedfrom dcauslons at the Asian
Dvelopment Bank, the Land Tenure Center at the Universty of Wisconsin,
the University of Minnesota, and the World Bank Written comments and
suggestxion by AS.P. Brandaw, D. Bromley, J. Bruce, M. Carter, R.
Christiansen, E Hayaml, M. Lipton, S. Migot-Adholla, K Otsuka, M. Roth,
V.Rutan, and TN. SrWnivasan were particularly helpfid.



TABLE OF CONTENTS
Glossary
Introduction
Part I: The Historical Legacy
1.   The Emergence of Property Rights in Land
2.   Extracdng Tribute and Rent from Peasants
3.   Success and Failure in Reform
Part II: Analytical Controversies
4.   Farm Size and Productivity
5.   The Effects of Land-Credit Links and Policy
Distortions on Land Sales Markets
6.   Incentives, Land-Credit Links and Land Rental
Markeu
Part III: PoUcy
7.   Land Registaion and Xtling
S.   Land Tax
9.   Regulations Limiting Land Sales
10.   Fragmentation and Consolidation
11.   Restrictions on Land Rentals
12.   Redistributive Land Reform
13.   Decollectivization
Epilogue on Methodology
Annex 1:    Intervention to Establish and Support Large Famns
Annex 2:    How Market Lnperfections Affect the Farm Size - Productivity Relation
Bibliography
i



GLOSSARY
Irrespective of their historical and cultural or Ideologia origins, the following terms are used In this
paper with the definition given below:
Co.2ectlve Farm: A farm jointly owned and operated under a single management for the benefit of
and with work input from the owners of the collective.
CommuWa Cwnershp System: A system of land ownership in which specific plots of land are
assigned temporarily or permanently to members for family cultivation, while other areas are held in
common for pasture, forestry, and collection of wild plants and gamo. Individual plots mayor may
not be inheritable or tradeable in internal rental or sales markets. But sales to nonmembers are
always forbidden or subject to community approval.
Contract Fannlng. A contract between a farmer and a p-archr 'er in advance of the growing season
for a specific quantity, quality and date of delivery of an agricultural output at a price or price
formula fixed in advance. The contract provides the farmer an assured sale of the crop and sometimes
provides for technical assistance, credit, servicca, or inputs from the purchaser.
Cori*: Unpaid labor and sometimes the service of draft animal provided by serfs, tenants, or
usufruct right holder to the owner of the manorlal estate.
Debt Peonage, Bonded Labor Series: A tribute payment or labor service originating in a defaulted
loan.
Family Fanm: A farm operated primarily with family labor, with some hiring in or out of labor.
Family farming systems may be socially stratified, with wide dispersion in farm sizes and technology
levels.
Radenda. A manorial estate in which part of the land is cultivated as the home harm of t'e owner
and part as the family farms of serfs, usufrctuaury right holders, or tenants.
Home Farm: That part of the manorial estate or large ownership holding culdvated 'Jy the lord,
landlord or owner under his own management using corvde and sometimes partly remunerated labor.
Landlord Eftate: A manorial estate in which all of the land is cultivated by tenants or usufructuary
right holders.
Junker Estate: A large ownership holding producing a diversified set of commodities operated under
a single management with hired labor. Laborers do not receive a plot of land to use for their own
culdtivation as pant of their remuneraton, except perhaps for a house and a garden plot.
Large Commerdcal Farm: A large ownership holding producing several different commodities
operating under a single management with a high degree of mechanization using a few long term
hired workers who may reside on the farm and seasonally hired workers who do not reside
permanentI) on the farm.
Manodl Estate: An area of land allocated temporarily or as a permanent ownership holding to a
manorial lord who has the right to tribute, taxes, or rent in cash, in kind or in corvde labor of the
peasant residing on the estate. ITis paper uses the sme term whether the peasants are there by
1



choice or are bound by restrictions on their mobility. Manorial estates can be organized as ha,<indas
or as landlord estates.
Rcat is used in several ways:
* Rsudrfa    t  'hTe residual payment to a productive factor in inelastic supply afier all factors have
been renumerated at their respective market rates, whether the other markets are competitive or not.
* Rent-secing rent. The additional reward received as a result of regul-tions and restrictions that
raise the level of rewards above its undistorted level. Where markets are thin or uncompetitive,
measuring rent-seeking rent may be very difficult.
* Land rent: A tenant's payment to a landowner in a voluntary contractual relationship. Rent may
be paid as a fixed or share payment in cash, kind, or iabor sarvices. It may or may not be equal to
residual ren  If the reservation utility of tenants has been reduced by distortions associated with rent
seeking, the land rent includes a component of rent seeking.
Reservraon udiky or reserwtion wage. The level of utility (including the risk ataibutes) or the wage
which is available outside the manorial estate to a potential tenant or worker on a manorial estate.
Share contract: A rental contract in which the tenant is paying a portion or all of his rent by
delivering a certain proportion of the output, the crop share, to the landowner.
SatJbnws: A farm belonging to the state and operated like a Junker estate or a large commercial
farm under a single management with a largely resident labor force paid in wages, and sometimes,
profit shares in cash or in kind. Laborers may be allocated a small garden plot.
Swplhw: Output or labor available over and above that required to reconstitute and maintain the
energy and life of peasants, serfs or slaves.
bsa.e: A payment in cash, kind or labor services to a landlord based on restrictions on mobility
and/or other forms of state-sanctioned coercion. Tribute may also be called rent or corv6e.
U    qtfauatVuarj ,fghi . Rights to use the land. May be temporary, long-term, lifetime, or inheritable,
but always exclude the right to unrestricted sale of the land.
Wage, aanton: A large ownership holCing specializing in a single crop under a single management
using wage labor, a large share of which resides on the plantation but does not receive more than a
garden plot for self cultivation as part
of the remunion.
2



INTRODUCTION
This paper began as an inquiry into the efficiency and equity consequences of rental
and sales markets for agricultural land In the developing world. Most of the work on the relationship
between farm size and productivity strongly suggests that farms that rely mostly on family labor have
higher productivity levels than large farms operated primarily with hired labor. If that Is so, why have
markets for the rental and sale of agricultural land frequently not reallocated land to famUy farmers?
Why do extraordinarily unequal distributions of ownership and operational holdings persist in many
parts of the world? Why has land reform seemed to be necessary to change these land ownership
distributions?
What began to emerge from this study was the clear sense that the great variations in
land relations found across the world and over time cannot be understood in a simple property rights
and markets paradigm. Section 2 explains the idealized sequence of the emergence and definition of
property rights which occurred in only few areas of the developing world. As that paradigm would
have it, increasing land scarcity leads to better definition of rights, which are then traded in sales and
rental markets that are equally accessible to all players. The outcome should be the allocation of land
to the most efficient uses and users. Yet this often did not happen, as great observed deviations from
efficiency demonstrate.
An examination of the historical evolution of land rights shows the reason for the
deviations: rights over land and the concentration of ownership observed in most developing countries
at the end of World War II are outgrowts of power relationships. Landowning groups used coercion
and distortions in land, labor, credit, and commodity markets to extract economic rents from the land,
from peasants and workers, and more recently from urban consumer groups or taxpayers. Such rent-
seeking activities reduced the efficiency of resource use, retarded growth, and increased the poverty
of the rural population. How these power relations emerged and what legal meens enabled relatively
few landowners to accumulate and hold on to large landholdings. The terminology describing
agricultural production relations varies as much as the relations themselves do. We use a consistent
set of terminology and provide a glossary of definitions.'
A lag litete eaborats the imi",*on of paa model of land ue follwing the utcad   of v.
Thuen  for th optimal u# of land  .   aociad  problems of localied monopolies (Fujita and Thib
3



BIecse land ownership distribution has often been determined by power relationships
and disortions, and becuse land sales markets do not distribute land to the poor (the key point of
secton 5), land reform has often been necessary to get land Into the hands of efficient small fazily
owners (section 4 shows that they are indeed efficient). The successes and failures of reform in
market and socialist economies and the perversior of reforms In both these systems, manifested in
large commercial farms or collectives, are discuss&  ;3 section 3. The social cost of failing to
undertake reform-peasant revolt and civil war-are also considered.
But land reform would not be necessary if there were economies of scale in
agriculture beyond those that a family could take advantage of with a given level of technology. In
that case it would not have been necessary to use power to aggregate large holdings or coercion and
distortions to recruit workers. And in modem times it would not have beun necessary to subsidiza
lage commercial farms so heavily through credit subsidies and other distortions. Voluntary
tasactions in undistorted markets would have achieved these ends, and small peasants might have
found it attractive to join collectives. Section 4 examines the work that hits been done on the
presence-or not- of economies of scale in agriculture, Plnding in measurenients of the relative
effcienc of small versus large farmis only exceptional cases which are consistent with the myth of
the efficient large farm.
Similarly, if land sales markets could alloczwe land from inefficient large owners to
small family farmers, lana reform would not be necessary. Abolishing the special subsidies to lage
fas  and the conditions that permit coercion would be all that would be required to lead to the
brekup of large farms through sales to small farmers. Showing why sles markets are often not
capable of facilitating these efflciency-twhancing transfers - covariance of risks, imperfections In
credit markets, distortions in commodity market and subsidies to large farms are among the reasons -
is the topic of Section 5.
Section 6 then shows that tenacy and sharecropping are not as ineffAcient as often
assumed. They are second best adaptations to incomplete or distorted markets for labor, credit, and
risk diffulsion. Such rental agreements are also necessary to allow large ownership holdings to be
I (...oa _ned
1986), rgi and  urbon plniig, and the dorminant of lad values (Rndall and Castle 1985). 7he
rifarmis cited provide a good oveview of this litratr
4



oprated by tent as small family farm units. Regulating tenancy or outlawing it has perverse
efficiency and equity effects for the poor.
The sections of the paper are grouped in k ee parts. Part I covers the history of land
relations and the legacies it leaves today. Part II covers the whree analytical controversies surrounding
economies of scale, and the efficiency of the land rental and the land sales market. Part m discusses
the major land policy issues left behind by the various distortions and successful and unsuccessful
reforms in the developing world. These iniclude land registration and titling, lar.d taxation, regulations
limiting land sales and land rentals, fragmentation of land, redcstributive land reform and
decollectivization. Policy implications are discussed using the insights gained in the previous sections.
Finally, the methodological epilogue examines how var:ous strands of economic
theory have contributed, or failed to contribute, to the explanation of variations in policies, distortions
and land relations over space and time.
PART I: THE HIS     RICAL LEGACY
1. The energence of property rights In lhnd
7Te critical issue in land-ablndant settings is access to labor, not lancd. At low
population densities, there is no incentive to invest in soil fertility, and because fertility is restored by
long-tree fallow, ownership security is not required to induce investment. When population densities
rise, fallow periods are gradually shortened until the lRnd is continually cultivated. Then plows,
maure, artificial fertilizers, and other invesments and labor-intensive methods are required to
maintain soil fertlity (Boserup 1965, Ruthenberg 1980, Pingali et a., 1986). Marginal lands are also
brought under cultvation requiring higher investments still to make them productive. Now ownership
security becomes an important incentive for making the required investments. As the demand for
credit to finance inputs and investments in land improvements rises, the issue of land as collateral
becomes important.
lTus as population density increases private rights to land emerge in a slow and
gradual process that exhibits great regularity (figure 1, arrows 1 to 4). Boserup's (1965) discussion
of this proces is unsurpassed and so is quoted here at length:
S



Virtually all the systems of land tenure found to exist before the emergency of private
property in land seem to have ihis one feature in common: certain families are
recognized as having cultivation rigihts within a given area of land while other families
are excluded .... 'Free' land disappears already before the agricultural stage is
reached. Tribes of food collectors and hunters consider that they have exclusive rights
to collect food and to hunt in a particular area....
Under the system of forest fallow, all the men,bers of a tribe .... have a general right
to cultivate plots of land.... Tbis gerera' right to take part in the cultivation of the
land which the group dominates - or imagines to dominate - can never be lost for any
member of the cultivator families. They may voluntarily leave the territory for a time,
but they can then reclaim their right when they return ....
.a distinction must be made between the general cultivation riZht - as described
above - and the more specific right a family may have to cultivate a par,cular plot of
land. Under all systems of fallow a family will retain the exclusive right to the plot it
has cleared and cultivated until the harvest has been reaped .... But if, after the lapse
of the normal period of fallow, the family does not re-cultivate a given plot, it may
lose its right to this particular plo; .... Thus, the general cultivation right is an
inseparable element of the status as member of the tribe and, therefore, in principle
inalienable, while the specific right to cultivate an individual plot is lost by desuetude
,... As long as a tribe of forest-fallow cultivators has abundant land at its disposal, a
family would have no particular interest in returning to precisely that plot which it
cultivated on an earlier occasion. Under these conditions a family which needed to
shift to a new plot would find a suitable plot, or have it allocated by the chief of the
tribe....
But the situation is apt to change with increasing population, as good plots become
somewhat scarce. Under such conditions, a family is likely to become more attached
to the plots they have been cultivating on earlier occasions....
6



Hunter-gatherer            T e r r i t o r i a I  r i g h t s    t o   h u n t   a n d   g a t h e r
3~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Emergence of agriculture                     -       external market
land grants
0~~~~~~ 1
Forest fallow  I                                                      S~~~~~~ ~~~~~~~~~~~~lavery  External tabt
Forest fallow          1enera   I1h t a  t o     o  I t              I v a t *  a n d  9 r  a Z a               reservrla
9 n 9 r a I                                             6~~~~   U 
2
Em  eroen ce   o I  rIghts   to   apeaIf Ic   pI0t0
Bush fallow                         R 3  n  9 h t a  t o  I f I b tz I e  a n d I o *  I a a d     Slave      Indenturt,
15
Family Farm                              .                                Plantatlabor
________ ________ _______ ________ _____Plaantatio
communal tenure                   Manorial systemns                                      Plantatlo
tonants. corvee labor
self-cultivation of the home farm
I grater 'rigl to      
4       to efol il"|o [*
Permanent                               t                                                              15
unrestricted                    greater rlghtJ to overlords
cropping               right to sell                                                            Abolition ol
I                     ~~~~~~~slavir y
Family Farm    Landlord Estate                         Hacienda                I              L
owner-operated               entirely                  ten nt plus                   Plntatior
tenant-operated             owner-operated
home farm              Contract Farming



Permanent           Family Farm                Landlord Estate                       Hacienda.                      Wage
cropping  ...mamas.Isauve 11  *eaamu-.p.rmd     *       @waer.sparatsd       Plantation
croppingLndroorrr 
.  _  .
Land reformTen1
Junker Estate
*am.s3vI@Iei                      I                         lowner-operated\
r*eht is s@el                 12|
4                                       Land reform                                             \
7_____________6_________      subsldlde
ol ectlve an                                         Contract
Contemporary    Family Farm                                                  Large mechanize 
Systems              *,.-prld13                      State Farans   14    commercial farm                   Farming
Titilng and Land RKegletrlom     Dooollooltlwhallea      Eliminatlio of subslidie        ReodltrlbutlI  Lead Note,
POLICY           aegulatlsa of Land S01e     s                             RedlelilbatlvO Land Retori      Land Ta:atleu
Resulallef e* Laud R1at1le                                Teneasy regatatlos             c"*.sol l
QUESTIONS   Flagmemtltloead Co aoollidmtis                                 land ait  iloa
Land Taxatlsn



At this stage, when the attachment of individual familles to Individual plots becomes
more permanent, the custom of pledging land is also likely to emerge. If a family
does not need to use a given plot for a certain period It may pledge it to another
family .... subject to the condit:on that the land must be returned, upon request....
This custom o; pledging .... must be distinguished sharply from the sale of land
where the former occupier of the land loses all rights in it.
Thus, the attachme At of individual families to particular plots becomes more and more
important.... As more and more land is subject to specific cultivation rights, little
land will be available for redistribution by the chief....
As long as the general right of cultivation has not lost all its importance a sharp social
distinction exists in rural communities between cultivator families on one hand and
families without cultivation rights on the other, the latter group consisting of
strangers, whether they be slaves or free... even those strangers, who are not slaves in
a legal sense, are nevertheloss left with no other choice than to do menial work for
chiefs or for ordinary members of the dominating tribe....
Under both long- and short-fallow systems the land lying fallow at any given time is
at the free disposal for grazing by domestic animals belonging to families with
cultivation rights....h. Ile cultivaors' communal rights to use fallow land for grazing
will usually survive long after the general right to clear new forest land has
dsappeared.... (Boserup 1965, pp 79-80
Boserup's discussion makes clear that property rights in land are not simple and are
arely unresticted. As land becomes more scarce, general and Inheritable cultivation and grazing
rights are complemented by rights to resume cultivating specific plots after fillow (srrow 2), to
inherit specific plots rather than just general cultivaion rights, to pledge or rent out the plots, to use
them as coilateral in informal credit transactions, and to sell them within the community (arow 3).
When the right to sell Includes sales to members outside tta community (arrow 4), the last vestiges
of general cultivation rights are lost amd private property rights are complete. General rights survive
only as grazing and collection rights on communal grazing areas and forests, whc-e soils are usually
unsuitale for crop or intensive pasture production.
9



Even where communal land rights and management systems prevail, as in indigenous
communities of the Americas, or tribal communities in Asia and Africa, families have strong specific
laud rights. These rights p;ovide substantial 'ownership' security as long as the plots are farmed by
individual family units (Noronha 1985; Downs and Reyna 1988). Land rental and sales usually occur
withiln the community, especially among close kin.
While the internal rules and structures of these systems exhibit a bewildering variety,
all communal systems have one thing in common: Sales to outsiders are either forbidden or subject to
approval by the whole community.
The right to sell is often proscribed by laws that assign ultimate ownership to the state
or that regulate the land tenure of tribal or indigenous communities. Colonial powers often legislated
a uniform system of communal tenure to be applied to all land held by indigenous populations
(although tribal societies have often circumvented formal prohibition of land sales; Noronha 1985).
Under communal tenure family-owned plots can be used only for pledging in informal credit markets
and not as collateral in formal credit markets.
2. Extracting tribute and rent from peasants
History has few examples of the uninterrupted transformation of general cultivation
rights to land into owner-operated family farms (along arrows 1 to 4 in figure 1). Nearly always,
there has been an intervening period under a class of rulers who exacted tribute, taxes or rent from
cultivator families (arrow 5). The landholdings of these overlords (referred to here, for expositional
simplicity, as manortal estates, whatever the cultural or historic setting) were allocated temporarily or
as permanent patrimony or ownership holding, along with the right to tribute, taxes, or rent (in cash,
kind, or corvee labor) from the peasants residing on the estate. Frequendy, peasants' freedom to
move was restricted by bondage or by prior claims to land by members of the ruling group. The
rights of the ruling group were acquired and enforced by violence or the threat of violence and
institutionalized in tradition, custom, and the law and order forces of the state.2 The rights took
2 For Westan Europe, North and Ihomas (1971) intepret te rig  to triute as the emergence of a
contract between peasts nd maoril lords, with the lords provding protecon and other public goods i
exchange for tribute. Ihis view ipore the asymmety in the possesdon of the means for violence and judicial
pow.
10



numerous forms and left historical legacies in the distribution of land once land rights became fully
private. Again, Boserup (1965) says it best:
Above the group of families with cultivation rights is usually found an upper class of
tribal chiefs or feudal land who receive tribute from the cultivators.... The emergence
of a kind of nobility or aristocracy often seems to follow the introduction of short-
fellow cultivation with animal draft power.... Usually the position of a cultivator with
regard to his rights in land does not change because a feudal government imposes
itself and levies taxes and labor services. The cultivator families continue to have their
hereditary cultivation rights, both general and specific, and redistribution of land by
village chiefs may continue without interference from the feudal landlords. Nor does
land become alienable by sale; grants of land by overlords to members of the nobility
and others are simply grants of the right to levy taxes, and do not interfere with the
hereditary cultivation rights of the peasants. In other words, the beneficiaries of such
grants do not become owners of the land in a modem sense.... (pp. 82-84)
An analytical structure for the evolution of agrarian relations
For an analysis of the evolution of agrarian relations and the associated land
ownership distribution, several points are key. The first is that favorable agricultural conditions
generate the potential for rent-seeking rent or surplus and provide an incentive for groups with
political and military power to try to capture the rents or surpluses. The second is that under simple
technology there are no economies of scale in farming and that independent family farms are
economically the most efficient mode of production except for a very limited set of plantation crops,
(see section 4). Compared to large farms based on hired or tenant labor, owner-operated family farms
save on supervision costs of labor or eliminate the inefficiencies and supervision cost constraints
associated with tenancy.
Therefore, where population density is very low, peasants will establish their own
farms in the bush and thereby escape paying tribute, taxes or rent to the overlord. Extracting tribute
under these conditions requires coercion. Or the utility of the free peasant must be sufficiently
reduced so that they will offer themselves voluntarily as workers or tenants to holders of large tracts
of land at wages, rents or crop shares that provide the same level of utility as would independent self-
11



cultivation.3 Coercion is no longer necessary. Utility can be reduced by changing the free peasants
access to high quality land. Large landowners can also try to increase the supply of labor or tenants to
their holdings by Influencing governments to intervene through differential taxation of owners and
workers in large and small holdings, or by limitations on market access that drive down prnfitabillty
for independent peasants and thus reduce the reservation price of labor. Such economic distortions
increase the rent that goes to large-scale farmers at a cost to the economy of lower productive
efficiency.
When peasants can freely establish their own farms, It becomes very difficult to
operate large farms with hired labor under a single management. With simple technology, there are
usually no technical economies of scale (section 4). Lumpy inputs such as draft animals provide for
declining economies of scale at very small farm sizes. For larger farms the same draft-animal and
driver combination has to be repeated several times over, leading to constant technical rets.
Disincentives associated with hired labor give the family-operated farm a cost advantage over large
farms: for family members, there are no hiring costs, they have greater incentives to work than do
hired wage labor because they receive a share of profits, and third, site-specific learning costs are
lower.
Renting out entire small farms to sharecropper families (share tenancy) or granting
usufructuary rights to peasants in exchange for tribute allows large landowners to circumvent many of
the disincentive effects inherent in large wage-based faming and take advantage of the tenant fmily's
labor. Share tenancy has some incentive costs of its own, however, (section 6) and even under fixed-
rent tenancy there are problems of supervision and moral hazard.
Once a labor supply becomes available, large landowners can organize their
operations either as landlord esmas, with the entire estate culivated by tenanted peasants, or as
hacendas, with workers cultivating portions of the hacienda for their own subsistence as tenants or
holders of usufructuary rights and providing unpaid cornie or labor services to cultivate the home-
farm of the owner (see glossary). Since share teonats do not receive their full marginal product,
landlord estes based on a lump-sum rent payment would be the most efficient form of operation,
followed by landlord estates based on share rents. The hacienda would be less efficient since labor
3 Taking into acount any risk tduction t  lanoner may be able to provide as pat of the barin.
12



tenants have few incentives to invest, and landowners' cultivation of the home farm entails labor
supervision cost. These points are more fully elaborated now.
Coercion: As Boserup (1965) points out, 'Bonded labor is a characteristic feature of
communities with hierarchic structure, but surrounded by so much uncontrolled land suitable for
cultivation by long fallow methods that it is impossible to prevent the members of the lower class
from finding alternative means of subsistence unless they are made personally unfree' (p.73). Four
ways have traditionally been used to tie labor to large farms: slavery, serfdom, indentured labor
contracts, and debt peonage.
Meillassoux 1981, shows that for merchant slavery in which the slaveholders
purchase, rather than capture slaves, they must produce for the market to finance the slaves.' In areas
with sparse populations of hunters and gatherers and with ties to external markets, such as in the
United States' Southeast, the East Coast of Brazil, and the South African Cape, large farms had to
import slaves as workers (arrow 6) The native hunter-gatherers were too few to provide a steady
labor supply, or simply moved away. Large farms in areas with access to abundant labor reservoirs
such as the sugar islands of the Caribbean and Mauritius, Ceylonesfe (Sri Lankan) and Assamese tea
plantations, Malaysia, Sumatra, and South Africa were able to rely on indentured labor instead of
slaves (arrow 7). lhe workers had to be indentured to prevent them for the period of indenture at
least from establishing plots of their own or going into mining. Laws and police forces were used to
enforce indentured labor contract and to ensure the recapture and return of escaped slaves. The
capital cost of slaves, the cash requirements for recruiting indentured labor frOm distant lands, and the
4 Mesailloux also shows that these systems of merchant slaery were dependent on systems
of aistocratic skvery which engaged in the reproduction of the slave population through
raids and warfare on widely dispersed subsistence-oriented peasant populations. Domar
(1970) relates ownership rights in people - slavery and serfdom - to land abundance, which
makes extrating residual land rents impossible. What he did not distinguish is that slavery,
the purchase of the labor force, requires high levels of capital, which can be financed only if
there is a market, while serfdom involve ex tracting tribute without a purchase transaction,
and so no market is needed.
5 For a discussion of the transition from slavery to serfdom, see Mesailloux 1991.
13



absence of cash markets for food in fact implied that these systems could be used only for crops that
bad an export market.'
Serfdom or bendage could be used in somewhat more densely populated regions with
a settled peasant population and production primarily for only local consumption (arrow 5).7
Peasants would have had to move to more marginal lands to escape bondage. Slaves could not be
imported because there were no export earnings with which to purchase them. Overlords obtained the
right to tie subsistence-oriented populations to the land and to extract tribute or labor services. This
pattern arose during feudal periods in Western Europe, China, and Japan, and pre- and post
columbian America, and survived in Eastern Europe until the late nineteenth century (Blum 1977).
Debt peonage or bonded labor, another form of coercion, survived in many areas
even under high population densities. Where manorial estates had to compete with mines for labor
and therefore faced acute labor shortage, as in Guatemala and Mexico in the nineteenth century or in
South Africa in the twentieth century, vagrancy laws kept a pool of potential workers in prison for a
variety of petty offenses (see table 1). In South Africa farmers could invest in prisons in exchange for
rights to prison labor; these rights could even be traded.
Economic distortions
Where coercion was no longer possible, or sufficient, influential groups were able to
get governments to intervene to create economic distortions that would generate a labor supply for
their farms. Once population density was high enough for long falow agriculture to replace hunting
and gathering, peasants would establish independent farming operations in areas without slavery and
6  nTo temperate zrn of the AMericas (Canada, North Estern US, Southern Brazil, and Argentina)
ecwapd slavey because their producs could not be exportd competitively to temperate zon Europe unil the
advent of the s  itmshp and the railroad at a time when davery had gone out of dyle. The tropical and
sbtopical crops sugar, cotton, and tobacco facd no competition in Europe mankets.
7 Mesailloux (1991) also shows that these systems of merchan slavery were dependent on
slaves for systems of aristocratic slaveiy which engaged in the reproduction of the slave
populbaon through raids and warfire on widely dispersed subsistence-onented peasant
populapions.
14



bondage. With identical technology and a competitive output market, cultivation of the home farm
with wage labor would not be competitive with the free family farm because of incentive
disadvantages and labor supervision costs.
To get free peasants to move to the manorial estate required lowering expected utility
or profits in the free peasant sector in order to reduce peasants' reservation utility - expected utility
from family farming, including the risk attributes of the corresponding income stream - or shift their
labor supply curve to the right. This was achieved through four mechanisms:
Reducing the land availablefor peasant cultivation by allocating rights to
"unoccupied' lands so that they went to members of the ruling class only and thus
confining free peasant cultivation to infertile or remote areas with poor infrastructure
and market access. (rable 1 lists a variety of cases from all continents in which access
to high quality land was restricted). Farm profits or utility on free peasant lands were
thus reduced by the higher labor requirements for producing a unit of output on poor
land, by increased transport and marketing costs, and by increased prices for
consumer goods imported to the region.
*     ItImposing d5iferendoJ taxation by requiring free peasants to pay tribute, hut, head or
poil taxes (i cash, kind, or labor services) while often exempting workers or tenats
in manorial estates or taxing them at much lower rates. Such systems were used
widely in Western Europe during the feudal period, in ancient Japan, China, India and
the Ottoman Empire, and by all colonial powers (table 1). Tribute systems in Eastern
Europe and Japan survived into the second half of the nineteenth century. As long as
free peasants can pay tribute or taxes in kind or cash and have equal access to output
markets, taxation alone may be insufficient to bring forth a supply of workers or
tenants. They were therefore often complemented by output market inventicas.
*     RestrIcting market access, by commonly setting up cooperative or monopoly
marketing schemes that buy only from the farms of the rulers. The prazo system in
Mozambique combined rights to labor and tribute from peasants with monopolies on
inputs and outputs. In Kenya the production of coffee by Africans was prohibited
15



Table 1: Intervention to Establish and Suort LargIe Farms.
COUNTRY             LAND MARPT VVENTIONS                             TAXS AND  TBRVENTIONS IN LABOR AND
IOUTUIr MA1S                                , 
ASLA
Idia (Norah)        LAnd rans fin lot entuy                          Haced system; 4th cne  WC
Corece labor, from 2nd contr
Chin. (South)                                                        LImitations on pea n mobility; caSO
Tax eneption for swves; ca 300
Gentry exemption from   es & labor srvices; ea
1400
Jap                 Ecludve lnd rights to developed wsteland; 723    Tribute exemption for cleared and temple land; 700
Io and Sumata        Ad  grant to companies; 1870                     Indentued labor. 19th cenuy
Culivation System; 19th century
iwpimnes            Lan grSts to mostic    ler; 16th ce_y             Encomienda
Rimioato
Tax exemption for hacienda woren ; 16th cenatry
Sdri Lhan           Land appropdration; 1840                          aunation tax eonyt, 1818
Indenture labor 19th century
EUROPE.
Pnusa               Land rams; from 13th conuy                       Monopoes on millig and alcobol
Retorictions on hbor mobility, 1530
Land  form legisations; 17S0-180S
Ruwsa               lAnd grAs; from 14th century                      Resictions on pe   mobility:
Service tem ; IS65                                   - Exit fees; 1400/50
- Pobidden yea; 1588
- Bnerfwnt- 1597
- Tndabiliy of secr 1661
Hoan farm w pt from taxata  1580
De btpompge; 1597
Monopoly on commee; until 1830
S. AMERICA:
Chai                Land grant (ereda de  m); 16th ceouy              Eionends; 16th century
Labor ervic   (Wt); 17th centy
bIport dutis on beef; 1890
Subidies to mechanizstiom 19S50i6
El Salvdor          Gra  of public land; 18S7                         Vesnay lws; 182S
dli  of commu  nd; 1882                          Exenption from public an miitary smo  for
___ ___ ___ ___ __ ___ __ _lu__ _l   bandm s  end  thi  workers;  1647
uah         te mala   Retlmen of  dln   16t centr                     Cal trute; 1540
Manmiento; ea 1600
Debtpeonago; 1877
Mexico                   t t of lodis; IS40                              i    ;  1490
propratio of com _nal luand; 1                   Trt exemption for hacienda worke  17th c.
Dobt peongp; 1790
Rctu of debtorn to haciendas; 1843
Vagncy lws 1877
Viceryai of         Lad gSant; 1540                                   Encomianda; 1530
hay                 Resettlement of Inian (consr_gacla); 1570         Mhu: Exemption for hacienda wokear, ISS0
rift  ad expropriation of Inian lad; 17dh ceumry   Slavery of Africa; 1580
16



AICAs
A    a               MMng; co 1840                                    Tax oxemption fo worker on Eluopen fm; 1849
Land gru under ettlment progms; 1871              Credit proviskin fow vLP -|"t   settlerc
'Sealers' law' 1873
Atoag                 ndM concesIons to  uropeans; 1838, 1865         Slavery; until 1880
Vaganq law; 1875
Egypt (Ottoma)      Land grns; 1840                                   Corve. hbor, from 16th centwy
Corv. exemption for farm-workers; 1840.
Lan tax oxemption for lage badlords; 1856
Credit and mrketing  ubsidie, 1920 ad 1930.
K|uR                Land conwesions to Europea; ca 1900               Hut nd poll tues; from 190S
No African land purchaus outside rcsot; 1926     labor Pase; 1908
Squatter laws; 1918, 1926 and 1939
Restritions on Africans' maket acces; from 1930:
- Dual price sydem for maize
- Quanine and for" dstackig for livestock
- Monopoly markeoing acato
- Prohibition of African export crop civation
Subsidies to mechanizaton; 1940c
Sokotho Caliphate    Ln gmuns to settler; 1804                        Slavery; 19th century
Malawi              Lan allotmen  to Europea; 1894                    Tax eduction fr fam-wore; ca 1910
Mozambique          Comprhendve rights to lse under prmo; 19th        Labor tribute; 1880
csorry                                           Vagnucy law; 1899
Aboiion of African trade; 1892
Forced cultivation; 1930
Sou Africa          Native rserve; 19th ceonury                       Slavery and indentured hbor, 19th cenury
Pasud-commu   tenure in eaves 1894                Resrooe on Africn  mnobiii; 1911, 1951
Native Lads Ac; 1912                             Mono"  maren, from 1930
- Demarcof r                                   Prion labor. ea 1950
- E_imion of tenancy                            Dirct and indirct subsid; 20th cenuy
- Prohibition of African lnd purcase outside
Tanayia             Lan gat to setles, 1890                           Hut tax and corve requirements; 1896
Comnpulsory cotton production; 1902
Vagrancy lw  (work card); 20th cenury
Exchlson of Africans from cedit; 1931
.___ _ __ __ _ __ _ _   Markei  coops to deprs Africn   price; 1940
Zimbao              Roeerve; 1896 and 1931                            Pdb and hut ta; 1896
Dscrimatinm again  tnan_ , 1909
M_oy  mar et  boards, fiom 1924
- Dua price systm in maime;
- Faced detockin in livestock; 1939
outright until the 1950s. European monopolies on sales of tobacco in Zimbabwe and
Malawi were directly transferred to large farms after the countries gained
independence.
17



Coipning agrlcidturalpublic goods and services (roads, extension, credit) jo the
fawms of the ruders or subsidizing these farms directly was another means of incaeasing
their profitability relative to peasant farms.'
Sometimes the four types of distortions were supplemented by coercive interventions
in the labor market - vagrancy laws, debt peonage, and rural slavery are examples - to make it easier
to retain workers or tenants on manorial estates.
Since these four mechanisms involved legal or customary rules backed by the state,
they required a coalition between the overlords and the state. The combinations of distortions used to
establish manorial estates under conditions of low population density have been remarkably similar
across continents and over time (table 1). The earliest recorded incidence we found was in the
Arthasastra in the fourth century B.C. Once members of the ruling group began to establish viable
aricultural production geting enough workers for their estates required interventions in more than
one market. The most common pattern was to combine restrictions on land use with differential
taxaion. Groups with widely different cultures, religions, and ethnic backgrounds - Ottomans, the
Hausa and Fulani in Africa, the Fujiwara in Japan, and all European colonial powers - imposed such
systems on people of the same or different ethnic backgrounds when faced with similar material
conditions. Material conditions of production rather than culture seem to have led to the emergence
of the distortions.
Production relations on the manorial estate
On both landlord estates and haciendas, corvee, all of part of the land is cultivated by
peasants under tenancy contracts or usufructuary rights. In the hacienda, the unpaid labor services of
peasants who hold usufruct rights to some plots on the estate is used to cultivate the home farm of the
owner. Corvee may include the services of their draft animals and plows. The labor services of
tenants constitute all or a part of their rental payments for the use of the land. Peasants may be free
to leave the manorial estate or may be bound to it. Sometimes peasants receive a wage payment in
In Zimbabwe, Aficans had been encouraged to cultivate maize through the Master Farmr Ptogrammi
h lat 1920s when European farmer found it more profitable to grown tobacco and cotton. When those
m    coliapsed monopoly matketing and dual price systems were introduced and the Master Farmer Program
was abandoned, with responsible officials publicly declaring that they had never intended to "teach the Natives
to grw maiz in compedtion with European producers (Phimister 1988:235).
18



addition as part payment for their labor. Often, the resident labor force is complemented by
seasonally hired wage workers.
The extreme variation in the names and details of these arrangements and in their
local evolution over time has long stood in the way of comparative analysis in a single theoretical
framework. Yet common elements seem clear.
Landlord estates were prevalent in China, Korea, Japan, Eastern India, Pakistan, Iran,
Egypt and Ethiopia. In many colonial environments, it was easy for landlords to restrict peasants'
alternatives and maintain control over land and labor and sometimes over output mark-ets. Haciendas
emerged as the predominant form of manorial estates in Algeria, Egypt, Kenya, South Africa,
Zimbabwe, Bolivia, Chile, Honduras, Mexico, Nicaragua, Peru, and other countries in Latin
America, in the Philippines, in Prussia and other parts of Eastern Europe.
The home farm of the landlord often vastly exceeded the area actually cultivated. A
major purpose of the huge landholdings was to restrict the indigenous population's possibitities for
independent cultivation, and much of the land remained under forest or fallow or was devoted to
extensive livestock grazing. At the height of the feudal period in Western Europe between one-quarter
and one-half of the total area on manorial estates was cultivated by the owner in the home farm. On
Latin American and African haciendas, that share was initially a much lower, one-tenth (Palmer 1979;
Chevalier 1963).
Many historical accounts have noted the lack of competitiveness and limited
profitability of large-scale cultivation of home farms relative to landlord estates in which all land is
rented out. Tbat relative disadvantage Is also confirmed by a range of quantitative studies. Records
for the eighteenth and early nineteenth centuries show that in all of the cases investigated hacienda
owners in Mexico would have been better off by renting out all of their land at rents actually paid by
tenants rather than cultivating their home farms (Brading 1978). Many overlords survived
economically against competition from independent producers only because of their access to capital
markets and large-scale storage of maize which could be sold at high prices in poor years (Florescano
1969). The same applies to many Chilean and Peruvian haciendas in the sixteenth and seventeenth
centuries which yielded a return on capital of about 4.5 percent, considerably below the market rate at
which the overlords borrowed funds to keep up their living standards. They were able to repay
mortgages only because of a rapidly devaluing currency and the appreciation of their land (Moerner
19



1973:204). Labor productivity and total production on the patrons' plots were about half that on
tenants' plots in Peru and one quarter in Ecuador (Pearse 1975:91).
What explains the total amount of tribute, surplus, or rent that could be extracted
from the peasants on the manorial estate? The predominant explanation for European estates was a
demographic-economic model based on Malthus and Ricardo (see, for example, Postan 1973; Le Roy
Ladurie 1966, and 1985; North and Thomas 1971; Brenner 1985; Holton 1977) that relates tribute
burdens to relative scarcities of land and labor. Before ruling groups controlled most of the land or
were able to coerce labor, attracting or retaining peasants to manorial estates in areas of low
population density required that peasants' utility on the manorial estate exceeded their reservation
utility for subsistence farming in the bush or in areas from which they had to be induced to emigrate.
In Europe east of the river Elbe such terms usually included a grant of hereditary usufruct rights.
Initially, most corvee labor was devoted to the construction and maintenance of infrastructure.
As long as population densities were low, corvee requirements had to be regulated
and enforced by the state. But as rising population densities and increased land scarcity reduced
peasant mobility, it became possible to increase the amount of tribute extracted and to increasingly
rnsform that tribute into obligations to work on the landlord's home farm Labor requirements, of
two to three days a week in feudal Europe, nineteenth century Russia, Kenya in 1918, and Central
and South America, began to rise with growing land scarcity. In Kenya, corvde requirements for
squatters and their families had risen to five days a week by the end of the colonial period (Resident
Labor Ordinance of Kenya, 1939).
This simple demographic-economic model falls to explain, however, why European
regions reacted so differently to the plague-induced declines in population in the fourteenth century.
The associated drop in tribute contrbuted to the erosion of serfdom in Western Europe, but led to the
reimposition of serfdom in Eastern Europe. In the debate over the demise of feudalism in Europe,
Bremner (1976, 1982) clearly established that economic factors such as population density and market
access alone are insufficient to determine the income distribution between peasants and lords in the
manorial estate. At best, they determine not the actual amount of tribute or surplus that could be
extracted from peasants but rather the maximum potential amount The lords' success in extracting
tribute depended on their political power to claim the land, monopolize markets, and control the
movement of peasants relative to the power of peasants to resist these efforts.
20



Barganing between peasants and lords and the distribution of Income
The amount of rent extracted thus depended on the outcome of a bargaining game, the
political confli^t, or the class struggle over the definition of 'property rights" in the widest sense.
That means that the cohesiveness of the landlords relative to that of the peasants and the success of
the alliances they could forge - with the King, the bureaucrac), other production sectors, the fnancial
sector, and external interests - are central to an analysis of change in the instruments of surplus
extraction to landed classes.
In the bargaining over the terms of income distribution between peasants and
landlords on the manorial estate, two sets of issues must be dealt with. One is to define the admissible
set of property rights and of coercive or voluntary exchange relationships, including the instruments
used to enforce such relationships. ITis problem includes the ability of overlords to impose
restrictions on peasant mobility and output markets, the broad terms of legitimate leases (inheritable
usufruct, long-term leases, short-term rental), the forms of rental payment available (cash, kind,
labor, fixed rent, crop share) and the sanctions (eviction, physical punishment, fines) or instruments
that ca be used to enforce such changes. The other is to determine optimal mix and level of use of
each intrument for maximizing surplus extraction, taking the available options as given. Although
this question is more amenable to economic analysis than is the problem of the admissible set of
instuments, there has been little formal modeling of it, even for environments without coercion (see,
for example, Carter and Kalfayan 1990; Carter and Zimmerman 1992; and Sadoulet 1992).'
This second problem could be set up as a bargaining process between landlord and
potential tenants. The landlord who maximizes his income or utility subject to the tenant's reservation
utility consaint, determines the terms of the tenancy, the size of the tenant's plot, and the size of his
own home farm according to the following considerations: he can set the overall rent burden of the
tenat. He can partition the rent into corvde, fixed rent payments in cash or kind, and crop shares,
9  Cintr and Kalfayam (1990) show that the combtion of a labor supevion constrint and a wordng
capital constaint can rsault in the emergence of tied labor contracts. Carter and Zimmerman (1992) provide a
dynamic extension of this model and demonsbta  the emergence of a number of the salient charateristics of
dual agrian societia aS a conquace of credit and labor sipervion problems. Sadoulet (X992) explains the
emergec of labor sorvice teancy as a device adopted by the ladlord in order to enfore an optimal level of
nsurance agpinst deft by te tenant in the case of crop failure. Covaiance of yields between the landlord's
hnorn fatm and tenants' plots is igored howevr. But in yar of crop failure the tenants' labor has no value on
tIe hom faram ether, and forcing him to provide it only leads to extra supervision costs. Sadoulet's explanation
hereo  fails.
21



each having its own incentive problems. He can choose the amununt of land allocated to home farm
cultivation, knowing that incentives are required to bring for,"% effort and that supervision is costly.
He can choose the size of the plot allocated to the tenants, knowling that family farms provide high
incentives to produce but may lead tenants to concentrate on dt'wr own plot and not supply sufficient
effort for home farm cultivation.
With peasants free to leave, the major constraint faced by the landlord is that he
cannot drive the utility received by his tenants below their reservation utility - the utility they could
receive working in the free peasant sector outside the manorial estate or in an urban labor market.
ITe tenant, for his part, can vary the labor effort on his own farm or leave for frontier areas,
indigenous reserves, or urban labor markets. So even without coercion or the ability to affect the
reservation utility, the landlord seems to have an abundance of instruments for driving the tenant
down to his reservation utility. Without further restrictions on the bargaining problem, its solution
may be indeterminate.
Constraints on the bargaining problem imposed by the state - restrictions on peasant
mobility, on the size of parcel to be allocated to peasants in inheritable usufruct, or on the tribute,
rent and corvde requirements, for example - can simplify the structure of the bargaining problem for
specific historical settings. But these outside regulations did change, albeit slowly, in response to such
forces as population densities and political conflict, so, they can not truly be regarded as exogenous.
Thus the complexity of the problem remains.
Rent seeking, coaUtions and conflict
The analytical problem becomes even more complex if it incorporates rent seeking or
surplus extraction through efforts to change the set of instruments available to landlords. A coalition
or class of landlords can try to induce the state to manipulate the reservation utility of peasants and
may succeed if peasants or workers are poorly organized to resist the change. We have not found any
models addressing these choices or game theory problems formally, but the literature is rich in
discussions of changes in the degree of coerciveness of the systems and of changes in other
instruments. North and Thomas (1971), for example, in an informally stated bargaining model,
analyze thd choice between tribute in cash or kind and corvee labor, suggesting that corvee was
preferred over tribute in kind where output markets were limited, and the relative prices of goods
22



were highly variable. There are many other examples, however, of frontier societies without external
markets in which tribute was collected in kind.
While the bargaining problem has received little formal analysis, manorial systems
have sometimes been interpreted as the outcome of an efficiency-enhancing contract between peasants
and landlords: the landlords provide protection and other public goods (which are produced with
economies of scale and require some specialization) in exchange for tribute or rent (North and
Thomas 1971, for example). This is a plausible interpretation for land-abundant settings, where
tribute rates or labor rents have to be set low enough to attract immigrants. However, there are two
major problems with this view.
First, it ignores the asymmetry between contracting parties in access to weapons,
laws, and public investment budgets. The systematic use of these instruments throughout history has
depressed the utility of peasants and workers to far below the reservation utility that would obtain in a
system without such symmetric access. Moreover, there is little doubt that substantial deadweight
losses and dynamic inefficiencies have been associated with taxes and tribute, with inequalities in
factor ratios between farming sectors, and with restrictions on access to credit and output markets.
Second, the contract view ignores the likely competition in rent seeking between
landlords, which would add to the deadweight loss associated with restrictions. Competitive rent
seeking, the literature shows, is likely to result in the dissipation of the rent into such rent-seeking
costs as competitive armies, arsenals, and fortiScations, which provide no consumption value.
Brenner (1985) argues that at the height of the feudal period, rents were completely dissipated into the
costs of competing in the system. Periodic conflicts over the right to extract rent have caused
destruction and decline in many flourishing kingdoms and empires, so the efficiency characteristics of
the contractual system are only third or fourth best.
Conclusion
The major issue in land relations, then, is the evolution of the relationship between
peasants and landlords over time. The best developed literature in this area relates to the demise of
the manorial estate, corv6e, and bondage and the emergence of capitalism in Europe. Dobb (1976)
23



interprets the emergence of capitalist farming and the loss of rights to tribute as the consequence of
increased population density alone, while Sweeney (1976) emphasizes the role of increased access to
markets. Brenner (1985) shows that these explanations alone are inadequate, arguing the need to
introduce the cohesiveness of the two groups and the strength of the coalitions they can form witk
lings or urban groups. Holton (1977) also discusses these issues, as well as broader non-economic
theories). In particular, Brenner stresses the importance of the cohesiveness of the peasant community
in resisting attempts by the lords to increase the instruments available to them or the intensity of their
use.
3. Success and failure in reform
How does the manorial estate disappear? Again Boserup (1965) explains succinctly:
"Ibe process by which the feudal landlord tenure [the manorial estate] is abandoned may take
different forms: sometimes the position of the feudal landlords in relation to the cultivators is
weakened; they lose their power over all or most of the peasants and they end up as private owners of
their home farms only [figures 1 and 2, arrows 8, 10, and 11]. In other cases, the feudal landlords
succeed in their efforts to completely eliminate the customary rights of the culdvators, and they end
up as private owners of all the land over which they had feudal rights, whilst the cultivators have
sunk to the status of tenants-at-will. England, of course, is the classical example of this last kind of
development' pp 79-87. In transitions of the first kind the peasants end up with the land rent, while in
those of the second kind, the landlords retain the rent.
Since land reform involves the transfer of land rents from a ruling class to tenant workers, it
is not surprising that most largescale land reforms were associated with revolts (Bolivia), revolution
(Mexico, Chile, China, Cuba, El Salvador, Nicaragua, Russia) conquest (Japan and Taiwan), or the
demise of colonial rule (Eastern India, Kenya, Mozambique, Vietnam, Zimbabwe). Attempts at land
reform without massive political upheaval have rarely succeeded in transferring much of a country's
24



land'° (Brazil, Costa Rica, Honduras) or have done so very slowly because of a lack of political
commitment to provide the funding to compensate owners (see section 5).
The outcome of land reforms has been conditioned by three factors: whether the
system was a landlord estate or a hacienda system, whether reforms was gradualist with compensation
or took place all at once, and whether the reform was undertaken in a market or a socialist economy.
We consider the first two factors in the context of the third, the type of economy.
Reform In market-based economies
Rapid transition from landlord estates tofamiyfaims in a market economy (figure 2,
arrow 7) has led to stable systems of productions relations. The organization of production remains
the same family farm system. The only change is that ownership is transferred from large landlords to
tenants who already farm the land and have the skills and implements necessary to cultivate their
fields. Government involvement in the transition has often been substantial from a ceiling on the size
of landholdings and the amounts to be paid for the land, to the establishment of financial obligations
of beneficiaries. Many reforms, that followed this pate  provided stronger incentives for tenant-
owners to work and invest in their farms and led to increases in output and productivity. The
10  Horowitz (1993) models land reform as the outcome of a Nash bargaining betwe  to agents
representing landed elites and the poor. Eh party can either agree to a reform propoul or iniiate 'revolt',
defined as a lottery over the thr outcomes *victory for te rich', 'victory for the poor', and 'maintemane of
the status quo'. The power-stucture which, in the cas of revolt, determines the probabilities for each of hse
events i take to be exogenous and time-invariant.  his leads to the definiton of a safe rform plan as tho
evolution of landholdins over time which consitus a Nash equilibum in tie bargaing pm  betwee
lbdlord and pests which, at any  t in tim, pvides each party with a level of utility at least equal to
teir expetod udtliy i th case of revolt Horwitz shows that in the cas of risk neuality (i) th  exdst a
unique safe reform juan for every initial distrbution of lanoldings whih can entail either rediutionu   from
the rich to the J io or accum atio of land by the rich; (ii) for any given power st r, tho extent of land
asfer is tho greatr of the higher of the initial imbalances in landholdings; (iii) except in pcial cases, the
sfe land reform plan is a prolonged process con sing of a sequence of individual reform events rather than a
one-tme rdistnbution. This approach is the first formal model in which the depedence of the equilibrum
ladholding pattemn on the power stucture is clearly elaborated. Ihe determinants of power, such as coalitions
wi third groups and intemal cohesiveness are not modeled, however, but from the model it is clear that
changes in the power strcture (such as the changes taing place in many parts of the world after 1945) and the
insumets available to landlords to reduce peasants' reservation utility will have major implications for th
stdbe land distbutiao
25



resulting systems have had great stability. Since the end of World War II, landlord estates in Bolivia,
large areas of China, Eastern India, Ethiopia, Iran, Japan, Korea, and Taiwan have been transferred
to tenants in the course of successful land reforms.
Theoretically, the productivity gains associated with such reforms come about because
of improved work and investment incentives associated with increased security of tenure. These gains
may be modest if tenants had to compensate landowners at near-market prices, if security of tenure
had already been high, if cash-rent contracts had prevailed, or if the disincentive effects associated
with share-tenancy had been low as suggested by Otsuka and Hayami (1988). Empirical evidence
shows that the reform of landlord estates led to considerable investment, adoption of new technology
and increases in productivity (Callison 1983; Koo 1968; King 1977; Dorner and Thiesenhusen 1990)
and that costs to the government of complementary investments supporting the transition in ownership
structure, such as infrastructure, housing, training in management skills, were low because the
structure of the smallholder production system was already in place.
By contrast with the relatively smooth transition from landlord estates to family farms,
reform of hacienda systems has been very slow and difficult. The outcome has frequently been the
emergence of large owner-operated Junker estates with greatly increased home farm cultivation
(arrow 10). Junker estates produce a wide variety of crops and livestock products using a hierarchy
of supervisors, permanent workers who sometimes are given a house and garden plot, and external
workers hired on a seasonal or daily basis. Junker estates are less specialized than plantations, which
produce and process a narrow range of crops (discussed in section 4 on economies of scale), and less
capital-intensive than large-scale commercial farms.
Expansion of the landlord's home farm at the cost of land cultivated by tenants for
their own use would be associated with losses in efficiency. Therefore, rational landowners would not
establish Junker estates unless induced to do so by such external constraints as the threat of land
reform or restrictions on tenancy designed to protect tenants' rights. Anticipating such reforms,
landowners often tried to reduce their exposure to expropriation by evicting tenants who usually are
the beneficiaries of land reform. The lack of competitiveness of Junker estates with the more efficient
smallholder sector made Junker estates an unstable form of production relations and led to intensive
lobbying for protection and for subsidies to introduce and expand mechanization.
26



By substituting subsidized capital for labor, the Junker estate was transformed into a
large-scale mechanized commercialfarn (arrow 11) that no longer depended on large amounts of
labor. Intensive mechanization of large commercial farms reduces the potential for land reform since
there are not enough families with farming skills and implements available on these capital intensive
farms to result in the establishment of efficient small farms able to rely on low-cost family labor. A
similar result can be achieved by converting haciendas or junker farms to livestock ranches, which
requires very little labor.
The early rounds of land reform in Prussia gave freehold property rights to hereditary
tenants, requiring them to services give up one-half to one-third of their hereditary land to the Junkers
r s compensation for the loss of their corvee services. Fearing that further land reform would include
tenants at will or holders of nonhereditary usufruct rights the Junkers evicted many of the remaining
tenants and reverted to cultivation with hired labor.
In Latin America, ever since the Mexican Revolution in 1910, land reform
movements have legally enshrined the principle that land belongs to the tiller and that indirect
exploitation of the land through tenants constitutes a cause for expropriation. The Brazilian Land Law
of 1964 puts a low ceiling on rental rates and crop shares and conveys permanent usufruct rights to
tenants after a few years of tenancy by protecting them from eviction. Similar provisions exist in
some land laws in Asia (Chuma and associates 1990). Restrictions on tenant cultivation in South
Africa had different roots - they were imposed to make tenancy less attractive to Africans who were
needed as workers in the mines. Whatever the motivation, these legal restrictions on tenancy induced
owners of haciendas to evict their tenants and to expand home farm cultivation with hired labor, or
shift to ranching, which requires little labor, or to adopt mechanization.'1
That Junker estates emerged only in response to pending land reform and tenancy
restrictions supports the view that there are no technical economies of scale in unmechanized
" deJanvry and Sadoulet (1989) argue that the threat of land reform and their ability to lobby in coition
with the urban sector for subsidies and provision of public goods led large landowners to mechanize and make
the transition from haciendas to large mechaized commercial farms in Colombia (1961-68), Ecuador (1936-
57), Peru (1964-69), Venezuela (1959-70), and in Chile (after 1972). In Ecuador, two separt stages can be
distinguished. Widespread eviction of tenants and the formation of Junker estates, until 1957 was folowed by a
period of increased emphasis on the family-farm sector together with widespread mechanization (1958-73).
27



agriculture and that the incentive problems associated with supervising hired or corvde labor far
exceed the efficiency losses associated with long-term whole-farm tenancy contracts. To compete
successfully with family farms, Junker estates had to find ways to reduce their labor costs or to
increase their revenues. Having lost their rights to rent or labor services from tenants or workers,
landowners often sought to secure rents from the expanding urban and inJustrial sectors through trade
barriers and subsidies for mechanizing production (de Janvry 1981). Trade barriers, by banning or
reducing foreign agricultural competition forced consumers to subsidize Junker estates or commercial
farms. Examples include the German Zollverein at the end of the nineteenth century (Gerschenkron
1965), tariffs on beef imports in Chile in 1987 (Kay 1992), and selective price support to products
from large-scale units in Kenya, Zimbabwe, and South Africa (Deininger and Binswanger 1993).
Subsidies for mechanization led to the transformation of nearly all Junker estates into mechanized
commercialfarms (arrow 11). Huge sums were provided either through direct mechanization
subsidies, as in Kenya, or through cheap credit, as in South Africa, Zimbabwe, and virtually all of
South America, where real interest rates were even negative (Abercrombie 1972). Mechanization
eliminated the need to rely on hired labor and resulted in widespread tenant evictions even in
countries with cheap labor - hardly an optimal transformation from a social point of view.
In some market economies baciendas were converted to communalfamiyfarm systems
(arrow 11). Communal tenure was adopted first in Mexico's ejido system and later, under land
reforms in Bolivia, Zimbabwe, and elsewhere. Beneficiaries were granted inheritable usufrucuary
rights, but constraints on land sales and rentals often prevented using the land as cottateral for credit.
Attempts to provide alternative sources of credit tirough special bansl or credit programs proved
ineffective (eath 1992; World Bank 15991). In Mexico, recent constitutional amendment legalizes
land rental and sales within all ejidos and allows each ejido to remove restriction on sales to
outsiders, by a majority vote, effectively converting the ejidatarios to owner-operated family farms.
Reforms In sodalast economies
Reform in socialist economies (figure 2, arrows 10, 11, and 12) has followed
different paths. Landlord estates in the former Soviet Union, Vietnam, and China were initially
converted intofamilyfarms (arrow 10), in much the same way as in market economies. Ihe
redistrnbuted farmlands were later consolidated into single management units or collectives (arrow
28



13), In which land is owned and operated jointly, under a single management. Families do not operate
their own plots as they do in systems of communal ownership.
In Algeria, Chile, East Germany, Mozambique, Nicaragua, and Peru, Junker estates
or large commercialfanns were converted directly into state farms (arrows 14 and 15). In most
cases, workers continued as employees under a single management, with no change in internal
production relations. Over time, the organizational differences between collectives and state farms
tended to disappear.
A desire to maintain presumed economies of scale in production and related activities
(input supply, marketing) or to educate the beneficiaries of reform during a limited transitional period
(Chile), motivated the establishment of collective and state farms. But to achieve efficient production
collecives have to deal with two incentives problems. One is to provide incentives to workers, a
problem addressed by the adoption of piece-rate remuneration systems designed to reward labor at
least partially on the basis of effort. Even where members of collectives were not able to divert effort
to private plots, lack of incentives and of disciplinary measures by central management led to serious
labor shortages following the transformation of private into collectively owned farms in Cuba
(MacEwan 1981) and Nicaragua (Enriquez 1992).
The other incentive problem concerns investment and savings decisions, which are
made jointly by the colective. Bonin (1985) shows that as long as equity financing is precluded and
members cannot market their share in the cooperative, the representative worker will not make
efficient investment decisions. Mitchell (1990) also examines problems associated with the
intertemporalallocation of consumption and shows that the distribution of decision-making power
between old (who would rather consume) and young (who prefer to invest) determines the rate of
growth for a cooperative enterprise. Successfil collectives tend to degenerate into capitalist enterprises
(or wage-labor-operated state farms) by successively substituting cheaper wage laborers for more
expensive members (Ben Ner 1984). McGregor (1977) provides a theoretical justification and
empirical examples of the tendency of cooperative enterprises to disinvest and to reduce membership
in order to increase current consumption by members. Barham and Childress (1992) showed that
Honduran collectives decreased their membership over time by about one fifth. Thus, the problems
associated with prvision of workers' effort and intertemporal consumption proved at least as serious
29



in collectives as in haciendas (Bonin and Putterman 1986; Putterman 1989). The poor performance of
agriculture under a collective mode of production is well documented and It is not surprising that the
expected increases in production from economies of scale were not usually realized (see, for example,
Colburn 1990 for Nicaragua; Ghai, Kay, and Peek 1988 for Cuba; Ghose 1985, Wuyts 1982, and
Griffin and Hay 1985 for Ethiopia and Mozambique, Lin, 1990 for China). Once given the chance to
do so, members of collective farms often voted to redistribute plots to family-szed farms.'2 True
economies of scale would induce economically rational farmers to establish collective forms of
production (Putterman and Giorgio 1985). In the absence of other possibilities of insurance,
collective forms of production would be chosen, due to the implicit insurance they provide against
noncovariate risks, even in the absence of economies of scale (Carter 1987). However, cooperative
production does not insure against covariate risks. Empirical evidence indicates that social ties may
be a less costly way to insure against risks that are not covariate (Walker and Ryan 1990).
In China, agricultural output in the first six years after decollectivization in 1978 grew
by 42 percent, with most of the growth attributable to the change in production organization (Lin
1992, Fan 1991, McMillan et al. 1989, Nolan 1988). Vietnam experienced similar productivity gains
from breaking up large unmechanized collective farms into tiny family units (Pingali and Xuan 1992).
The small family farms in these densely populated countries expanded the labor input and were able
to reduce machinery and ferdlizer use. Clearly, the incentive advantages of individual farming
outweighed any efficiency losses due to the exremely small size and fragmentation of farms
(Wenfang and Makeham 1992).
Under different conditions, as in Algeria and Peru (Melmed-Sanjak and Carter 1991),
the privadzation and breakup of mechanized state farms or collectives has been less successful.
12 Ortega (1990) offers quantitative evidence for the decline of the collective sector dtoughout Latin
America. In Pemu, tho absence of economies of scale led reform bendiciaes to effectively subdivide reform
colloctives by cncentaing effort on their private plots and to ptess for legal subdivisions and individual land
titles (Kay 1983; Horton 1972; McClntock 1981). Collctivas failed in Zimbabwe and were soon abandoned in
favor of a smallholder-oriented strategy (Weiner 1985). Similrly, colectives failed in the Dominicam Republic
and were replaced by cooperatives, with individually owned plots (Meyer 1991). Land reform cooperatives in
Panama are highly indebted and use labor far below profit-maximing lvels (ie  1987). Algerian
production cooperatives experienced low pductivity  membership dsertion, high ue of mnion, and
cnderable underemployment of the workforce (Pfeiffer 1985; Trtman 1985). The same patten of declinig
output and transformation into a 'collective Junker estate' has been observed in Mozambique (Wuyts 1985).
30



Mechanization of these large farms had occurred and had reduced the number of workers or tenan
before their collectivization. When these collectives were turned over to their relatively few remaining
workers, the resulting family farms were relatively large and unlike in China and Vietnam could ant
be operated efficiently without additional hired workers or high levels of mechanization. But hiring
additional workers dilutes the incentives advantage of the family farm, and the farms had neither the
access to subsidized credit nor the large amounts of equity needed to finance hired labor or the
mechanization. To make reform work under these capital-constrained conditions and reap the
efficiency benefits of family farming may require including more beneficiary families in the reform
program than those employed on the highly mechanized farms, by resettling landless or near landless
workers from outside the farms (Part III).
The social cost of delayed reform: revolts and dvil wars
Maintaining an agricultural structure based on relatively inefficient hacienda systems
is cosdy. In addidon to the static efficiency losses" there are dynamic efficiency losses associated
with the reduced profitability of free peasant cultivation and the accompanying lack of incentives to
invest in physical and human capital in the sector. Ihen there are the resource costs used in rent-
seeking to create and maintain the distortions that support the large farms and contribute to rural
poverty and inequality. In a competitive rent-seeking equilibrium these costs are equal to the rents.
The distortions reduce employment in the sector, imposing an equity cost. Finally, the social costs of
failing to reform have often included peasant uprisings and civil war.
Consider Brazil, where the social costs of continued massive distortions in favor of
large farms have been substantial (Binswanger 1991) even without violence. Between 1950 and 1980,
agricultural output grew at a remarkable 4.5 percent a year, land area expanded at 1.5 percent a year,
but agricultural employment grew at only 0.7 percent a year. Over that period, the large-scale farms
evicted most of their internal tenants and workers, many of whom migrated to urban slums or ended
up as highly insecure seasonal workers without farming skills. An alternative growth path based on
13 Quantitaive estimates of this efficiency los amr scare, but Loveman (1976) estimate that Chile could
have saved roughly $100 million a year in agricultual imports during 1949-64 had the 40 percet of land left
Ucutiva  byw l    ndlords be  cutimVated
31



smaller family farms could have provided rural employment and self-employment opportunities for
many of these people and gainfully absorbed a substantial share of the rapidly growing population.
In most cases, protracted and violent struggles have significantly reduced the
performance of the agricultural sector and the economy as a whole. While peasants have rarely been
the initial protagonists in radical class struggles or revolutionary movements many revolutionary
movements took refuge in remote areas of limited agricultural potential - sometimes designated
"communal areas", "reserves", or "homelands"- where peasants have provided both active and passive
support for guerrilla fighters. Many analysts have emphasized the important role of peasant discontent
in incidents of regional and national violence (Moore 1966; Wolff 1968; Huizer 1972; Migdal 1974;
Scopol and Scott 1976; Christodoulou 1990; and Kriger 1991). The losses from such conflicts are, of
course, difficult to measure, but some notion of their magnitude can be gauged from the duration and
intensity of such struggles as these cases show:
*     In Mozambique, peasants escaped from forced cultivation, vagrancy laws, and forced labor to
inaccessible rural areas, which were the main centers of support for the Frelimo guerrillas
from 1961 until independence in 1975. (Isaacman & Isaacman 1983). Land reforms which
were initiated after independence, however, resulted in highly mechanized collective farms
and did not address the problems of the freehold sector. Violence continues to this day.
e     In Zimbabwe large scale eviction of some 85,000 families from European-owned farmlands
during 1945-51, led to a general strike among Africans in 1948 and provided the basis for
peasants' support of ZANU (Zimbabwean African National Union) guerrillas in 1964,
(Mosley 1983; Ranger 1985; Scarritt 1991 and Kriger 1991). Guerrilla fighters took up the
peasants' grievances over unequal distribution of land and state interference with production
and used the TIbal Trust Areas as bases to attack European farms. While a substantial
settlement program after independence provided land to Africans, a number of shortcomings
limited the success of this program (see Binswanger and Deininger 1993). Policy distortions
remained in place despite evidence that large farms are not more efficient than small holder
farmers (Masters 1991) and land reform continues to be a major political issue.
32



*     In Guatemala, communal lands were In effect expropriated in 1879 by a law giving
proprietors three months to register land titles after which the land would be declared
abandoned. Most of the "abandoned" land was then allocated to large coffee growers.
Redistribution attempts in 1951-54 were reversed following a military coup in 1954, when
virtually all the land which had been subject to land reform was returned to the old owner and
farms expropriated from foreigners were allocated in parcels averaging more than 3,000
hectares (Brockett 1984). Since then, there has been a repeated pattern of suppressinn and
radicalization of resistance. Suppression of the cooperative movements of the 1960s led to
formadon of the guerrilla army of the poor (EGP) in 1972, with Its main base in Indian
highlands. Peasants responded to a wave of government-supported assassinations in 1976 with
the formation of the committee for peasant union (CUC) in 1978. Government massacres of
protesting peasants followed (Davis 1983). Almost 40 years after the first attempt at reform,
continuing peasant demonstrations signal the cost of failure.
*     Smatlholder land in El Salvador was similarly appropriated. A decree of 1856 stated that all
communal land not at least two-thirds planted with coffee would be considered underutilized
or idle, and would revert to the state. Communal land tenure was abolished in 1888. Sporadic
revolts led to such countermeasures as the 1888 "security tax" on exports to finance rural
police forces, a 1907 ban on rural unions, and the creation of a National Guard in 1912
(McClintock 1985). Areas where land pressutes were particularly severe emerged as centers
of the revolt of 1932, during which some 10,000 to 20,000 peasants were killed (Mason
1986). Guerrillas promising land and other agricultural reform gained considerable support in
rural areas in particular following tenant evictions in the cotton growing lowlands during
1961-70. Ihese evictions led to a 77 percent decline in the houseplots available to tenants as
the number of tenants dropped from 55,000 to 17,000. Violence continued to escalate until
1979, when reform-minded officers engineered a coup and introduced land reform in an
attempt to preempt a shift in popular support to the FMLN-FDR guerrilla forces. Narrow
eligibility rules sharply limited the number of beneficiaries of land reforms and more than a
decade of civil war ensued. 'he peace accord of 1992 mandates additional land reform.
*     Colombia also demonstrates the perils of incomplete land reform. Conflicts over land between
tenants and large-scale farmers at the frontier escalated from isolated local attacks in the early
33



1920s to more coordinated tenant actions by the late 1920s. While various kinds of reform
legislation were considered during the 1930s, the law finally passed in 1936 vested rights in
previously public lands *ith large landlords rather than the tenants cultivating the land (Le
Grand 1982). A series of tenant evictions followed, leading to a quarter century of violence
(1940-65) during which guerrillas recruited support from peasant groups. Land reform
legislation in 1961 and 1968 regularized previous land invasions but did nothing to improve
the operational distribution of land holdings and far fewer peasants benefitted from the
reforms than had previously been evicted (Zni.osc 1989). Peasant land invasions intensified
during the early 1970s, leading to the declaration of a state of emergency after 1974. Regional
mobilizations, strikes, and blockades flared up again in 1984, indicated that the conflict is not
yet resolved.
Much of the rural support for the Shining Path guerillas in Peru can be traced to the exclusion
of most of the highland Indians from agricultural bUnefits and the benefits of agrarian reform
of 1973 which benefitted primarily the relatively few workers in the coastal area. As a result,
more than half the departments in ae country have become virtually inaccessible to
government forces (McClintock 1984), and public investment in these regions halted, inducing
further economic decline and large-scale migrations to the cities, thus exacerbating social
tensions and conflicts. Poor economic management during the 1980s and continued activity by
Shining Path have led to capital flight and economy wide decline.
Other countries that have experienced prolonged conflicts over land include Angola,
ChUle, and Nicaragua. While the policies that created and maintain dual land ownership distributions
do not necessarily lead to violent struggle - other intervening factors are likely to be important - they
clearly played a significant role in many cases.
PART U.: ANALYTIC CONTROVERSIES
The first question which is central to the analysis of past and future reforms in
grculura land relations is: Are junker estates and large mechanized farms economically more
efficient than smaller, family-operated holdings? The answer is important because if they are not,
34



equalizing the ownership distribution or breakig up collective or state farms Into family farms would
enhance both efficiency and equity. In examining the relationship between farm size and
productivity, we look first at the sources of economies of scale: economies of scaIe in processing
plants that are transmitted to the farm and generate wage plantations, lumpy inputs that cannot be
used below a certain minimum level such as farm machinery and management skills, and advantages
in the credit market and in risk diffusion arising from larger ownership holding (section 4). We then
summarize the empirical findings on scale economies and diseconomies.
This leads to the second central question for land reform: if, as we find, large operational
holdings are usually inefficient, why do large landowners in market economies not rent to family
farmers (section 6)? Tbe rental market has historically been the most important mechanism to
circumvent the diseconomies of scale associated with large ownership holdings despite the incentive
ssues associated with tenancy and sharecropping which are reviewed in Section 6. Yet the history of
land reform shows that long-term rental of entire farms often implies a high risk of loss of land to
tenants, and long term tenancy is no longer an option. Short-term rental of parcels of land cannot
create small family-operated holdings. But if tenancy is no longer an option, what prevents the land
sales markets from bringing ownership holdings in line with the optimal distribution of operational
holdings? Our analysis in Section 5 shows that it is the result of imperfections in other markets,
brought about by land-credit likagtes and policy distortions.
4. Fsrm Size and Productivity
Economies of scale In processing
Plantations have historically been established to produce specialized export crops in
areas of extreme land-abundance and therefore have had to import slaves or indentured labor. But
even after the abolition of slavery or indentured labor, wage plantations survived in selected crops as
highly specialized large ownership holdings using hired labor to produce a single cash crop. Most
workers lived in labor camps on these wage plantations and had no subsistence plots of their own to
cultivate.
35



Labor is the largest component of total costs. Grigg (1974) and Courtenay (1980)
discuss how the ability to use labor nearly year-round favored the organization of production of these
crops under plantations, rather than with tenants or outgrowers. Tree crops such as oil palm, rubber,
and tea have the most even demand for labor. Labor demand is more seasonal for sugar and coffee,
although irrigation (for sugar) or specific processing (for coffee) can help even out demand.
Wage-based plantations continue to exist for tne typical plantatdon crops - sugarcane,
bananas, oil palm and tea because of another technical characteristic. Economies of scale arise from
the processing or marketing stage rather than in the farming operations and are transmitted to the
farm because of the need to process the crops within hours of harvesting (Vnswanger and
Rosenzweig 1986). Only for these crops can wage plantations compete with smallholders without
relying on coercion to acquire labor.
Economies of scale in processing alone are not a sufficient condition for plantations.
The sensitivity of the timing between harvesting and processing is crucial. Easily stored products such
as wheat or rice can be bought at harvest time in the open market and stored for milling throughout
the year. Therefore, the economies of scale in milling are irrelevant for the organization of the farm.
In the case of sugarcane, by contrast, harvesting and processing must be carefully coordinated. If cut
cane is left unprocessed for more than a day, much of the sugar is lost to fermentation. And to keep
the expensive capital stock operating throughout most of the year processing cane into sugar, cane
must be planted at different times of the year, even at times when the sugar yield is not at its
maximum. Independent farmers would be unwilling to plan cane during those times without
compensation. One way to circumvent this problem is for sugar factories to run their own plantations,
with a single manager who decides on the tradeoffs between harvesting cane at suboptimal times and
leaving the capital stock idle. Another way is contracfarming (Hayami 1992; Glover 1990).
Contracting with small farmers is widespread throughout India, Thailand, and elsewhere where
sugarcane was introduced into an existing smallholder system.
Production of bananas is another ecample of the coordination problem. Mature
bananas must be put into a cold boat within 24 hours of harvest, an immense challenge for the
plantation and shipping company. Coordination is required to ensure that the boat will arrive when the
bananas are ready to be shipped and that a boat can be flled when it arrives, - for that reason, some
36



of the world's largest owner-operations are banana companies whose holdings include dozens of
plantations operated by hired managers and workers. In Central America, when legislation made it
more difficult for multinationals to own plantations, the major banana companies increased their
supplies by buying from contract farms. These farms typically have hundreds of hectares and their
contracts are so tight that they virtually remain managed by the multinationals (Ellis 1985).
Similarly, rapid deterioration of the harvested product together with economies of
scale in processing are the main factors leading to the continued cultivation of tea and oil palm on
plantations. Thus the superiority of the plantation depends on a combination of economies of scale In
processing with a coordination problem. Plantations do not arise - or do not survive once labor
coercion is abolished - unless both these conditions exist. Bananas for local and national markets,
which are supplied by individual trucks requiring little coordination, are supplied by family farms all
over the world. Similarly, traditional unrefined forms of sugar such as muscovado in Central
America, where processing did not involve economies of scale, were produced by family farms even
in economies dominated by sugar plantations. In many countries coffee and rubber are also cultivated
undaer smallholder systems. They have lower capital requirements for processing than do sugarcane,
tea, or oil palm, and therefore, have a smaller optimal cultivated area associated with a single
processing unit. Despite their even labor demand over the year, the plantation mode of production
has therefore declined sharply at the expense of smallholder production.
The different outcomes for plantations following the abolition of slavery also support
the combination hypothesis. United States cotton and tobacco plantations which had no coordination
problem, abandoned large-scale cultivation and rented the land out to their former slaves, creating
landlord estates (arrow 17). The same thing happened in Latin America, except that some farms
became landlord estates and some haciendas (arrows 16 and 17). Slave-perated sugar plantations in
the Caribbean and South America, however, converted to wage plantations (arrow 15). There are, of
course, other factors at work as well determining what precise pattern of production relations results
after slavery is abolished. Klein and Engerman (1985) distinguish three patterns according to relative
lad abundance and the presence of government intervention.
Today, wage plantations survive in areas where they were first established under
conditions of low population density and with a large land grant. Where the same crops were
37



Introduced into existing smallholder systems, contract farming prevails. Processors seem not to have
found it profitable to form plantations by buying out smallholders and offering them wage contracts.
This suggests either that the coordination problem associated with plantation crops can be solved at a
relatively low cost by contract farming or that imperfections in land sales markets are so severe that it
is prohibitively expensive to create large ownership holdings by consolidating small farms (section 5).
Lumpy inputs
Draft animals for plowing were the first lumpy input in agriculture. Because of the
difliculty of farming using rented draft animals (Binswanger and Rosenzweig 1984), smdl1 farmers
who lose their draft animals frequently rent out their land until they can acquire new animals (Jodha
1984). Farm machinery - threshers, tractors, combine harvester - are much lumpier than draft
animals. Tractors and harvesters reach their lowest cost of operation per unit area at a much larger
scale than do draft animals, so the optimum operational farm size rises with their introduction. Karl
Marx and his followers believed that the economies of scale associated with agricultural
mechanization were so large as to make the family farm obsolete. Yet small owners can rent out their
land to larger operators (consolidators) rather than sell it, as the ejidatarios in irrigated areas of
Mexico have often done. So the initial economy of scale associated with machines does not imply that
reverse land reform is needed in areas with many small ownership holdings.
Machine rental can permit small farms to circumvent the economies of scale
advantage associated with machines in all but the most time-bound of operations, such as seeding in
dry climates or harvesting where climatic risks are high, where farmers compete for first service and
therefore prefer to own their own machines. ' But threshing can be done at any time of the year and
as in European agriculture in the late nineteenth century, the expansion of stationary threshers in
developing countries today reflects a well developed, efficient rental market. Harvest combines are
often rented in the developed and developing world. Most Midwestern U.S. farmers rent them from
operators who foliow the progress of the harvest season from Oklahoma to Canada. Tractors too are
widely rented out for plowing to small farmers in Asia, Africa, and Latin America, but the markets
14 BinsuwaWr and Rosenzweig (1986) discuss the limits to natal markets impsed by moa hazard snd
msonality.
38



are not as problem free as those for threshers (World Bank 1984). Rao's (1975) analysis of India,
shows that small farms' productivity advantage over large farms initially disappeared following the
introduction of tractors in Northwest India, but once the size of operational holdings was adjusted
upwards, the smaller farms re-emerged with higher productivity rates.
Tbus, the economies of scale associated with machines increase the minimum efficient
farm size, but by less than expected because of rental markets. The use of draft animals and machines
- lumpy inputs - leads to an initial segment of the production function that exhibits increasing returns
with operational scale, but these technical economies would vanish when farm size is increased by
replicating the optimal scale of lumpy inputs or when rental markets make the lumpiness of machines
irrelevant. Under constant technical returns to scale and with perfect markets for land, capital, and
labor, the ownership-distribution of land would be irrelevant for production and would only affect the
distribution of income. Landowners would either rent the necessary factors of production (abor and
capital) and make zero profits operating their own holding or, if there were transaction cost. in the
labor market, rent in or rent out land to equalize the size of operational holdings.
Management skUls like machines, are an indivisible and lumpy input, so the better the
manager, the larger the optimal farm size. Technical change strengthens this tendency: fertilizers and
pesticides - and arranging the financing to pay for them - require modern management skills. So does
the marketing of high-quality produce. In an environment of rapid technical change, acquiring and
processing information becomes more and more important, giving managers with more formal
schooling and technical education a competitive edge in capturing the innovator's rents.
Therefore, optimal farm sizes tend to increase with more rapid technical change.
Some management and technical skills, like machinery, can be contacted from specialized consultants
and advisory services or provided by publicly financed extension services. Contract farming often
involves the provision of technical advice. But key farming decisions and labor supervision cannot be
bought in a market. So limits on management skills will lead to an upward sloping segment in the unit
cost curve as operational holding size increases.
Access to credit and risk diffusion
39



Land, because of its immobility and robustness, has excellent potential as collateral,
making access to credit easier for the owner of unencumbered land (the issue is discussed In detail in
section 5). Rural credit markets are difficult to develop and sustain. There is therefore severe
rationing, which can be partly relieved by the ability to provide land as collateral. The high
transaction costs of providing formal credit in rural markets implies that the unit costs of borrowing
and lending decline with loan size. Many commercial banks do not lend to small farmers because they
cannot make a profit. Raising interest rates on small loans does not overcome this problem, since it
eventually leads to adverse selection (Stiglitz and Weiss 1981). For a given credit value, therefore,
the cost of borrowing in the formal credit market is a declining function of the amount of owned land.
Land ownership may serve as a sign of creditworthiness in informal credit markets as well.
Access to credit is particularly important in developing countries because they usually
lack other intertemporal markets to insure against crop or price risks. Insurance is sometimes
available for very narrowly defined specific risks such as hail or frost, but only for very large farms.
Forward markets are often banned or discouraged by policy intervention. An interested local insurer
would have enough information to overcome the moral hazard problem, but the covariance of crop
yields makes the risk uninsurable at the local level. A national insurer could overcome the covariance
problem, but lacks the local information to overcome the moral hazard problem. The absence of a
market for multi-risk crop insurmce is the result of the combination of moral hazard and the local
covariance of production risk. The absence of crop insurance and forward markets confers special
importance on access to credit as an insurance substitute, but the combination of covariance and moral
hazard also sharply reduces the potential of financial intermediation in rural areas (Binswanger and
Rosenzweig 1986).
Providing funds to overcome emergencies is a common function of informal rural
credit markets. But the amounts small farmers can borrow for consumption are usually tiny - and
often at high interest costs (Binswanger 1985; Christensen 1989; Morooka and Hayami 1990; Udry
1990; Deaton 1991). Investigations into how farmers and workers cope with disaster show that credit
finances only a small fraction of their consumption in disaster years (Jodha 1978). Access to formal
commercial bank credit therefore gives large modem commercial farms a considerable advantage in
risk diffusion over small farmers without such access.
40



Farmers and workers with little or no access to credit can attempt to diffuse their risk
by relying on accumulated reserves and wealth, social relationships, and risk-sharing arrangements in
land, labor, output and input markets (Jodha 1978; Bidinger and others 1990; Rosenzweig 1988;
Deaton 1990; Sharp 1990). Wealthy individuals can self-insure much more easily than the poor both
directly, as a consequence of their wealth, and indirectly, because geographically dispersed social
networks on which they can rely in years of (ocally covariate) poor harvests. Wealthy farmers should
therefore be better able to accumulate profit-maximizing portfolios than poorer farmers, giving them
an allocative efficiency advantage."5 In land-scarce environments, the bulk of a farmer's wealth is in
the form of land, so large ownership holdings are correlated with a better ability to diffuse risks
through the wealth effect and land's robustness as collateral for credit. Forescano (1969) suggests that
in high risk environments, the superior ability of land-rich individuals to diffuse risk through storage
and better access to credit markets might have been an important reason that otherwise unprofitable
demesne cultivation survival in the face of competition from family farms.
Evidence on farm size - productivity relationship
The literature demonstrates that imperfections in a single market would not be
sufficient to introduce a systematic relationship between farm size and productivity per unit of land.
For example, if credit is rationed according to farm size, but all other markets are perfect, land and
labor market transactions will produce a farm structure that equalizes yields across farms of different
operational size. But if there are imperfections in two markets, land rental and insurance, or credit
and labor, a systematic relationship can arise between farm size and productivity.
Srinivasan (1982) has shown that under conditions of fixed farm size (no land rental) and no
insurance, uncertainty and risk aversion can lead to an inverse relationship between farm size and
productivity, provided that absolute risk aversion does not increase and that relative risk aversion does
not decrease with wealth. With credit and labor market imperfections, the relationship is not
necessarily inverse. For example, Feder (1985) and Carter and Kalayan (1989) demonstrate that
with certain model parameters, the combination of credit and labor market imperfections can lead to a
is As explained in Binswanger and Rosenzweig (1986), they are not able to provide icsran  to small
farmers bocause covunance of inome would require large reserves in order to be able to offer craedible
conrct.
41



U-shaped relationship. Eswaran and Kotwal (1985) obtain an inverse relationship by adding a fixed
cost of production to labor and credit market imperfections. Generally, the presence of multiple
market failure can explain a variety of farm size distribution and productivity structures.
The implications of imperfections in labor, credit, and land markets are illustrated by
Feder (1985) whose model is replicated in Appendix 2. By assumption, the efficiency of hired labor
depends on the intensity of supervision by family labor, implying that family labor and hired labor are
complements and that the amount of labor effort or 'efficiency' units supplied increases with
supervision.
If credit and land rental markets are perfect, the supervision constraint alone would
lead each household to lease in or lease out the amount of land required to maintain a uniform ratio of
family labor endowment to operated area. The ratio of effective labor input to operated area would be
constant for all cultivators, whatever the distribution of land ownership. No farm size-productivity
relationship would exist.
But if there is a binding constraint in the credit market whereby the supply of working
capital depends on the amount of land owned, the optimal size of the operational holding would vary
systematically with size of the owned holding even if land rental markets were perfect. The magnitude
(and direction) of this variation would depend on the relative elasticities of output with respect to
effective labor and of labor effort with respect to supervision.
Now, if, in addition to a supervision constraint and a credit constraint, there are no
rental markets for land - whether by law or because of the threat of land reform - a negative relation
between farm size and land productivity is likely to emerge. Of course, the capital cost advantage of
large farms does not necessarily lead to higher investments on the farm if the capital can be invested
elsewhere in the economy at higher returns than in agriculture.
The Evidence for Diseconomnes of Scale
The discussion thus far suggests several approaches to the measurement of the farm
size-productivity relationship:
42



*     Since the supervision costs vary with the operational holding size while the capital constraint
is related to the ownership holding size, the separate effects of operational and ownership
holdings should be distinguished in any test of the farm size-productivity relationship. To
eliminate errors resulting from the raw correlation of farm size and household size,
regressions of an efficiency indicator on operational and ownership holding size should also
include the number of adult family members who can act as supervisors. None of the existing
studies has taken full account of these distinctions.
*     Proper measures of relative efficiency are the difference in totalfactor productivity between
small and large farms and the diference in profits, net of the cost offamily labor, per unit of
capital Invested. Using market prices to measure productivity assesses differences in private
efficiency. Using social opportunity costs as a measure eliminates the impact of distortion and
measures differences in social efficiency. Few studies have made this distinction.
*      Most of the literature has analyzed physical yields of specific crops or the value of
agricultural output per unit of operated area. These are not relevant measures of overall
private or social efficiency since they are but partial productivity indices that do not take into
account differences in input and labor use. Because part of the adjustment to incentive
problems and other market imperfections is to vary the output mix so as to save on the factors
with the highest scarcity value in the specific farm, focusing on a single crop is inappropriate
except in monocrop farming systems. Individual crop studies are therefore not relevant to the
farm size-productivity relationship problem.
*      Normalizing any productivity measure by total land area or regressing it on land area raises
severe measurement problems because agroclimatic potential and land quality differ across
regions. Ihe same problem afflicts any comparisons that involve pooled data or use the means
from several regions (e.g., Thiesenhusen 1990; Deolalikar 1981). Land quality differences
within regions are often so large that adjustments must be made for those differences if
productivity is measured per unit area rather than per capital invested (Bhalla and Roy 1988).
Only if there is no correlation between land quality and farm size is such an adjustment
43



unneeswary16 - or if the differences arise from farmer investments in tubewells, land
levelling, drainage, or the like.
The following test of the farm size-productivity relationship is one way to take these
considerations into account describing not a causal relationship but a multiple correlation:
PIK = g(OPOW,H,Z) with exected signs g8 <,g2> Ogy>O, (1)
Where K is assets, L is labor, P is private or social profits net of private or social cost of
family labor, OP is operated area or value of operated land, OW is owned area or value of
owned land, H is the number of household workers, and Z is a vector of exogenous land
quality, distance from infrastructure, and exogenous land improvement variables. g, should
be negative because of rising supervision costs. g2 should be positive because ownership
provides better access to credit. And & should be positive because family members have
incentive to work and can supervise.
None of the studies of the farm size-productivity relationships have applied this full
specifications and few studies have even looked at total factor productivity or farm profits net
of the cost of family labor. So we must be content to summarize the findings of farm-level
studies within small regions that look at value of output per operated area. Typical findings
are presented in table 2, which is extracted from Berry and Cline (1979) and similar results
are found in a range of other studies.'7
"Both distress sales (Bhagwati and Chakravarty 1969) and differential patterns of invesmet (Sea 1964)
coud explain theoretically why small farmers could systematicaly end up with higher quality land within a
given villae. Few empirical studies exist at a sufficiently disggregated village level to confirm this asociation
For six villages in semi-arid India, Walker and Ryan (1990) reject the existence of a systemtic association
betwe  farm size and land quality.
1  For SiX Lain American countries Lu and YotopoUlos 1971, and 1979, Barralough and Collarte 1973; for
nortesten Brai Kutcher and Scandizzo 1981 ; for fifteen countria in Africa, Asia, and Latin America
Comia 1985; for the Indian Punjab Sen 1981; for India and West Bengal Cater 1984; and for Ildia
disggrgated into seventy-eight agroclimatic zones Bhalla and Roy 1988. Dyer 1991 descnbes the army of
intument used by large produce  in Egypt to increase their competitiveness with small farmers,
deostaing that large producers can successlly lobby for meaure to counteact the inverse fam-size
productivity relationship. Te need for such rent-seelng implies the coniued validity of this relationship
although Dyer interpret it to mean the opposite.
44



TABLE 2: Farm-size productivity differences, selected countries
Fam size                        Northuat Brazil          Punjab, Pakistae          Muda, MaLkydi
Small farm                          563                        274                     148
(heatr)                         (10.0-49.9)                 (5.1-10.1)               (0.7-1.0)
Larva fm                             100                       100                     100
(hectarft)                         (SOO+)                    (20+)                   (5.7-11.3)
Note:       *100 - bqet fam size compared with second smallest farm size. Second smallest fam size
used in calculations to avoid abnormal productivity results often recorded for the smallet plots.
lable 4-1. Northeastern Bazil, 1973; Production per Unit of Available Land Resource, by
Farm Size Group, p.46. Index taken using average gross receipts/areas for size group 2 (small)
and 6  arge), averaged for all zones excluding zone F, where sugarcane and cocoa plantations
skew productivity average for argo farms.
able 4-29. Relative Land Productivity by Farm Size: Agricultural Census and FABS Survey-
based Estimates Compared, (1968-9) p. 84. Index taken using value added per cultivaed acre for
second smallest size roup and luagest
'Table 4-48. Factor Productivity of Muda River Farms by Size, Double Croppers, 1972-3 p.
117. Index taken from value added in agriculture/relong (0.283 ha = 1 relong).
Soue:       Berry and Cline (1978)
Those studies support the following generalizations:
*     Ihe productivity differential favoring small farms over large one increases with the differences
in size. That means it is largest where inequalities in land holdings are greatest, in the
relatively land-abundant countries of Latin America and Africa, and smallest in land-scarce
Asian countries where farm size distributions are less unequal.
*     The highest output per unit areas is often achieved not by the smallest subfamily or part-time
farmers but by the second-smallest farm size class, which includes the smallest full-time
farmers. This suggests that the smallest farms may be the most severely credit constrained.
*     Plantation crops as represented by sugarcane production in Brazil, do not exhibit a negative
farm size-productivity relationship (Cline 1971; Kutcher and Scandizzo 1981).
*     When land is adjusted for differences in quality using land value or exogenous land quality
measures, the negative productivity relationship weakens but does not disappear, especially
where it is very large.
45



*     Introduction of the green revolution technology in India led to a weakening but no the
disappearance of the raw productivity differentials (Bhalla and Roy 1988).
Three studies came closer to the specification in equation 1. For the Muda River region
of Malaysia, Berry and Cline (1979) found that value added per unit of invested capital for the second
smallest farm size group exceeded that of the largest farm size group by 65 percent, more than the
difference in value of output reported in table 2. The use of value added adjusts for costs of
purchased inpUts, but this measure is still likely to bias the test in favor of small farms to the degree
that small farms use labor more intensively than do large farms. But since the result holds for raw
output, the negative relationship would probably hold as well if the test were based on net farm
profits. The results suggest that well-developed rental markets, as in the Muda area for tractors and
threshes, enable small farmers to circumvent the economies of scale associated with tractors, leaving
labor supervision costs to dominate."
In the second study, Berry and aine (1979) first split the data for Northeast Brazil (see
table 2) into agroclimatic zones, which sharply reduced the observed negative relationship. 'Social'
profits were then calculated by imputing a real opportunity cost of 15 percent to capital and valuing
family labor at 0, 50 and 100 percent of the minimum wage, a wage rarely paid in agriculture. Even
when family labor is valued at the full opporuity wage, social profits are dearly higher by 23 to
150 percent for the second smallest farm size group (10 to 50 hectares) than for the second largest
and the largest farm size groups (200 to 500 hectares) in four of six non-sugar growing zones. For
the two zones where the relationship does not hold as clearly (Bahla and Sertao), the weakness of the
results appears to be due to paucity of obsvations (Kutcher and Scandizzo 1981). The negative
' Only a few studies explicitly test for the sepambility of fimily and hired labor. Pitt and Rosenzweig (1986)
show for a sample of Indonesian fannmes that profits are independent of the short-tm health status of the
household head, but since short-tem illne does not interfere with supervision the result says little about
whether wage labor can complement family labor on a pmnet bi Deolaliar and Vijverberg (1987) reject
the hypothesis of perfect substitutabiltybetween family and hired labor based on samples from ndia and
Maaysia, but beas they estime a pduction function using coss-ction data, atistical problems vitate
their findings. Benjamin (1992) estimates a demand fimction for aegate labor services. He rejects the
hypothei of nonseparability for Indneian rice fam  aon the basis of the joint lack of significane of
demogaphic variables. Since his model includes ar  havested a a dependet variable, it does not allow for
adjustmens of area operated (via rt) in response to family siz. In effect, then, the model meures only the
conditional impact of demographic variables, given opeatod  ea, on the demad for hired labor. The fact that
ea opeated (which, has sigificant influence on labor demand) s correlated with family composition suggests
that a strong supervsion constraint might be found if the unconditional effect were considered.
46



productivity relationship still holds in the technologically advanced Agrt .. region, where
mechanization was most pronounced if social profits are considered.
In the third study, Rosenzweig and Binswanger (1993) estimate a profit function similar
to equation (1) which include total assets, the composition of the asset portfolio, family labor,
education, age, and the onset date of the monsoon. They use the complete ICRISAT panel data from
ten villages in high-risk semi-arid India to estimate a model that allows for separate testing of
technical economies of scale on the one hand and the impact of supervision cost advantages of poorer
farmers relative to the capital cost and risk diffusion advantages of wealthier farmers on the other
hand. Fixed-effects estimation techniques were used to eliminate problems of land quality
differences. The results reject the hypothesis that the composition of investments reflects technical
scale economies. They support the hypothesis that the asset portfolios of farmers are signifcantly
affect by farmers' risk aversion, wealth, and the degree of monsoon onset variability (a measure of
weather risk). In an environment of slowly changing technology, the profitability of the portfolio is
not affected by formal schooling, but it does rise with age, a proxy for experience. Profits (net of
their wage costs) also increase with the number of adult family members, suggesting that their
contribution arises from their management and supervision function.
Rosenzweig and Binswanger also estimate the impact of weather risk and wealth on the
risldness and profitability of farmers' asset portfolios. Figure 3 plots the profit per unit of asset for
four wealth classes as a function of rainfall variability (onset of the monsoon). The profit rate of
farmers at the eightieth percentile of wealth is insensitive to Increases In weather risk, suggesing that
they are confident enough in their ability to diffuse risk through credit, savings, or social relationships
that they do not need to choose portfolios that reduce risk up front at some cost in profits. Farmers in
the 20th percentile, however, sharply reduce the profitability of their portfolios as rainfall risk rises.
Despite these portfolio adjustments, this high risk environment with relatively little
mechanization and slow technical change, the smaller farm size groups have higher profits per unit of
wealth at all levels of rainfill risk observed in the data The supervision and labor cost advantages of
family labor are apparendy greater than the advantages that the lumpiness of management skills ad
machines and the better access to credit and other risk-diffusion measures confer on large farms. Only
in the most risky environments does the advantage of the poorer farmers nearly disappear.
47



Figure 3.
Profit-Wealth Ratios and Weather
Variability, by Wealth and Class
034-
029 -                                             *_
0-24 _                                                         -_-_
19- .'*
0I09 1
9-6      11-6     13-6     156      17-6     19-6     21-6     23 6
Monsoon onset standard deviation (weeks)
3. Profit-wealth ratios and weather variability, by wealth class. Pcrcentiles:    , 20
--,  40th; --, 6oth;---, 8oth.
Note:      The onset date of 4p.e Monsoon was the single most
powerful of eight different rainfall characteristics to explain
gross value of farm output.
48



Using a nonparametric approach to estimate a production function for Wisconsin farmers, Chavas and
Allier (1993) study farms in a very modern and dynamic environment. They flnd virtually no scale
economies in dairy production and only very limited initial scale economies due to lumpiness of
Inputs.
Condusion
Most of the empirical work on the farm size-productivity relationship has been flawed by
methodological shortcomings, and has failed to deal adequately with the complexity of the issues
involved. Studies that come to grips with some of the shortcomings and use a more refined
measurement of land quality and a productivity variable Instead of simple yields find that even in
fairly technologically advanced and mechanized areas, such as the Muda scheme in Malaysia or the
Agreste region of Northeast Brazil, small farms retain a productivity advantage over large farms. This
finding suggests that rental markets can substitute to a certain degree for the indivisibility of machines
and some management skills. The methodologically sounder study based on the ICRISAT data
confirms both the mechanisms leading to differential performance by scale and the superiority of
smaller farms in an environment with litde mechanization and slow technical change. While there is
evidence on the negative relationship between farm size and production, more work is needed on this
subject. Such work should follow the lines sketched out in equation 1, using recent farm level data
for developing coutry regions with high quality agroclimatic and soil conditions, substantial
mechanization, and dynamic technical change.
S. The effects of land-credit Unks and policy distortions on land sales markets
The farm-size productivity studies indicate that for given technology, factor prices, land
quality, and farming sklUls there is an optimal operatdonal holding size at which the disincentive costs
of adding more workers filly offsets the economies of scale from lumpy inputs, access to credit and
management skills. Taking into account differences in hrming skills and land quality, this finding
translates into an optimal distribution of operational sizes. For any given distribution of ownership
holdings, one would expect tenancy and land rental markets to bring the distribution of operational
holdin  close to that optimal distribution. f incentive problems associated with tenancy are minor
49



and can be ignored, the distribution of ownership holdings would be Independent of the distribution of
operation holdings, since large landholders would simply rent out their land with no loss in efficiency.
But if legal restrictions on tenancy make this option infeasible or unprofitable we need to
ask whether the sales market will bring about a more nearly optimal distribution of ownership-
operational holdings - that is, whether it will be profitable for the owners of large and relatively
unprofitable farms to split them up and sell them to small family farmers. Covariate risk, imperfect
Intertemporal markets, and policy distortions affecting the functioning of the land sales market will
prevent this market from achieving a first-best solution. But increases in efficiency arekstill likely to
result from sales transactions that transfer land from bad to better managers.
Covariate risks and Imperfect credit markets
Land is often a preferred store of wealth, so with imperfect inter-temporal markets the
utility derived from landownership will exceed the utility derived from farm profits. Its immobility
makes land a preferred form of collateral in credit markets which confers additional utility from
landownership, especially in an environment where production risk cannot be insured.
'he collateral value of land and the high positive correlation of incomes in a given area
imply that there would be few land sales in periods of normal weather."' Landowners would be
made better off by selling land only if they could ear a higher return from the sales proceeds than
from cultivating or renting out the land. So, where non-agricultural investment opportunites for ru
residents are limited and national credit markets are underdeveloped little land will be supplied for
sale in normal years. Tbe number of bidders for land is constrained by the level of household savings
since mortgaging the land would be unprofitable. Because land has collateral value, its equilibrium
price at given credit costs will always exceed the present discounted value of the income stream
produced from the land. Mortgaged land, however, cannot be used as collateral for working capital,
so the owner does not reap the production credit advantage and thus will be unable to repay the loan
out of increased income from the land. With imperfect insurance markets, only unmortgaged land
1 Such paucity of lad sales is also observed in developed countres where land sales makets arn usually
very thin.The perete of rnmland transfend an avemg each year is 3 % of the total in the US, 1-1.5% in
Bditain, 1.5% within tho white sector in South Africa, 0.5% i Ireland and Kenya (Moll 1988:354).
50



yields a flow of income or utility, the present value of which equals the land price. As discussed, if
land ownership provides access to credit and helps in risk diffusion, the buyer has to compensate the
seller for the utility derived from these services of land (Feder and associates 1988). Since only
unmortgaged land provides these services, a buyer relying on credit cannot pay for the land out of
agricultural profits alone. Thus land sales are likely to be fianced out of household savings, so that
the purchased land can be used as collateral for working capital. This need to purchase land out of
savings tends to make the distribution of landholdings more unequal, despite the greater utility value
of land to smaller owners arising from its insurance value and their lower labor costs.
Spatial covariation in yields suggests that in particularly good crop years, when savings
are high, there would be few sellers and many potential buyers of land. Good years are thus not good
times for land purchases. In bad crop years, farmers would have litdte savings with which to finance
land purchases. And in particularly bad periods - say after consecutive harvest failures - moneylenders
would be the only ones in the local rural economy with assets (their debt claims). Moneylenders
would prefer to take over rather than sell the landholdings offered as collateral by defaulters since the
price of land would be lower than average in bad years. So, in bad crop years land would be sold
mainly to moneylenders as distress sales, or to individuals with incomes or assets from outside the
local rural economy. We should expect, then, that in areas with poorly developed insurance and
capital markets, land sales would be few and limited mainly to distress sales. Results from India and
Bangladesh confirm this hypothesis. Farmers in India experiencing two consecutive drought years
have been found to be 150 percent more likely than other farmers to sell their land (Rosenzweig and
Wolpin 1985).
The implications of different mechanisms to insure against risk on distress sales and the land
ownership distribution are demonstrated by a comparison of the evolution of ownership holdings from
about 1960 to 1980 for predominantly agricultural villages in India and Bangladesh (Cain 1981).
These villages faced very high environmental risks but were characterized by distinct differences in
mechanisms of risk-insurance: In Mharashtra, India, an employment guarantee scheme operated
throughout the period and attained participation rates of up to 97 percent of all households during
disasters. Such schemes were absent after the major flood episodes in Bangladesh. With other
Insurance-mechanisms either absent or exhauted, 60 percent of land sales in Bangladesh were
undertaken to obtain food and medicine. Downward mobility affected large and small farmers equally,
Sl



suggesting that even large farmers had insufficient possibilities to diffuse risks. 60 percent of the
currently landless had lost their land since 1960 and the Gini coefficient of landownership distribution
increased from 0.6 to almost 0.7. This contrasts sharply with the Indian villages where land sales for
consumption purposes accounted only for 14 percent and were incurred mainly by the rich to meet
social obligations. 64 percent of land sales were undertaken in order to generate capital for productive
investment (digging of wells, purchase of pumpsets, children's education and marriages), leading to
an equalization of the land-ownership distribution in India, and suggesting that the poor were not only
able to avoid distress sales, but actually could acquire some la,' as rich households liquidated
agricultural assets to be able to pursue non-agricultural investment.
Historically, distress sales have played a major role in the accumulation of land for large
manorial estates in China (Shih 1992) and in early Japan (Takekoshi 1967) and for large landlord
estates in Punjab (Hamid 1983). The abolition of communal tenure and the associated loss of
mechanisms for diversifying risk are among the factors underlying the emergence of large estates In
Central America (Brockett 1984).
Moral hazard, covariance of income, and collateral value of land imply absent insurance
and imperfect credit markets. In such environments, land sales markets are likely to become a means
for large landowners to accumulate more land. Even where markets for labor, current inputs, and
land sales and rentals are perfectly competitive, weak intertemporal markets for risk diffusion may
therefore prevent land sales markets from bringing about pareto improving trades and an efficient
farm size distribution - an illustration of the theorem of the second best.
The Impact of policy distortions
The existence of common policy distortions intensifies the failure of the land sales
market to distribute land optimally. Consider an idealized case of competitive and undistorted land,
labor, risk and credit markets. The value of land for agricultur  use would equal the present value of
agricultural profits capitalized at the opportunity costs of capital. If the poor have to borrow to buy
land at its present value, the only income stream available for consumption is the imputed value of
family labor. The remaining profits go to pay for the loan. If the poor can get the same wage in the
labor market, they are no better off as landowners than they would be as wage-laborers. Ard this
52



example assumes ideal conditions, with the poor paying the same interest rate as most creditworthy
borrowers.
Anything that drives the price of land above the capitalized value of the agricultural
income stream thus makes it impossible for the poor to buy land without reducing their consumption
below the level of their potential earning in the labor market.
The most important factors and distortions are the following:
*    With populations growing and urban demand for land increasing, the price of land is expected
to appreciate, and some of this real appreciation is capitalized into the current land price.
Robinson and associates (1985) find much higher imp'icit rates of return (cash rents to land
values) to farming in predominant agricultural states in the United States than in states where
nonagricultural land demand is high. The impact of closeness to urban areas on agricultural
land prices is well known. Since these returns are realized only when the property is sold, the
only way a poor person could tap into that income stream is by regularly selling off a small
parcel of land to pay the interest costs - hardly a feasible option for small landowners.
*    In periods of macroeconomic instability, nonagricultural investors may use land as an asset to
hedge against ination, so that an infation premium is incorporated into the real land price. If
epected inflation is fully reflected in interest rates, inflation alone will not affect agricultural
land prices (Feldstein 1980). But if inflation is higher than expected interest rates, and if land
is perceived to be no riskier than alternative assets, excess demand for land will increase the
price of land as a speculative asset. Inflation and changes in real reurns on alternative uses of
capital are the main factors explaining changes in land prices for the United States (Just and
Miranowsld 1989). For Iowa, in addition to fundamentals, (the present value of the discounted
future income stream), an additive fad term closely associated with expected inflation has a
significant impact on land prices (Falk 1991). In a simulation using results of econometric
Although overvaluation due to mispeetion - bubbles - would lead to observationally equivalent
predictions, myopic behavior on the part of land purchasers seems a more satisfactory explanation. On the
poubilities for rational bubbles see Asako (1991) and Diba and Grossman (1988). Empirical and experimental
evidee   an bubbles is provided buy DeLng and Shlcifer (1991), Smith and associated (1988), and Evans
(1986).
53



estimation for Brazil (1966-89) Brandao and Rezende (1992) find that six percent of the
Increase In land is attributable to credit subsidies, 28 percent to macroeconomic instability
(inflation).
*    Credit subsidies are capitalized into land values, as shown in the Brandao and Rezenda (1992)
study and by Feder and Associates (1988). For the U.S., Shalit and Schmitz (1982), show that
most of the increasing debt on farm real estate during 1950-78 was translated into higher land
prices, whereas farm income increases had a much smaller impact.21 Even where there are
no credit subsidies, large landowners have a transactions cost advantage in securing credit,
which is capitalized into land values and may even block access to mortgage credit altogether
for small borrowers altogether.
a     Many countries exeempt agricultural income from income tax, and even where there is no
general exemption, agricultural income is de facto subject to lower tax rates. These preferences
will be partly or fully capitalized into land values. Since the poor, pay no taxes and so cannot
benefit from the tax break, they do not receive the corresponding income stream. Any other
subsidies or tax preferences for large farms similarly increase the difficulty the poor have in
buying land.
Where any of these factors push the price of land above the price justified by the
fundamentals of expected agricultural profits in the absence of distortions associated with farm size,
the poor have difficulty buying land. Even If they are provided with credit on market terms that
difficulty persists unless their productivity advantage from lower supervision cost is very large. Of
these factors, nonagriculturl demand, inflation, credit constraints, and credit subsidies have been
investigated empirically; income tax preferences for agriculture have noL Most of the empirical
studies concentrate on the United States since the paucity of land transactions in developing countries
21 Whiloe tis dem  ate the sgnificane of the polcy and institutional enirment in aggregae models,
icroconomic evidence on the impostnce of credit ratining on land prices is limited. Carter (1989), Carter
and Kalfaya (1989), and Cart and WVibe (1990) use a roughly calibrad model to determine the reservation
price of lnd as a function of frm-sz and obtain a U-shaped cumre. Becase of the roughness of their dat,
the resuts indicate orders of magnitude rather than exact figpues, but they are certainly in the approprate
dirton.
54



makes research difficult (Melichar; four other studies; Hallan et al; Barhema). More work needs to be
done.
6. Incentives, land-credit Unks and land rental markets
As long as there are imperfections and/or distortions in other markets, land sales
markets are unlikely to bring a skewed distribution of land ownership holdings closer to an optimal
distnrbution of operational holdings. The question, then, is whether land rental markets can increase
efficiency by improving the access of the poor to land under conditions in which they can not buy
land.
Land tenancy markets might not increase efficiency if tenants lack incentives to invest in land
improvements, to work hard, or to apply sufficient inputs. These problems will be particularly severe
under sharecropping aangements, with the tenant receiving only a share of the marginal product of
the inputs (the Marshallian inefficiency). Quantitative measurement of the inefficiency associated with
share contracts in different environments is nceay to determine the importance of such
disincentives. The empirical discussion shows that the inefficiencies of share-cropping, measured at
the farm level, are not large.
Despite the disincentives associted with tenancy and sharecropping their widespread use
all over the world suggests that, in an enviroment, where capital constraints and risk considerations
make fixed rent tenancy contracts infeasible, share rental contracts may in fact emerge as efficiency
enhancing, especially if the incentive problems associated with them are low. Since both the
theoretical literature (Otsuka and Hayami 1988; Otsuka, Chuma, and Hayami 1992) and the empirical
literature (Bell 1988) have been reviewed recenty, the discussion here is brief.
Choice of contract and the inetive problem
In the basic model of land-leasing, renting out land under a fixed-rent or share coneract
or employing wage labor are substitutes along a continuum on contractual choices (Otsuka, Chuma
and Hayami 1992). The landlord maximizes income by choosing the number of tenants, the fixed
55



payment and the output-share parameter subject to the constraint that tenants achieve their
(exogenously given) reservation utility. The tenant determines the level of effort that will maximize
utility, yielding an effort-reaction function.
The basic model consists of a constant returns to scale production function Q = OF(e,h)
where Q is output, e is effort, h is number of tenants, and 0 is a stochastic element. The landlord's
income is y = h[l-&)Q - pl, and the representative tenant's income is Y=caQ + p. The fixed rent
contract is obtained for (ar= 1,p<O), the pure wage contract for (a=O,0>O); and {0<a<< 11 with
arbitrary p (often assumed to equal zero for simplicity) characterizes the share contract (Otsuka and
Hayami 1988).
Under conditions of certainty and the rather unrealistic assumption of perfect
enforceability of effort, all contracts lead to equivalent outcomes and the choice of contract type does
not matter (Cheung 1967). If the assumption of perfect effort enforceability is dropped, the tenant
receives only a fadicn a of their marginal product for all but the pure cash rental contract.
Therefore, with effort unobservable and under conditons of certainty (or equivalent, risk neutrality),
the fixed-rent contract clearly dominates the fixed-wage and the share contract and will always be
chosen in equflibrium (Otsuka, Chuma, and Hayami 1992). Given supervision costs for workers or
sharecroppers any type of contract other than fixed rent would result in an undersupply of effort by
the tenant or worker, which would lead to lower total production.
With risk aversion and uncerinty, a share contract provides the possibility of pardy
insuring the tenant against fucations in output. Where intertemporal markets are weak or
unavailablo, there would then be a trade-off between the risk-sharing properties of the fixed-wage
contract under which the worker's residual risk is zero, and the incentive effects of the fixed-rent
contract, which is associated with the optimal supply of effort. (Note that with effort as the only
variable input, effort supply completely determines total production.) Under these empirically relevant
conditions of risk aversion and uncertainty, the one-period contract can achieve only a second-best
solution since increases in the output share parameter above the second-best equilibrium value, while
increasing expected production, are no longer Pareto superior since they lower the risk-averse tenant's
utility by exposing him to greater uncertminty.
56



Recasting the problem in a multiperiod context and allowing for reputation effects,
however, provides options for bringing this second-best opdmum closer to the firstbest outcome.
Otsuka, Chuma, and Hayami (1992) discuss the conditions under which, In a multiperiod context, the
threat of loss of reputation will prevent the landlord from cheating, and so the fixed-rent contract will
tend to dominate the fixed-wage contract as it does in the certainty case just described. They argue
that in relatively closed villages, such Implicit enforcement by the community may be strong enough
to bring the inefficient outcome under the unenforceable contract closer to the first-best outcome even
if risk is present. This conclusion is consistent with the empirical observation that fixed-wage
contracts are found only where the institutional environment discriminates against tenancy contracts
(see section 3 and below) but is inconsistent with the overwhelming prevalence of share-cropping
relative to fixed-rent tenancies.
Choice of contracts and factor market constraints
There is considerable theoretical justification and evidence (Bliss and Stern 1982; Pant
1983; Nabi 1985; Binswanger and Rosenzweig 1986; Skouflas 1991; Shaban 1991) that where
markets for lumpy inputs such as management skills and draft animals are imperfect, households seek
to achieve the optimal operational holding size through land tenancy contracts. Given the nontradable
factor endowment, land rental would be expected to increase efficiency if a fixed-rent contract is
chosen. The relevant question is whether share contracts would increase efficiency as well.
A limit on the working capital available to the tenant (or to landlord and tenant) because
of imperfection in the credit market, can lead to the adoption of a share contract as the optimal
solution to the bargaining problem. Laffont and Matoussi (1981) show that a worldng capital
constant imposes limits on the share parameter a that may make the first-best fixed-rent contract
unfeasible. Their model implies positive correladon between the tenants' working capital and his
output share a. A positive correlation between working capital and output in the share contract but
the absence of such an effect in the fixed rent contract would indicate presence of an incentive
problem. Consideration of the joint working capital available to tenant and landlord does imply a
negative relationship between the landlord's working capital and the tenant's share. All of these
predictions are confirmed empirically for a set of data from Tunisia. his direct dependence of the
opdmal choice of contract on the working capital available to both landlord and tenant may account
57



for the coexistence of a variety of contracts in the same environment among parties with roughly
similar risk aversion but different endowments of working capital.
Thus the main reason that nterlinked contracts and cost-sharing arrangements are so
common may be that they implicitly provide the credit or insurance a tenant needs in an environment
where credit and insurance markets are imperfect (Otsuka, Chuma and Hayami 1992). The traditional
interpretation that these interlinkages are devices used by landlords to bring the second-best
equilibrium closer to the first-best outcome by increasing the tenant's supply of effort (Braverman and
Stiglitz 1982; Mitra 1983; Braverman and Srinivasan 1981) requires strong assumptions that are
generally not satisfied in developing countries (Otsuka, Chuma, and Hayami 1992).
A tenant may be able to meet only part of his (working) capital requirements in the
credit market because of the limited suitability of unharvested crops as collateral - and at higher
interest rates than the landlord would get by offering land as collateral. The landlord is in a better
position than other financial intermediaries to provide a tenant with implicit credit and actuarially fair
insurance because of economies of scope in supervision and informational advantages concerning the
value of the tenant's unharvested crop. Since the amount of credit provided will be related to the
tenant's expected future income, the landlord can set the contractual fixed payment to zero and stfll be
free to adjust the interest rate, or accept the customary interest rate and adjust the fixed payment and
share parameter, to realize an optimal outcome (Otsuka, Chuma, and Hayami 1992).
A popular form of implicit credit is the landlord's provision of inputs to the tenant under
a cost-sharig arrangement. Providing inputs this way is formally Identical to an implicit production
loan which, like interlinked contracts, would be adopted where credit markets are imperfect. Static
analysis of costsharing a em     may thus be inpropriate if credit constraints are to be taken
filly into account. Calculating the implicit interest rate charged for such production loans would help
determine the equity and efficiency consequences of share-cropping arrangements. Tbe few empirical
studies that have been done suggest that the interest rates may not be significandy different from those
charged by moneylenders, rates reaching as high as 50 percent (Fujimoto 1986) or even more
(Morooka and Hayami 1986). Where there are imperfections in credit markets, it is possible to
derive the precise conditions for share contracts under which the benefits from overcoming the credit
58



market Imperfections would be greater than the loss resulting from the Marshallian inefficiency
(Shaban 1991).
If tenants are assumed to be willing to bear greater risk as their wealth rises (decreasing
absolute risk aversion) then both working capital constraint with imperfect credit markets and risk
aversion by tenants would generate observationally equivalent outcomes. For a sample of Tunisian
farmers, Laffont and Matoussi (1988) found that a credit constraint rather than risk aversion led
farmers to prefer share over fixed-rent contracts. Since credit and insurance are at least partly
substitutable, it is likely that improvements in financial markets and in the insurability of risk will
lead to a decrease in share contracts in favor of fixed rent contracts. Such a shift should result in a
gain in overall efficiency since fixed-rent contracts have superior incentive properties.
None of the land rental models discussed here, or even Marxian-inspired models of
semifeudalism (Bhaduri's 1986) considers the tenant's reservation utility - usually the market wage -
to be endogenously determined. Instead they explain inefficiency and inequities as consequences of the
contracts themselves, despite the fact that such contracts are entered voluntarily by both parties (see
epilogue). But in light of the discussion in section 2, it would be surprising indeed if landlords with
some political power did not try to find ways to reduce the reservation utility of potential tenants and
workers.
Governments the world over have been concerned about the efficiency and distributional
implications of such tenancy arrangements, which in essence depend on the relative bargaining power
of each of the contractig partners. Tenancy and sharecropping have therefore been heavily
regulated. The empirical evidence suggests that government interventions into these have had little
success in achieving their stated objective of protecting tenants, which is hardly surprising given the
market imperfections leading to the emergence of share tenancy, and the design difficulty to welfare-
improving interventons. Historically, land reform that resulted in establishing owner-operated farms
appears to have been a far more successful way of addressing the equity question.
Evidence on efflidency of tenancy arrangenents
59



and draft animals were significantly lower on sharecropped plots than on owned parcels. No
statistically significant differences in efficiency were found between owned plots and plots rented on a
fixed-rent basis, supporting the hypothesis of the productive efficiency of fixed-rent contracts. Other
results point in the same direction (Sen 1981), and Otsuka and Hayaml's (1988) review of the
literature finds, at most, small efficiency losses associated with tenancy.
Government interventions have given rise to efficiency-reducing share contracts in a
number of countries. Otsuka, Chuma, and Hayami (1992) argue that in India, where the Bell and
Shaban studies were conducted, there were a number of government constraints on long-term fixed
rent contracts. That implies that the 16 percent in production losses adjusted for land quality derived
by Shaban are likely to constitute an upper bound.
L     transactions to circumvent imperfections in credit markets have been important in
West Africa in the past (Robertson 1985), and continue to be observed in a number of developing
countries where credit markets are absent or credit is highly rationed. Usufruct mortgage is still
reported to be common in Bangladesh (Cain 1981), Java (Morooka and Hayami 1986), and Tbailand
(Fujimoto 1988). In the Philippines, tenancy transactions emerged as a credit substitute in response to
limitations on the transferability of land (Nagarajan and associates, 1991).
Where there is imperfect information, on tenant' unobservable characterisics, landlords
may inpret the tenants' acceptance of certin types of contracts as a signal that can be used as a
self- selecing screening device (Newberry and Stiglitz 1979). The preference for tenants who already
possess some land and draft animals, which is well documented in the literature (Quibria and Rashid
1986; Shaban 1991) points in the same direcdon.
Tenancy has often been described as a rung on the 'agricultural ladder' that rises from
worker to share tenant, to fixed rate tenant, to owner and permits farmers to acquire capital and
gricultural knowledge. In a static firamework this has been modeled by making production a fimction
of tenats' effort and the landlord's provision of management skills (Eswaran and Kotwal 1985).
Aldtough quantitative evidence is limited, Reid (1973) argues that this function of tenancy played an
important role in the U.S. South after the abolition of slavery. Lehman (1986) noted the importance
of tency in facilitating capital accumulation in the intergenerational transfer of farm holdings in
61



Ecuador. Tenancy would be expected to facilitate capital accumulation where land is abundant relative
to labor and where rents or owners' shares are low. Longitudinal studies of changes in tenancy
pans and capital accumulation over the life cycle of tenants would help shed more light on the
relative importance of this phenomenon in different environments.
Condusion
Because of the productivity advantage of small over large farms, it is more profitable for
large landowners to rent out land under fixed-rent contracts than to work it using hired labor, if
markets and information are perfect. If effort is unobservable and credit is rationed or insurance
markets are imperfect and tenants risk averse, the first best fixed-rate tenancy contract may no longer
be attainable and a second-best share contract would be adopted instead.
Empirical investigations show that share tenancy arrangements under a wide variety of
conditions are a highly flexible tool for adjusting to such constraints with relatively modest losses.
Since effort is not fully enforceable and even limited enforcement is likely to be associated with some
cost, the adoption of share tenancy (or wage labor) contracts would still be associated with some loss
of efficiency. Removing the conditions that prompt the emergence of share tenancy are likely to lead
to modest efficiency gains and will be more effective than legal prohibition of such contracts (see Part
II). Greater efficiency gains may be associated with the removal of the distortions that lead to the
adoption and perpetation of wage labor contracts, and large commercial farms rather than fixed-rent
or share contracts.
PART m: LAND POUCY
Many institutional arrangements in land markets emerge as a result of attempts by the
contracting parties to overcome problems of asymmetric information, moral hazard, and covariance of
risk. Other types of institutional aangements represent interventions by the government or
community intended to produce outcomes that are more congruent with the society's objectives than
those that would be generated by market forces alone. Here, we wI define efficiency and equity as
the main objectives underlying intventions in land markets, although equity can be considered a tool
62



for achieving the more encompassing objective of minimizing social tensions. The two objectives are
not always compatible; in some situations interventions that facilitate greater equity would reduce
efficiency, and vice versa. But not always.
Four broad types of land distribution and production relations remain today (figure 1),
each with its own characteristic policy problems. Family farm systems under freehold or communal
tenure face problems of access to credit, land registration and titling, tenancy regulation,
fragmentation, and taxation. Communities in which communal tenure prevails face decisions about
allowing sales to outsiders. Where large scale mechanized commercial farms coexist with low wages
and unemployment, governments have to consider ways, such as the elimination of tenancy
regulation, the elimination of agricultural subsidies and/or land reform, to make farm size
distributions more compatible with equity and efficiency objectives. In wage plantation systems,
contract farming and taxation are the important issues. For collective and state farms systems the key
concern is whether decollectivization or privatization should aim to establish large commercial farms
or small family farms.
7. Land registration and titling
The issues
Land dtles an registation reduce the problems of asymmetric information and thus
provide the instittional framework to facilitate land sales. Such transfers can enhance efficiency by
transferring land from bad managers to better farmers and by facilitating the use of land as collateral
in the credit market. Transfers of land, which are facilitated by land tides may negatively impact
equity as well as efficiency if economic and insdtutional distortions encourages accumulation of land
by influential or wealthy individuals. As establishment and maintenance of land tidies is not costless,
whether to introduce titding has to be based on a comparison of the benefits of land tites over and
above existng arrangements to regulate land transactions and the likely cost of such arrgemes.
In the early stages of agricultural development, transactions in land take place mainly among
individuals who are members of the same community and who generally share information about the
63



rights enjoyed by a renter or a seller, and about rights to specific tracts of land. With more advanced
agriculture and increased mobility, communal constraints on sales to outsiders are abandoned and
transactions are increasingly with individuals who are not members of the same community. The
scope for asymmetric information increases, generating inefficiencies in the land market since the
price of land may no longer reflect its true social value and the extent of land transactions becomes
less than optimal. To reduce these informational inefficiencies and the associated welfare losses,
societies develop institutional arrangements to reduce risk, such as the requirement in the Arthsastra
(4th Century B.C. in India) that land transactions be conducted in public with witnesses or the
establishment of a centralized public register that tracks land plots and those who have rights over
these plots. As early as 600 B.C., the Bible describes a land transaction between the prophet
Jeremiah and a relative in which a written record of the transaction was kept in two copies with a
certain priest in Jerusalem.
Public registers provide potential buyers or renters of land with a way to verify that the
rights they are about to purchase belong to the seller. A functioning legal system and effective
enforcement mechanisms are other institutional arrangements designed to reduce the uncertainty
related to land transactions. Without such arrangements to reduce the risk of challenges to land rights,
the incentives to invest and to work hard are weakened. It is often more efficient to reduce the risk
through the provision of public goods (and records, police, judiciary), than trough the private
individual allocation of resources (guards, elaborate fences).2
Asymmetric Information and risk are at their extremes in frontier areas, where specific
plots have no previous owners, though the government usually claims formal ownership. While often
the land is subject to a general claim by trbal groups who have been using it for hunting, gathering,
horticulture or livestock grazing, some of it is also claimed by individuals who have migrated from
other areas. Since there Is no culturally unified community from which to ubtain knowledge, the
administrative infrastructure Oand record, offices, courts, police) typically becomes overloaded by
claims and counter claims. It is not uncommon then to find private (and necessarily segmented)
institutions protecting property rights over land (gunmen, fortified property). And because the
2 In Ugand  and COte d'Ivoie, Ind privuiiaon significandy decreased rids and rdacon coss
associated with transferng lnd relt in inasd land trasfr, a factor commonly associated with higher
productivity in agrculture (Barrows and Roth 1990; Atwood 1990).
64



Institutions for recording property are not well developed, land claims based on forest clearing can
lead to excessive deforestation (Southgate, Sierra and Brown 1991).
Institutional arrangements for land records and title documents also have beneficial
implications for credit markets. In lending, asymmetric information provides ample scope for moral
hazard. Collateral has long served as a means of minimizing the efficiency losses associated with
asymmetric information and moral hazare nnd land has traditionally been viewed as an ideal
collateral asset in areas where land is scarce (Binswanger and Rosenzweig 1986). For land to be
useful as collateral, however, the lender needs to be assured that the borrower-operator has the right
to dispose of the land by sale or the transfer of use rights. Thus the documentation of land rights
makes land a form of credible collateral, affects the willingness of lenders to make loans and may
make credit markets more efficient (Feder, Onchan, and Raparla, 1988). Where the inability to use
untited land as collateral for credit is the relevant constraint, the issuance of titles can provide a
solution in the long term. But other constraints such as small farm size prevent the operation of a
credit markets, land titles may fail to be useful until these obstacles are not removed (Atwood 1990).
Under ideal conditions government intervention in land registration is theoretically
neutral in its effect on equity. In practice, however, titling c3n lead to greater concentration of land
and to the dispossession of groups that have enjoyed land rights under a customary system that
predated the formal system. When titling is introduced, wealthier and beter connected individuals
may use their information advantages to claim land over which other, less informed, individuals have
customary rights. Even when there are no information advantages, titling based on the on-demand
principle involves fixed and relatively high transacton costs for surveys and bureaucratic processing
that put smallholders at a disadvantage. The equity-reducing impacts of titing on this basis are well-
known. The introduction of selective dtling on demand greatly facilitated the emergency of haciendas
in Central Luzon (Hayami and Kikuchi 1984), Guatemala (Cambranes 1985), El Salvador (Lindo-
Fuentes 1990), and Nicaragua (Newson 1987). In Bolivia during the 1980s, the tiding agency granted
titles to very large farms in the Eastern Lowlands within one to two years, while applications from
smallbolders without the benefit of helpful lawyers have an average processing time of 12 years.
Bruce (1988) notes that land grabbing by influential individuals during tiding programs who are able
to use the rules in their favor did more to facilitate land concentration than transactions in the land
market following the issuance of tides. The profitability of consolidating several small untitled
65



holdings and geting a single title provides incentives for w'althy individuals to buy out smalholders
and to concentrate their own holdings. Titled land also provides advantages in the credit market
(Feder, Onchan and Raparla 1988) that are likely to increase income disparity.
The policy implications
To avoid these undesirable effects, titling programs should be accompanied by publicity
campaigns to ensure widespread lnowledge of the rules and procedures. Both equity and efficiency
considerations argue that titling programs by systematic rather than on demand. Efficiency is
increased through economies of scale and equity by the fact that all claims in an area are registered at
the same time. The ryorwari sys. a introduced by the British system In Southern India around 1820
and similar systematic ttling progrmns elsewhere show that conflicting claims can be dealt with
dhrough a relatively quick administative procedure rather than through lengthy and costly legal
channels.
Because dtling programs can be expensive, the issue of optimal expenditure is relevant
(see Malik and Schwab 1991). Feder and Feeny (1992) have demonstrated that when individual
wilingness to pay for tidtling determines the aggregate public expenditure, there may be a tendency for
over-investment from a social wdfare perspective. Frequenty, some less costly arrangement than
foral titling may significantly lessen the problem of asymmetry of information. In Rwanda, the local
municipality issues affidavits that attest to he ownership rights of specific individuals over specific
tracts of land but are not based on precise surveys (Blarel and associates 1992). A lower cost system
was also used in Thailand prior to the inrduction of formal dtling (Siamwalla et. al. 1990, Feeney
1988).
Communal systems constitute a special case. Communal land is not considered adequate
collateral In formal credit systems because of constraints on sales to outsiders. Issuing individual titles
in communities that maintain such constraints may improve neither the security if tenure nor access to
credit, although indib ħual tides would be helpful to avoid barriers to the emergence of rental markets
within the community. Until the restrictions on trnsfers to outsiders are eliminated, a community tle
could be issued to ensure the community's security of ownership against well-connected outsiders.
Paeu (1992) advocates registering land as corporate property as a wa, of decreasing the costs
04.!



associated with tiding while reaping many related benefits such as insurance, flexibility of land
allocation, and the utilization of genuine scale economies in subsidiary activities. Experience with
group ranches in Kenya suggests that imposing group titles from above is unlikely to be successful
while issuing individual tides does not prevent farmers from taking advantage of scale where they
exist (Grandin 1989).
Another case for community titles concerns common property resources, such as
communal pastures, forests, or other marginal lands. Such areas constitute an important safety net for
the poor that may be particularly important in high-risk environments where alternative means of
insurance are unavailable. Community mechanisms for managing common property resources have
tended to weaken with economic development (Lawry 1991; Jodha 1986 and 1990), and privatization
of such resources in India has led to significant increases in yields. But the preservation of common
property resources could be desirable from an equity perspective since privatizing these lands takes
away a part of the social safety net for the rural poor. Providing a community tidtle for these lands can
protect communal rights from outside encroachment and prevent the poor from being excluded from
communal property. We need to learn more about the management and the relative importance of
such areas to specific social groups.
Assessments of the impact of individual titling on efficiency vary. Atwood emphasizes
that in a distorted environment, introduction of land titles may decrease equity and efficiency. Feder,
Onchan, and Raparla (1988) find that in Thailand, where possession of a tide can be considered
exogenous' output is 14 to 25 percent higher on titled land than on untitled land of equal quality.
The market value is also much higher for tidtled land than for untitled land of similar quality. Less
rigorous evidence is provided for Costa Rica by Sals and associates (1970), who estimate a positive
correlation of .53 between farm income and title security. Studies in Brazil and Ecuador also suggest
a positive association between farm income and tidtles (IDB 1986). But several studies have
demonstrated that the credit market advantages of titles account for the lion's share of their effects
and that ownership security does not significantly affect demand for tiding (Adholla and associated
1991). rTtling may have no significant effect at all when legal or customary rules limit land
>  If the decision to acquire title is endogenous, estimation of the effects of tiling using crosssectiol da
is subject to simultaneity bias (Boldt 1989; Stanfield 1990).
67



asactions and credit markets are weak. In Latin America where credit markets are more devdoped,
recent land dding programs appear generally to have led to increases in the value of land, without
encouraging increased concentration - at least in the short term - (Stanfield 1990).
68



8. Land Tax
The issues
In most developing countries, land taxes have evolved from tribute payments to feudal
lords or to a colonizing power. Because the taxes went to central government budgets, local
willingness to pay depended on strong enforcement by tax collectors, who shared in the revenues.
Inflation and the difficulty of centralized collection eventually led to the erosion or complete
disappearance of such taxes. Today, the policy question is whether to reinstate land taxes and,
perhaps to use them to finance investments and serviced in local jurisdictions, as is done successfully
in the United States. In theory, a tax on land has three main advantages over a tax on agricultural
output or exports: (1) if a land tax is based on the potential monetary yield of a certain plot under
normal conditions, a land tax has minimal disincentive effects; (2) it facilitates taxation of the
domestic agricultural sector while being much less regressive than poll taxes; and (3) if the tax basis
is changed infrequendy, a land tax does not discourage investment in land improvements.
If risk is high and insurance markets are unavailable or imperfect, introducing a
significant land tax (based on average incomes) can lead to increasing land concentration as Hamid
(1983) has shown for India. Under these conditions, a tax based on actual output, which acts as an
insurance mechanism in the same way as sharecropping does, might be preferable to a lump-sum tax
on land (Hoff 1991). It can be shown, however, that for realistic values of risk aversion, income
variation, and export taxes, producers would prefer a land tax, balanced by an equivalent reduction of
export taxes (Skinner 1991).
Administering a tax on land effectively and equitably requires having an official record,
or cadastre, of the size, value, and ownership status of each tract of land, its productive capacity and
information on the costs of outputs and inputs. Land tax administration also requires a property tax
law that assigns property rights and tax obligations and an administrative organization that keeps the
register up to date and assess, collects, and enforces the tax (Bird 1974). Even in the few developing
countries able to meet these conditions, land taxes are relatively unimportant, suggesting that the
administrative or political costs may be higher than the incentive advantages associated with a land
tax.
69



Progressive land taxes are often advocated as a means of making land speculation less
attractive and inducing large landowners to sell out or use their land more intensively (see Hayami,
Quisumbing, and Adriano 1991 on the Philippines). Landowners often find ways around such taxes,
however, from establishing dummy divisions of their holdings to lobbying for exemptions from
progressive rates associated with effective use of the land (as in Brazil), which sharply diminish the
effectiveness of progressive land taxes in breaking up large commercial farms. Such an approach was
applied and failed in Argentina, Bangladesh, Brazil, Colombia, and Jamaica (Strasma, Aism, and
Woldstein 1987; Bird 1974); Carter (1992) in a simulation model calibrated to Nicaragua finds that a
progressive land tax is unlikely to significantly alter the distribution of land. And even if such taxes
did work, it is not obvious why such an indirect approach would be politically more acceptable than
direct redistribution of land. Progressive land taxes are also likely to be associated with higher
administrative costs and protracted litigation.
Plicy implications
Where the administrative requirements - an up-to-date cadastre plus administrative
organization - are lacking, flat or mildly progressive land taxes based on rough classification of
holdings may still be useful for raising revenue and providing some modest incentives for owners to
sell off poorly utilized land. The United States has found success by assigning the administration of
land taxes to local authorities and earmarking tax revenues for local infrastructure and local
government services. By increasing the local visibility of the benefits fianced with the tax revenue,
this approach may increase willingness to pay a land tax. It may also reduce adminisrative costs since
local governmens should be better able to assess land values and land ownership.
9. Regulations dmiting land sales
Governments and local authorities have often placed restrictions on land transactions.
Restrictions are typically placed on land sales and rentals when major changes are introduced to alter
the land ownership pattern (redistnbutive land reform or settling programs). The restrictions are
designed to prevent an increase in the number of landless and in the social tensions that accompany
landlessness. Since these restrictions also prevent some transfers of land from worse to better farmers
70



or managers, there is likely to be some efficiency loss. Such restrictions are frequently evaded,
however, through disguised sales and rentls, which are likely to involve transaction costs that
constitute a loss to society.
Restrictions on the rights of land reform beneficiaries or settlers on state-owned land to sell the
land also reduce their access to credit. Often new owners are forbidden to mortgage their land during
an initial probation period. Since that period coincides with the establishment phase, when their need
for credit is most urgent, the efficiency losses may be considerable. Land rental contracts (usufruct-
mortgaging and kasugpong contracts ) that have arisen as credit substitutes, in some places, such as
the Philippines (Nagarajan, Quisumbing and Otsuka 1991) involve considerable efficiency losses.
Sometimes restrictions on sales are not total, as in communal systems that permit sales
only among members of the community. The welfare losses from the sales restrictions are less than in
the case of a total ban, but they are not completely eliminated.
In the early years after a redistributive land reform in areas where land markets are thin
and accurate information may not be available on the expected stream of incomes from the land, it
may be reasonable to impose a temporary restriction on sales of say, three to four years. That would
allow sufficient time to acquire lnowledge about a farm's potential and to avoid sales at prices below
the real value of the land, which would run counter to efficiency and equity objectves. Such
restrictions would not be needed, however, in areas where former tenants receive land they have been
tilling since they can be assumed to have adequate knowledge of the land. In the case of partial
restrictions under communal systems, the ban on sales to outsiders may serve a protective role in
environments where outsiders with strong political connections may attempt to take over land in the
community. Where appropriate institons for inragroup decNsion-making are available (Libecap
1986), permitting the community to limit sales and giving it the right to decide whether to eventually
allow sales tD outsiders may be an acceptable compromise between equity and efficiency concerns (see
Barrows and Roth 1990). As traditional social ties loosen or the efficiency loss from the sales
restriction becomes too high, groups are likely to allow sales to outsiders. The recent constitutional
reform of the land rights system In Mexico allows for free sales and rental within all efidos and for
decision-making by majority vote on whether to eliminate the restriction on sales to outsiders.
71



The most common means of restricting land sales are upper and lower bound size
restrictions and zoning regulations. Land ownership ceilings have often been imposed in an attempt to
break up large estates or to prevent their reconcentration. Among countries that have imposed ceiling
are Bangladesh (Abdullah 1974), India (King 1977), Indonesia, Japan, Korea, Pakistan, South
Vietnam, Taiwan, Egypt, Ethiopia, Iran, Iraq, Zimbabwe, Bolivia, Cuba, El Salvador, Guatemala,
Mexico, and Peru. While such ceilings can theoretically increase efficiency where a negative
relationship exists between size an,. vroductivity, in practice the ceilings have been evaded through
fictitious subdivisions or have become superfluous over time through inheritance. Ceilings were often
commodity specific providing much larger limits for sugarcane, bananas or livestock ranching.
Therefore, they encouraged inefficient conversion to products with the highest ceilings. Rarely did
ceilings alone enable the poor landless or extremely small farmers to vurchase land; rather, they
enabled farmers with medium-sized holdings, who had already acquired some equity, to enlarge their
holdings (Chile).
Despite these flaws and loopholes in practice, several studies do credit land ownership
ceilings with a major role in preventing new large consolidations after land reform (Cain 1981;
Mahmood 1990). In Japan and Korea, success in preventing the reaggregation of land may be
attributed as much to the availability of attractive investment opportunities outside agriculture and to
noneconomic factors sucL as attachment to land as to the ceilings on land holdings. Ceilings imposed
foLlowing a land reform that results in fairly homogenous holdings might be effective and less
disortionary in preventing massive reconcentration of land.
At the opposite end, resricdons on minmwn holding size are intended to prevent
excessive frgentation of farms. While it is not clear that fragmentation is always a negative
phenomenon (see below) a floor on farm size might provide a useful countervailing effect in a society
where inheritance customs lead to extremely small farms. Whether the intervention improves
efficiency depends on the specific circumstances. Also to be considered is that many restrictions on
subdivision of land or minimum holding size have historically been used to prevent ex-slaves, tenants,
/    and other powerless groups from acquiring ownership rights to land and thus eventually competing
with fanm  established by the rling group. Restrictions on the subdivisions of large farms in Kenya
and Zimbabwe have limited the prospects for land resettlement schemes (Leys 1974) and in these
circumstancs clearly reduced efficiency.
72



Governments often adopt zoning regulations, .e. assign specific uses to certain lands to
overcome environmental externalities rather than allowing market forces to determine land usage. In
urban areas, the objective of zoning is to prevent commercia or industrial activities from locating in
residential areas and creating noise and pollution. In rural areas zoning of land for agricultural use
provides benefits such as tax credits, exemption from assessments for urban type services, eligibility
for soil conservation programs, and proteceion from nuisance suits, but forecloses the option of selling
the land as a residential property.? In general, zoning is justified if negative externalities need to be
reduced by more than the cost of zoning enforcement.
Zoning laws established for social or environmental reasons may run counter to
economic incentives. Zoning may then need to be supported by some type of incentive mechanism,
and political support for implementation of the regulations becomes essential to their enforcement
(Barrows and Neuman 1990). If there are sharp conflicts between private profitability of land uses
and zoning regulations in a country with weak institutional infrastructure, and little popular support
for the zoning measures, zoning may lead to excessive rent-seeking and corruption. If zoning results
in the emergence of extensive rent-seeking the benefits may greatly decrease or even become negative
(Mills 1989). Zoning laws affect supply and demand for land and may lead to consumer mobility in
response to zoning Mfebout effects). The attempt to counteract production or agglomeration
externalities through zoning laws also generates the potential for rent seeking behavior by landowners
who either try to evade existing zoning regulations or lobby for the imposition of a set of laws which
would provide them with a differential advantage. All of these issues have been analyzed largely in
isolation of each other and a comprehensive analytical treatment is not yet available (Pogodzinsi and
Sass 1990).
X  Hmabeny and Barrows (1990) find that parcel caterics in geneal ddemine whvetr agicultusb
zoni  has positive or neptive pnce effects, in paicular parcl i and ditance from urban areas. (For a
toview of the effects of urbanizaioD on agnculture, swe Bhadra and Brmndao 1992.)
73



10. Fragmentation and consolidatlon
The Issues
While governments often intervene to prevent fragmentation of farm land, such
intervention is not always economically justified. Tbat requires that inheritance customs or other
exogenous forces be responsible for most of the fragmentation, that losses from fragmentation be
substantial, and that existing markets be unable to counter fragmentation.
While inheritance customs probably explain much of the fragmentation ^f.arm land, it
may also reflect conscious decisions by farmers seeking to reduce their risk by diversifying their farm
land and thus their crops (McCloskey 1975). 'Ibis factor is likely to be important where other risk-
diffusion mechanisms such as insurance, storage, or credit are unavailable or are associated with
higher costs than fragmentation. Fragmentation may also help to smooth out labor requirements over
time where labor requirements are highly seasonal (Fenoaltea 1976).
Among the disadvantages associated with fragmentation are physical problems (increased
labor time, land loss, need for fencing, transportation costs, and limitations to access); operational
difficulties (unsuitability of certain equipment, greater difficulty with pest control and management
and supervision, foregone improvements such as irrigation, drainage, and soil conservation); and
social externalities (need for extensive road and irrigation networks; Simons 1987). The few studies
which quantify losses fom fragmentation developing countries suggest that the losses involved are
modest, although further studies of the efflciency of farms or losses from fragmentation are clearly
needed. Indeed, Heston and Kumar claim that in Asia *it is hard to find instances where
fragmentation had involved high losses in output' (1983:211), and in Ghana and Rwanda, Blarel and
associates (1992) find fragmentation does not seem to hurt productivity and does improve risk
diversification and the allocation of family labor over time.
Policy impllcatIons
Relying on the market to eliminate   tation is liklcy to involve high transaction
costs to coordinte transfers among large mumbers of landowers. Transaction costs are much lower
74



under goverament programs, which are normally coercive and include a range of other development
initiatives, and returns can be high - Simons (1987) finds returns of 40 percent for France. However,
if the forces that led to fragmentation remain unchanged, land consolidation programs are unlikely to
have any long-term effect (Simons 1987; Elder 1962).
When should something be done about fragmentation? Experience in industrialized
countries shows that fragmentation becomes a serious constraint requiring intervention once it impedes
the ability to use machinery on a large scale in areas with a rapidly decreasing agricultural population
(Bentley 1987). This is rarely the case in developing countries, with their high population densities. In
addition, consolidation programs are likely to take a long time to complete, and they require
considerable human capital and well-developed cadastres and land titles. Immediate government action
to consolidate holdings does not appear to be a high priority in most developing countries, considering
the high costs and the potential reduction of interest in fragmentation as rural credit and insurance
markets improve.
11. Restrictions on land rentals
The !ssues
Governments have often introduced tenure securhy and rent control legisladon to protect
tenants from arbitrary eviction or to limit the amount of rent landlords can changc. The unintended
result has often been the eviction of tenants at the first hint of such legislation and the landlords'
resumption of self-cultivation on the home farm, resulting eventually in the formation of Junker
estates. In India, atempts to pwvide greater land security for tenants could be enforced only in stes
that imposed land ownership ceilings (King 1977), and even there, landlords found ways to evade the
legislation by signing tenants to short-term contracts which were exempt from protection, or by
rotating tenants from plot to plot.
Where rent controls have been effectively implemented and combined with protection
from eviction as in the Philippines or Taiwan, they do increase tenants' income, but since there Is no
transfer of ownership, they are still likely to result in dynamic efficiency losses. In the longer term,
75



unless landowners find ways to circumvent the restriction on rents, such policies are likely to reduce
incendves for renting out land, resulting in efficiency losses from constraints on adjustments in
operational farm sizes. Investment is also likely to fall on farms on which tenants have a protected
status since landlords are unlikely to invest heavily in land from which they are prevented from
evicting tenants while tenants' incentives to invest are weakened by uncertainty about the inheritability
of the protected status.
Bans on share tenancy or low ceUings on the landlord's share are widespread even
where other forms of land rental are allowed, such as the Philippines (Otsuka, Chuma and Hayami
1992), Brazil (Estatuta da Tierra 1964), Zimbabwe (Palmer 1979), South Africa (Bundy 1985),
Honduras, and Nicaragua (Dorner 1992). These restrictions are motivated in part by the common
belief that share tenancy is exploitative (because, under conditions of land scarcity, tenants are likely
to receive incomes close to their reservation wage) and in part by efforts to eliminate the Marshallian
inefficiency associated with share contracts. But if the choice of contract is endogenous and if share
contracts provide efficiency gains under circumstances of credit constraints and high risk and
supervision costs, simply prohibiting share contracts without changing the underlying framework of
market imperfections is likely to result in very slight gains in efficiency (Otsuka and Hayami 1988).
More likely, the bans will be ignored, giving way to disguised transactions or less efficient wage
labor contracs that improve neither equity nor efficiency. Tenancy has long been an important
transitional stage allowing peasants to accumulate capital and gain agriculural experience, so
elimination of sharecropping as a rung on the agrarian ladder will certainly not contribute to equity in
the long run. And considerable inefficiency in production may be associated with the absence of
sharecropping as an option, especially where restrictions on private ownership of land impede the
functioning of fixed-rent markets (Noronha 1985). Collier (1989) estimates static efficiency losses of
more than ten percent associated with unavailability of share contracts in Kenya. From al
perspectives then, bans on sharecropping and low ceiling on landlord's share have no merit.
76



12. Redlstributive land reform
The Issues
Most redistributive land reform is motivated by public concern about the rising tensions
brought about by an unequal land distribution. The common pattern is concentration of landownership
among relatively few large owners in an economy where labor is abundant and land is scarce. Thus
the masses of landless laborers and tenants who derive their livelihoods from agriculture receive
relatively less income because their only asset is labor. Redistributive land reform can also increase
efficiency, by transferring land from less productive large units to more productive small, family-
based units (section 4).X6 Yet, because of other market imperfection, land markets will not typically
effect such transformations of ownership patterns. The value of the land to large owners may exceed
the discounted sum of agricultural income smallholders can expect to receive despite their productivity
advantages from lower supervision costs if there are policy distortions favoring large owners or if the
access of small farmers to long-term credit haz  already been exhausted by mortgage-based land
acquisition.
Market values of land are determined in a way that prevents small farmers who lack
equity from building up viable farms and improving their standard of living while repaying their land
mortgage. Land reform schemes that require payment of the full market value of the land are likely to
fall unless special arrangements are made. In the simplest case, beneficiaries soon default and the
program ends. Many ambitious land reform programs simply run out of steam because full
compensation of old owners at market prices imposes fiscal requirements that the political forces are
unwilling to meet - that was the fate of programs in Brazil, the Philippines, and Venezuela. Some
programs attempt to avoid this problem by compensating landowners (with bonds) whose real value
erodes over time. Not surprisingly, landowners oppose this thinly disguised confiscation, and such
programs are politically feasible only in circumstances of political upheaval (Cuba, Japan, Korea,
Taiwan or Vietnam). Another approach is to finance land purchases through foreign grants or from
intal tax revenues or inflationary monetary expansion - or some combination.
X  Under circuma     of extem  poverty and lsndo   redistribution of land can also enhance efficiency
by iprovng t nutnritional wllbeig and thus the productve capacity of the population (Dasgupta and Ray
1986 and 1987, Mon. 1992).
77



Polley implications
Before any land redistribution program is ihtroduced, the implicit and explicit distortions
which drive land prices above the capitalized value of agricultural profits need to be eliminated.
Otherwise, small farmers will continue to have an incentive to sell out to larger farmers since the
eavironment would still favor large ownership holdings. In Brazil, the emergence of an agricultural
structure dominated by large farms owes much to the policy bias in favor of large farms (Binswanger
1987). The poor must be provided with either the land or a grant to help them buy it to compensate
for their lack of equity. Credit to beneficiaries for land purchases can only play a subsidiary role."'
The macro-economic and political environment also strongly affect the outcome of land reform
policies. In Chile, substantial increases in output followed the expropriation and redistribution of
almost 20 percent of the total agricultural land in 1964-70, much of it due to the increase in
investment induced by the favorable macroeconomic and political conditions (Jarvis 1985, 1989). In
contrast, output failed to increase significantly during the decollectivization and breakup irzo fawrily
farms in 1975-83, a period of extremely unfavorable government policies. Not until some of the debts
incurred to pay for the land had been forgiven and structural impediments affecting small farmers had
been eliminated did the program become fully effective. Removing distortions also lowers the amount
of grant assistance needed by small farmers to support their acquisition of land.
The type of manorial estate has a substantial bearing on the gains to be expected from
land reform. On landlord estates, would-be beneficiaries are already managing operational units so
land reform addresses primarily the equity concerns of society, transferring the entidtlement to land
rents while leaving operational farm structure largely unchanged. Potential efficiency gains are
associated with improved investment incentives and increased security of tenure (section 3). With
haciendas, the threat of land reform legislation often leads to the eviction of tenants and reductions in
the resident work force. The large commercial farms that result are more difficult to subdivide than
landlord estates or haciendas (de Janvry 1981, Castllo and Lehman 1983; de Janvry and Sadoulet
7 Organizations such as the Penny Foundation is Guatemaa have been able to buy land from owners and
distbute it to small faer.v with itle appan government subsidies (Foer 1992). These ca  usually
mvolve mome grnt elemet or subidy   the credit provnded to the smallholders, or the purchase of the land
below market prices on account of liabilities of the fomr ownr to govenment insituion or the worers
which ar forgiven as pat of the dea.
78



1989). Land reforms of Junker estates and large mechanized farms involve major changes in the
organization of production. The resident labor force and external workers have little or no
independent farming experience, and in many cases, neither the infrastructure nor the investments in
physical capital provide an appropriate basis for smallholder cultivation.
The availability of technology and of competitive input and output markets thus becomes
a crucial determinant for the potential of land reform to increase efficiency. Appropriate institutional
arrangements are needed to ensure access to extension services, credit, and markets. Such institutions
are especially important where land reform involves resettling beneficiaries on former Junker estates
or large mechanized commercial farms. To reap the efficiency gains of family farming under these
conditions seems to require increasing the density of family labor, and that may require resetding
landless workers from outside.2 Reform of these systems is likely to be difficult, but where the
alternative to reform is the perpetuation of large economic and social costs, including the possibility
of revolt and civil war, the cost of failing to reform may be enormous.
Opinions are divided on redistributive reform of wage plantations in the classic
plantation crops: banana, sugar, tea and oil palm. The fact that contract farming in these plantation
crops is practiced successfully in many parts of the developing world indicates that convertng
plantations to contract farming is feasible. Indeed, Hayami, Quisumbing, and Adriano describe the
successfil conversion of even a banana plantation into a contract farming system in the Philippines,
and strongly argue for bringing about more such conversations through a progressive land tax. The
efficiency gains from lower supervision costs associated with such a step are likely to be offset,
however, because of the genuine economies of scale in plantation crops.
Tryig to replace plantations with collectives rather than contract farming has been
unsuccessful. In Peru, the failure of collectivized sugar plantations to invest and their increased
esploitation of external workers who were denied membership rights led to strikes by collective
members that were put down by military intervention. Continuing losses - in part due to falling world
X  To oomm extent, credit and other public support can  sbtute for the advantage of mmily labor per
hocae  Loys (1978) found for Kenya that there ws8 very little difference in economic peformance between
high density  hm, with smal p!ots and low public investmen, and low denst sms with arg plots
and subal publc uppoft
79



sugar prices - provoked increased government intervention and the effective transformation of the
collectives into state farms (Kay 1952). In Malaysia rubber plantations which had been established on
a collective basis were split up and allocated to individual farmers at maturity to ensure proper
tapping (Pickett 1988).
13. Decollectlvization
The poor performance of collectives and state farms the world over is so obvious that
the question facing the liberalizing economies of Eastern Europe and the Commonwealth of
Independent States is not whether to privatize but rather how quicldy and in what form - as large
commercial farms or family farms.
Policy implications
The discussiotns in this paper imply that four issues appear to be of overriding
importance in determining this policy choice:
*    The small farm option is viable only if there are competitive input and output markets.
Otherwise the land rent ai4 the entrepreneurial rents from agriculture would be captured by the
monopolistic output marketers and input suppliers rather than by the new farm owners. Risk
diffusion mechanisms also need to be functioning adequatly else covariate weather or price
shocks can force distresc sales by new landowners, who do not have other assets or income
streams. Work on creating competitive input and output marketing systems and a viable
financial system therefore has to start before large farms are split up ino individual
landhchlings.
Experience from China, Vietnam, and East Germany shows that inputs and machinery services,
which have previously been supplied by the cooperative, are more efficiently provided by
private contractors who lease or buy the machinery stock from the cooperative in a competitive
process (Nolan 1988, Pingali and Xuan 1992, Pryor 1992). The Chinese experience also
suggests that farmers and machinery suppliers respond to the changes in operational holding
80



size by adopting a difference and generally more efficient pattern of mechanization (Ling
1991). TIhs suggests that tne excessive lumpiness of the existing machine stock is not a serious
constraint to smaller scale farming.
Agriculture research, extension, and other production support services take on special
importance since many farm workers are likely to lack the skills needed to manage their own
farms. Some of the structures that served quasi-governmental functions on collective and state
farms particularly by providing education and health services could be retained as well. They
might also eventually develop into independent cooperatives for supplying machinery, custwr
plowing machine rentals or for inputs and possibly credit - all in competition with the private
sector (see Nolan 1988; Pryor 1992).
*    Where capital skills, technology, infrastructure, or competitive markets for inputs and outputs
are lacking, enthusiasm for independent farming may be lacking as well. If only a few
entrepreneurs are willing to farm, the resulting farms are then likely to be too large for the cost
advantage associated with the use of family labor, and large commercial farms, heavily
mechanized or dependent on large numbers of hired workers, will emerge in their stead. Most
likely such large farms would continue to press for subsidies, emerge as rent-seekers from the
rest of society and, if successfiu, generate insufficient employment. Therefore, countries may
need to find temporary arrangements, including long-term land leases, that will provide a
greater number of households with opporunities to acquire the necessary skills needed to allow
the emergence of a structure of smaller family farms more consistent with the income and wage
levels and rural labor forces that can be expected for these economies in the next few decades.
EPILOGUE ON METHODOLOGY
Scholars of various ideological persuasiveness and methodological commitments have
attempted to explain the great variations in land relations over space and over time which have been
the topic of this paper. Much of the discordance among these scholars is closely associated with their
choice of modeling strategies and assumptions. This epilogue relates the analytical results and the
81



observed variations in land relations discussed in this paper to the minimum set of assumptions needed
to derive the results or explain the variations. We distinguish several levels of assumptions.
Lewl A assumes se:.-mterested behavior, such as expected utility maximization or other
forms of purposive behavior, of all actors, who compete on a level playing field in an environment
with risk using voluntary transactions, with symmetrically distributed information and exogenously
given endowments of land, capital, and skills. Technology is characterized by constant or diminishing
returns to scale. Virtually none of the variations in land relations discussed in this paper can be
explained with these assumptions alone.
Lewl B adds constraints in the credit market cr assumes that market is entirely absent.
Formal models of surplus value from Marx to the generalized version of Roemer (1982) use this
approach to explain capitalist exploitation and the endogenous differentiation of maximizing individual
economic agents - who operate in a competitive environment with voluntary transactions - into
economic classes as the consequence of differences in their exogenous endowments of physical capital
and absent credit markets. Eswaran and Kotwal (1985) apply Roemer's approach to agriculture,
imposing in addition constant costs (section 4).
Lewl C adds asymmetric information, moral hazard, and incentive problems, arriving at
the analytical apparatus of agency theory. As Stiglitz (1986) summarizes, these assumptions are
sufficient to explain credit rationing, thereby giving an analytical underpinning to level B models.
They also explain various combinations of reasons for sharecropping and interlinked credit (section
6). Ihese assumptions are also sufficient to establish the superiority of family farms, as discussed in
the mathematical model of Feder (section 4 and appendix 2) and the historically widespread use of
teona    by large owners of land at moderate to high population density to circumvent the diseconomy
of rale (section 2). Incentives issues of collectives are also analyzed with this analytical apparatus
(section 4).
Level C models provide little insight into the process by which large landownership
holdings could accumulate or be perpetuated in systems characterized by voluntary transactions and
competition, and with constant or diminishing returns.
82



Level D adds several material conditions relating specifically to agricultural production,
generating the analytical apparatus used by Meillassoux (1981) or Binswanger, Rosenzweig, and
McIntire (1986, 1987). The material conditions most frequently used in this paper are covariance of
risk and returns among farmers and workers in a given agricultural region, the immobility of land,
which - when it is scarce - makes it Into a preferred store of wealth (relative to stocks and livestock,
for example) and of collateral, and exogenously given population density and processing
characteristics of specific agricultural commodities.
Covariance creates enormous difficulties for intertemporal markets for crop insurance
and credit. Because of land's preferred role as store of wealth and as collateral, an insurance and
collateral benefit is associated with landownership. Together with the failure of intertemporal markets
this preferred role explains the prevalence of distress sales and the accumulation of large
landownership holdings even in a competitive environment with strictly voluntary contracts and
diseconomies of scale (section 5). Ihe potential failure of land sales markets to improve efficiency in
an environment with missing or imperfect intertemporal markets is a powerful and historically
relevant illustration of the theorem of second best (Lipsey and Lancaster 1957) of neoclassical
economics.
The explanation of variations over time and space of property rights to specific plots of
land (sections 1 and 2) requires the introduction of population density and its association with the
farming systems and the farm technologies, as explained by Boserup (1965). The seasonality of
production, the timeliness requirements of specific crops, and the economies of scale of the
processing plants or transport facilities required for them are necessary material conditions to explain
the survival, in only a few specific plantatio'% crops, of wage plantations in the absence of slavery or
indentured labor (section 4). Note that anthropologists, like Marvin Harris, who use behavioral-
materialist approaches also carefully specifying their detailed material assumptions, although their
themes extend well beyond those discussed in this paper.
Level E partly abandons the assumption of voluntary contracts (for the case of slavery
and bondage) and extends the analysis beyond individualistic approaches and transactions by
introducing rent seeking, coalition building, and the coercive power of the state to enforce laws.
Ihese additions facilitate the explanation of the use of bondage and slavery, tribute systems, state
83



allocations of preferential land rights and enforcement powers to ruling groups, distortions in
commodity and fctor markets, and distortions In public expenditures specifically intended to extract
rent and make large ownership or operational holdings competitive with independent family farms
(section 2 and 3). The historical literature has sharply differentiated between coercive and noncoercive
methods of rent extraction and has often equated the elimination of coercive means with the leveling
of the playing field. While there are certainly important qualitative differences between coercive and
noncoercive means, the differentiation seems to have obscured the continuity of rent seeking or
surplus extraction along alternative paths such as taxation of the free peasant sector, land allocation,
monopoly marketing, and the allocation of public spending.
Level E explains the emergence and persistence over time of highly dualistic farms size
structures as the result primarily of a rarely broken chain of rent seeking (sections 2 and 3). It
explains the poor economic performance of many such systems as the result of a dissipation of rents
into the cost of competition for them among zent-seeking groups.20 Within the chain of rent-seeking,
the officially sanctioned set of legitimate instruments of rent seeking may be progressively reduced by
gradually eliminating slavery and serfdom, tribute and corvde, and land rental, until only output and
factor market distortions and differential allocation of public expenditure remain. With exogenous
variation in the set of instruments available for rent seeking, this framework of analysis can explain a
substantial proportion of the variation over space and time in the level of use of each of the available
instruments. For given instruments, modeling at level D can also, in principle, investigate the income
distributions and efficiency costs associated with the resulting distortions, while the theory of rent
seeking behavior (Tollison 1982) can be used to investigate the extent to which rents are dissipated in
the process of competing for them.
reBner (1975, 198S) argues tt under faudalism the rents extmcted from pents by luoded elites were
almost completely dissipated and that tho esult  failure by peasants and landlords to rnvest in land
imwovem    and drft animals wa responsible for the extension of arble farming to marginal lands and the
declining productivity asocited with pwuin gwth in feudl European aiulre. Thus it was th  nt
oekigg itself that led to the Neo-Malthusian or Ricardian subsistenco cises of the twelfth and thinteth
centuries, daer than to popuation-induced positive Boserup-succes of investment,  age in technique,
inresed divsion of labor, and agiculura productivity growth. This explansion of sagnant or declining
prductivity is smilar to tht documented by Kreger, Sdhiff, and Valdes (1992) to explain trecet stgata
of agrulture and limited technical change in much of Africa as a cosqun  of the extrdinry high
taxaion of (mosty smallholder) agriculte in my Africn counties by urbn-domatd stes.
84



Finally, level F asks questions that are touched on only lighdy in this paper about what
determines endogenously the changes in the set of instruments available for rent seeking or surplus
extraction in a given country at a given time. Population density and its distribution over space
becomes and endogenous variable. The questions include the extensively debated issues of the demise
of feudalism and bondage (Marx; Dobb 1977; Brenner 1985); the abolition of slavery (Fogel and
Engerman 1977; Meillassoux 1991); the elimination of corvee, tribute, and debt peonage; the power
to monopolize output and input markets (Anderson and Hayami 1986); elimination of the land rental
option; and land reform (de Janvry 1981). The questions analyzed also include why revolt and
revolution are necessary in some cases, while in others a change in the set of instruments available is
successfully accomplished by reform, and why some reforms lead to stable and efficient production
relations, while others result in institutions that are unsuccessful in either equity or efficiency.
These are the grand themes of historians, classical economists, and Marxist historical
materialist analysis. These issues usually involve coalitions (or their breakdown) that, except in purely
agrarian societies, extend beyond opposing rural groups to include manufacturing, trading, financial,
bureaucratic, or foreign interests. Therefore additional exogenous elements (including material ones)
from outside agriculture must be factored into the exploratory framework. Much of the work on these
themes that we have come across neither explicidy specifies assumptions about the distribution of
information (level C) nor formally includes into the analysis specific material conditions of agriculture
(introduced in level D) or other sectors of the economy. And while rent seeking of level E is implicit
in the questions asked, and coalitions or their breakdown are discussed, the coalition building
associated with rent seeking is rarely modeled explicitly. There may be some gains to be had from
more formal considerations of these omitted elements and their incorporation into the structure of the
andysis of these grand themes.
85



Appendix 1
Interventions to Establish and Support Large Farms
The literature on emergence and evolution of manorial estates and the production relations
prevailing within such estates has focussed largely on examples from Europe (mainly Britain, France,
Germany, and Eastern Europe). This appendix, which explains table 1 in the text, provides evidence
on the establishment and evolution of large farm systems from a wider range of setdngs and covers a
longer time period.
The examples discussed here all suggest that neither the establishment nor the continued
existence of large farms were due to their superior economic efficiency and/or the presence of
economies of scale in agricultural production. The establishment of large farms was due to
government intervention in favor of large landholders via land grants and differential taxation.
Withdrawal of these privileges led either to their disintegration into landlord estates or to a shift
towards rent seeking and more subtle forms of support for large farms.
Asia
India (North)
Land market interventions. The hacienda system is already described in the Arthshastra from
the 4th century BC. In the first century, land grants comprising some ten or more villages each were
made to priests and to a few members of the ruling family and high officers of the state (Sharma
1965). This process of land grants 'culminated in the 11th and 12th centuries, when Northern India
was parcelled into numerous political units largely held by secular and religious donees who enjoyed
the gift villages as little better than manors' (Sharma 1965:273).
DifWerendal waton and labor levies. Corvee labor emerged in the second century and
remained prevalent until the tenth century. Between the fifth and tenth centuries, where population
density was high enough, as in Gujarat, Rajasstan, and Maharastra, permanent tenants were reduced
to tenants at will. Where population density was low, tenants and artisans were tied to the soil in the
same manner as serfs in medieval Europe (Sharma 1965).
China (South)
Diferend tation and labor kvies. The equitable land allotment system introduced around
600 under conditions of land abundance allocated land equally among all members of the commTunity
in return for tax payments. Slaves received the standard size of plot but had to pay only half the taxes
demanded from free men (Chao 1986). Peasants, however, could not escape the tax burden since
farmers who fled to uncultivated lands were returned to their village by the authorities. DeFrancis
(1956) quotes reports of 600,000 "refugees having been collected in a single year (544). To escape
the tax, many cultivators presented themselves as serfs or "bondservants" to large landholders or
monasteries, leading to the emergence of large estates. In a major land reform in 1369 under the
Ming dynasty, the estates were broken up into small freehold farms (Eastman 1988). Following the
land reform, tax captains were installed to administer tax collection from units of 110 households each
and to deliver grain taxes to government warehouses. Using corvee labor and bondservants, they were
also active in land clearing to expand their revenue base (Shih 1992). They accumulated modest
86



estates of their own thanks to their ability to provide credit. Increasingly heavy tax demands (to
finance wars) left many tax captains in a desperate situation.
The new gentry class that began to emerge in the fourteenth century was exempt from both
taxes and labor services. Since gentry landlords did not pay taxes, they were able to reap higher
returs from land and accumulate wealth. They were able to further increase their holdings after
periods of disaster by foreclosing on lands they had accepted as collateral for credit (Shih 1992).
These advantages made it easy for members of the gentry to accumulate land, decrease the tax
captains' revenue base, and finally buy out bankrupt tax captains, who by the end of the century had
lost most of their land to gentry landlords. As gentry landlords increased their moneylending
activities, small owners in financial difficulties had to resort to selling their land or selling themselves
to gentry landlords as serfs or bondservants, thereby obtaining partial exemptions from their tax
obligations. Gentry estates grew to several thousands of hectares in size, with a labor force of over
10,000. The estates were often split up into smaller farms of about 500 hectares, managed by
specially educated bondservants (Shih 1992).
Following the change from the Ming to the Qing dynasty in 1644, gentry landlords lost their
tax privileges. Declining population and greater opportunities for off-farm employment during 1630-
50 increased the amount of land available and, as in Westem Europe, improved the position of
peasants (Shih 1992). In the second half of the seventeenth century, the heritability of serf status was
repealed, and serfs were fully emancipated in 1728. Operation of a large home farm using wage labor
was no longer profitable, and landlord estates emerged (Wiens 1980), considerably improving the
position of tenants. Tenancy allowed operational holdings to adjust to household size and led to very
labor-intensive cultivation and high yields (Feuerwerker 1980).
Japan
Land market interventions. To provide incentives to make the investments required to transform
wasteland into paddy land, the land reclamation bill of 723 made such land the heritable personal
property of the developer. This provision led to the emergence of a separate category of private land
that was tax exempt and excluded from the communal tenure system in which land was redistributed
every six years among all members of the community (Takekoshi 1967).
DWrerental taxes and labor levies. In return for such land allotments, farmers had to pay
tribute in kind as well as special labor services of up to 140 days a year (Takekoshi 1967). Cleared
and temple lands, as well as land belonging to the nobility, were exempt frota all tribute
requirements. In order to obtain immunity from tributes, many landowners transferred their lands to
temples or members of the nobility. While they had to give up the heritable right to the land, original
landholders did in most cases continue to manage the land and home farm cultivation remained
minimal. Higher officials could accumulate manors of enormous size, but in turn had to commend
their properties to higher-ranking individuals to protect the immunity of their manor from tribute
requirements, leading to a complex tenure-hierarchy in which shares of manors and associated rights
to income were traded (Sato 1977). Around the end of the fourteenth century increasing land scarcity,
as evidenced by physical fragmentation of fields due to intergenerational transfers, led to a gradual
conversion to landlord estates (Keirstead 1985), which remained in place until the nineteenth and
twentieth centuries.
Java and Sumtra
87



Land market intervendons. The Agrarian Land Law of 1870 declared all uncultivated land
inalienable state property and leased it to European companies which established large scale
plantations.
Dferendal taxes and labor leves. These plantations were operated almost exclusively using
indentured labor (Breman 1989). Laws such as the 'coolie ordinance' from 1880 imposed severe
penalties on indentured workers who absconded and prison terms on anybody employing such
runaway workers, thus indicating the scarcity of labor (Stoler 1985). Large scale cultivation was
limited to these plantations. Where individual peasant holdings prevailed at the beginning of colonial
rule, authorities used the "cultivation system' (1820) to appropriate surplus without expending
resources for capital investment, and relying on traditional land tenure and labor exchange
arrangements. Ihis system required farmers to grow cash crops (predominantly coffee or sugar) for
the government on one-fifth of village lands in lieu of a land tax (Hart 1985). Both of these crops
were integrated into the local systems of rice or upland cultivation (Geertz 1963).
PhiHlppines
Land market Interventions. Land grants were given to private individuals and religious orders
after 1571 (Roth 1977) and by 1700 all of the best land was under the control of large estates
(Cushner, 1976).
Dfferential taxation and labor levkes. The Philippines, like countries in Ladn America, had
both encomlenda-the right to tribute in labor, cash, or kind from a particular region-and
repanlmlnto-which distributed workers for public works and private Spanish businesses. The
systems differed from those in Latin America, however, in that the right to labor services was
hereditary and often included whole villages. Workers on European haciendas were exempted from
heavy public works and from taxes, making hacienda employment highly attractive. Despite this
advantage, the lack of economies of scale led to almost immediate disintegration of rice-cultivating
haciendas into landlord estates. Moreover, by the nineteenth century, sugar production as well as
processing were controlled by tenants as well (Roth 1977).
Sri Lanka
Land market Interventions. Upland areas where slash and burn cultivation was practiced were
declared crown land in 1840 (Bandarage 1983) and sold to private cultivators, mainly British, who
established coffee plantations.
D&Terentlal taxation and iabor levies. Corvee labor was abolished on public lands in 1818 and
replaced by a grain tax amounting to 10 percent of gross produce. Export agriculture -all land under
coffee, cotton, sugar, indigo, opium poppies, and silk- was exempted from the tribute (Bandarage
1983).
While landed interests had successfully opposed the imposition of a general land tax, the
opportuniy to arn income from coffee cultivation, together with the absence of a totally landless
labor caste, severely limited the willingness of local people to supply labor to estates. Thus almost the
entire agricultural work force on coffee estates had to be imported: Census figures indicate that in
1871 and 1881, 97 percent of some 200,000 plantation workers were indentured Tamils, mainly from
India. The 3 percent of Singhalese plantation workers were mostly low-country artisans who were
88



paid competitive wages and used their position to accumulate capital for own land purchases
(Bandarage 1983).
Europe
Prussia
Land market interventions. Land grants in Prussia date from the thirteenth century and were
made to knights and nobles who were to colonize the largely unpopulated territory and provide
military services to the king. Initially, population density was so low that very favorable terms were
required to attract peasants: peasants received hereditary usufruct leases to about 32 hectares of land
each. Noble knights operated modestly sized demesnes of about two to three times the size which was
provided to settlers (Hagen 1985) to supplement the rents they received from peasants. They were
'not the master but the neighbor" of the farmer, and in economic terms they often fared worse than
full peasants (Luitge 1979). Depopulation caused by the Black Death increased the amount of land
available to the nobility who became 'land rich but labor poor". Productive use of this land could be
maintained only by attracting and settling new farmers, often on terms which were quite favorable to
the settlers.
Diferential taxation and labor levies. While settler farmers had a legal right to leave without
the lords' consent as late as 1484 (Hagen 1985), the Landesverordnung of 1526 no longer mentioned
the right of the farmer to take legal action against a landlord who would not allow him to leave (Abel
1978), indicating landlords' increased bargaining power (due to higher population density). Such
restrictions on peasants' mobility facilitated more widespread adoption of labor rents and an increase
in labor requirements from two days of service a week for full peasants in 1560 to three days around
1600 (Hagen 1985). Still, landlords had to rely on hired workers in addition to compulsory labor
services, estates were relatively small: In 1624, Junkers' demesne took up only 18 percent of the
cultivated land (Hagen 1985). The main benefit of labor services for landlords was the obligation of
full peasants to supply a pair of oxen or horses and a driver rather than the contributions made by
non-&ll peasants (nicht sparMlhge Bauern) to demesne cultivation.
Although landowners increased the size of their demesne by adding the land of families who
died during the plague years of the fourteenth century and the Thirty Years War of 161848, large
farms began to dominate in Prussia only after the land reform in 1807-50 (tge 1979). Three aspects
of the reform contributed to the emergence of large farms: the terms of separation requiring farmers
with hereditary or nonhereditary lifetime leases to cede one-third or one half their land to the Junkers
in return for freedom; the initial limitation of reform benefits to "full peasants* and its extension to
other peasants without long-term lease rights only in 1850 when, most people agree, it was "already
too late" (Dickler 1975); and repeal of tenancy protection laws, which had been in place since 1750.
These factors allowed Junkers to vastly increase their demesnes and to draw on an increased pool of
wage labor. The typical Junker style of cultivation with permanent laborers residing on house plots
emerged as the predominant form of production organization (Ltltge 1979). After farm workers
became free to migrate in 1868 and began moving westward (Wunderlich 1961), they were gradually
replaced by salaried and migratory seasonal workers, especially from Poland, where population
density was high and landlessness was widespread (Diclder 1975).
Input and output market Iterventions. From the earliest settlement days, knights had certain
rights of jurisdiction and monopolies on milling and on the manufacture and sale of alcohol.
89



However, the fact that they were willing to cede a good deal of their trade-related privileges to
entrepreneurs who engaged in land-clearing and attracting settlers from the west illustrates just how
pressing the labor scarcity was.
Russia
Land market intervendons. In the fourteenth century, princes, considering all land in their
princedom as their patrimony (votchina), granted land to nobles who could provide the labor force
necessary to cultivate the land and pay taxes. These landlords in turn had to attract peasants with very
favorable terms. In-kind payments (obrok) remained the predominant type of peasant obligation, and,
due to the limited ability to impose labor rents (barshchina), home farm cultivation was almost
nonexistent (Blum 1961).
In 1565, Ivan IV confiscated the property (votchina) of almost all the old princedoms,
converting it into state land (oprichnina) and then using it for land grants to reward servitors.
Servitors did not receive freehold title, acquiring only usufruct rights under service tenure (pomestye)
which became the dominant form of lay seignorial tenure. As a result, 'the personal possession of
landed property became a monopoly of a single class of Russian society-the servitors of the tsar'
(Blum 1961:169). As land rights could be terminated at will by the tsar, continued possession of the
land was conditional on the performance of service to the state. Indeed, landlords who could not
provide payment in service or money were evicted, and the class of servitors was subject to high
fluctuations, competition for labor was fierce, and home farm cultivation remained very limited. The
economic situation of the servitor was often precarious until tenures gradually became heritable in the
seventeenth century (Blum 1961).
Restrictions on Jabor mobility and dWerenata ta.taon. The extent of labor scarcity is
illustrated by continuous!y more severe restrictions on peasant cultivators' mobility. Between 1400
and 1450, the right of peasants to terminate leases and move on to another landlord was restricted to
two weeks each year. Even then peasants were required to pay formidable  "exit fees" (equivalent to
300 bushels of oats or 120 bushels of wheat; Blum 1961) before leaving. Landlords competed fiercely
for labor and resorted to 'labor pirating", i.e. attracting workers from other estates by promises. In
fact, such labor pirating became "the principal lawful way by which renters transferred from one lord
to the other", though illegal means were often resorted to as well (Blum 1961). The intduction in
1588 of "forbidden years" during which the peasants' right to move was temporarily suspended did
not prevent labor pirating because the law could not be enforced. Decrees in 1597 and again in 1607
bound all peasants to the place they were residing at the time of the census of 1592, which facilitated
enforcement of the law. The Assembly Code of 1649, which remained valid untfl about 1850,
abolished statutes of limitation on the return of fugitive peasants to their original landlord. It also
made serfdom heritable by prohibiting the peasant's wife and progeny from moving as well. After
1661, fines for peasant raiding had to be paid 'in serfs": for every iUlegal peasant found on a
landlord's holding, the landlord had to give up one of his own serf families. Serfs could be freely
sold; restrictions prohibiting the sale of serfs without land were unsuccessful. Serfs were also used as
collateral, to be auctioned off if their landlord went bankrupt. In 1859, two-thirds of all serfs were
mortgaged. After 1719, the privileges of peasants-mainly at the frontiers-who had escaped serfdom
were successively eliminated. They became serf-like state peasants, subject to taxes, quitrent, and
conscription. By 1850 more than 90 percent of the male population were serfs (Blum 1961).
90



In 1580, landlords' home farms (demesnes) were exempted from taxaticn. With revenue
requirements also rising, the tax burden on peasants increased substantially, significantly lowering the
potential return from cultivation (Blum 1961). Peasants responded by running off to the frontiers
where landlords were keen to attract labor and, because of temporary exemptions from taxes, were
able to offer better conditions.
Landlords attempted to tie peasants to their holdings through debt peonage. Under laws passed
between 1586 and 1597, a debtor automatically fell into debt servitude if he was unable to repay the
loan on time. He then had to work continuously for the creditor just to pay the recurrent interest.
Without any possibility of repaying the principal, debt servitors' only advantage over slaves was that
they were to be freed following the creditor's death (Blum 1961).
Input and output market Interventions. Since neither serfs nor state peasants were allowed to
engage in independent business until the 1820s or 1830s, landlords enjoyed a de facto monopoly over
commerce in their area, in addition to their formal monopoly on alcohol manufacture and sale.
Latin America
Chile
Lnd market Interventions. In the mid-sixteenth century, town councils, free of the central
supervision by a viceroy or governor that was common in Mexico and Peru, handed out land to
settlers 'with utmost generosity and ... in the face of royal legislation to the contrary" (Bauer
1980:4). In contrast to other Latin American countries, where the right to tribute was legally
distinguished from land grants, and the de jure protection of Indian communal land was enforced by
central authorities, encomenderos in Chile received land grants in the middle of "their" Indians'
communal lands early on. The encomenderos were thus provided with cheap and abundant labor
services such that 'by the 1650s landownership and encomienda were filly integrated ...[andJ the
encomlenda was absorbed by the land' (Bauer 1980:8).
Dfferentlal taxation and labor levies. Ihe main means to provide labor to the mines was the
mta which required all Indian settlements to supply a certain proportion of their labor force for
agriculture or public works, but in most cases the mines. Hacienda workers were exempt from the
mfta and many Indians sought refuge from the cruel forced labor requirements by joining the ranks of
the yanaconas, a group which had given up all ties, including land rights, to their original
communities and, living in total dependence on individual Spaniards, formed the nuclear labor force
of the Spanish estates.
A rise in demand for wheat from Lima in 1687 led to a considerable increase in such labor
requirements with landowners relying on either reconstituted encomlenda or on yanaconas who were
virtually enslaved and only given 3 days off a year to tend their house-plots (Pearse 1975). As on the
Eastern European Junker estates, able tenants were used as "labor brokers" and obliged to supply the
hacienda with workers (veones obligados or reemplazantes) nearly year-round (Kay 1977).
Input and ouwput market Interventions. Large wheat grownng farms in the Central region could
not compete against wheat produced on the more dynamic (and smaller sized) farms in the South and
were converted into livestock ranches. In order to protect them from competition from Argentina they
91



lobbied successfully for the imposition of import taxes on beef at the end of the 19th century. Such
taxes were maintained despite consumer riots caused by high food prices in 1905 (Kay 1992).
In this century, large landowners received special treatment to reduce the cost of mechanization. They
received exemptions from import tariffs and low interest rate loans; real interest rates on
mechanization loans in most of Latin America during the 1950s and early 1960s were actually
negative. Farmers in Chile, Argentina, Brazil, and Venezuela paid back only 50 to 80 percent of their
equipment loans (Abercombie 1972).
El Salvador
Land market Intervendons. Public land was granted to anybody who was planting it at least two
third with coffee from 1857 (Lindo-Fuentes 1990). A large land titling program, initiated in 1882,
which was intended to speed up the growth of coffee production, is thought to have directly affected
up to 40% of the territory of the country (Lindo-Fuentes 1990) and led to extraordinary concentration
of land ownership. The 1882 law required all occupants of ejido lands to register their claiLs (i.e.
prove that they were cultivating the land and pay the titling fee) within a period of six month. All
lands not claimed in this way was to be sold at public auctions. Illiterate Indians, were often not
aware of these requirements and well-connected individuals could take considerable advantage of the
legislation. The goal of establishing a successful export agriculture could have been achieved by
modernizing the credit system and providing education to Indians as well, in particular as Indians had
proven to be responsive to market incentives before. Choice of the land market as the instrument to
achieve the trnsformation illustrates the administrative difficulties as well as the power of the elites
who would benefit from such legislation (Lindo-Fuentes 1990).
DIferental taxation and labor levies. In 1825 vagrancy laws were passed requiring Indians to
carry work cards certifying their employment (Lindo-Fuentes 1990). The penalty for vagrancy was
Imprisonment. In 1847, landowners planting more than 15 000 coffee trees obtained exemption from
public and military services for themselves and all their workers.
Guatemala
Land market lntervenlons. While the Spanish made some land grants in Guatemala in the early
sixteenth century, their main land market intervention was resettlement of the Indian population in
centalized villages to facilitate tax administration and conversion of Indians to Christianity. They
limited their activities to ranching for which no land title was required (MacLeod 1973). Titles, which
were issued to Spaniards through land grants, became important only in 1590-1630, following a shift
to culdvation of indigo.
Dferential taxaton and labor kWes. Initially, Spaniards had little interest in establishing
intensive agriculture and collected tribute instead (such Indian tribute contributed more than 80
percent of royal government revenue; Brockett 1990). From 1540, tribute assessments were made in
cash, and the need for cash income was an important force inducing Indians from the highlands to
migrate to plantation areas 'MacLeod 1973). By the 1560s and 1570s, Indians who had migrated from
the highlands in this way constituted the majority of the coastal Indian population.
Beginig around 1600, Idian headmen were required to provide labor contingents
(mandamlento)-which could be as high as a quarter of the work force-for tasks of public interest
(MacLeod 295). Mandamlento labor was ideally suited to the seasonal demands of indigo processing.
92



Employment of Indians in indigo factories was widespread, despite its legal prohibition to prevent
futher decline of the decimated Indian population (Lindo-Fuentes i090). The mandamiento system
survived well into the 1880s, when it was used to provide cheap labor for European coffee plantations
(Cambranes 1985).
Debt peonage was legalized in 1877, and by forcing debtors to work off their debts, provided
landowners with official means of enforcing the continuation of a flow of cheap labor. Following the'
abolition of debt peonage, vagrancy laws were adopted in 1933 in response to the severe labor
shortage. All Indians who could not prove owner-operatorship of a minimum of 1.1 to 2.8 hectares of
land were forced to work-mainly on plantations-for 100 to 150 days a year to discharge their "debt
to society." The requirement to carry work cards facilitated enforcement (Pearse 1975).
Mexdco
Land market interventions. Resettlement of Indians beginning in 1540 deprived them of their
traditional lands and placed them on smaller, less productive holdings. While the intention of the
resettlement program was primarily to raise money for the crown by selling the Indians' land to
Europeans, the expropriations seriously reduced the productive basis of the Indian agricultural
economy (Gibson 1965; Taylor 1988).
Communal lands were expropriated in the 1850s, and as land became increasingly scarce,
fewer alternative opportunities were open to potential tenants. "The expropriation of communal
villages brought about two contradictory tendencies. On the one hand, cheap temporary labor became
more readily available than before. This made It economically less and less necessary for the hacienda
in central Mexico to rely on forced labor. On the other hand, as the haciendas acquired more and
more land, much of it of mediocre quality, they preferred not to work it themselves but to shift the
risk to sharecroppers and tenants. The condition of these occupants was so precarious that many of
them ... inevitably incurred debts with the hacienda which they could not repay" (Katz 1974:41).
DferentoJ taxation and labor levies. Spanish settlers received, after 1490, encomlendas, i.e.
rights to Indian villages from which they could etrac tribute in kind and labor services. Restrictions
limiting the use of tribute labor in agriculture were imposed in some regions, in order to secure labor
supply for public works.
In 1542, the original encomlendas were restricted to the right to collect tribute and the system
of repartinento was used to distribute Indian labor, supposedly in a more equitable way. While this
restricted the power of the original beneficiaries of the encomlenda, it worsened the lot of Indians
who still had to pay tribute to encomenderos and to render labor services under repartimientio.
Tribute requirements remained in place but could be avoided by working on haciendas (the
hacienda paid the tribute). Tribute was often required to be paid in cash, forcing many highland
Indians to migrate to lowland areas to obtain the necessary cash income (Moerner 1978).
Debt peonage was not significant in the early period of colonization, but it later acquired
importance as a means of tying laborers to the hacienda and lowering their wages. In 1790, 80
percent of peons in one area had a wtal debt higher than the legal limit; their average debt was
equivalent to eleven months' wages (Taylor 1972). As landlords let debt accumulate up to the point of
the expected future value of work performed, the system came very close to slavery (debt peons were
93



even being traded by redeeming the debt to their current employer). A law enacted in 1843 secured
not only state enfocement to Ocollect* debts incurfed to haciendas but also made it illegal to hire
laborers who had left their hacienda without paying their debts and required that they be returned
(Katz 1974). Vagrancy laws passed in 1877 and strictly enforced led to a considerable increase in the
employment of deportees and 'criminals" (Katz 1974).
Viceroyality of Peru (present day Peru, Bolivia, and Ecuador)
Land market interventiois. Beginning in 1540, land grants became common in this region, with
grants of 120-800 hectares being relatively easy to obtain. lhe main beneficiaries were the
encomenderos, i.e. Spaniards who had received rights to labor services from whole villages (see
below), since without Indian tribute labor to work the land, the latter was virtually worthless. Once,
all the land set aside for this purpose had been exhausted, around 1557, "private' Indian land was
expropriated and distributed among Spaniards (Gonzales 1985; Dav.es 1984).
In the coastal areas, resetement under Viceroy Toledo in 1570 moved Indians into newly
establisbed town-q where they were assigned farmlands of often inferior quality. Programs to review
existing Spanish land titles under which 'Spaniards could legally acquire land that they had previously
stolen from Indians by paying a fee to the Crown" (Gonzales 1985:15) were introduced in 1589. In
1641 the same pattern was applied even more rigorously to improve the financial position of the
Spanish crown: there were large-scale expropriations of Indian land, and all surplus land was sold to
Europeans. Indians 'suffered a considerable reduction in their holdings; they now possessed some of
the worst farmland in the valley" (Davies 1984:130). In the Arequipa Valley, adult married men were
allotted an area of only about half a hectare.
Diferentlal taxaton and labor levies. Beginning around 1530, the encomlenda conferred rights
to tribute (in labor, cash, or kind) from a particular region to Europeans, who replaced local
overlords. Holders of this privilege (encomenderos) were, at least at the beginning, completely
unregulated as to how much or what form of tribute to assess(Ramirez 1986). While many used lnbor
tributes to cultivate large farms, assessment of tributes in cash did reportedly force Indians to borrow
funds and sell off abandoned lands to repay their debt (Davies 1984). The right of individual
encomendwros to the exclusive use of Indian tribute labor for personal services was abolished about
1550, mainly to free labor for public works and the mines. The other benefits of encomlenda
remained, however.
With the abolition of encomienda, the Spaniards transformed the mita, an Incan institution for
recruitng labor for public works projects, into a permanent labor-recruitment arrangement for the
mines. In addition to paying tribute to the encomendero, each village had to supply a percentage of is
work force for "public works," which mostly meant work in the mines. As work for Spanish
haciendas exempted from the mlta and tribute requirements, many workers in the aldplano are
reported to have accepted work on haciendas. The class of yanaconas, who were resident on
haciendas and had completely abandoned their tribal identities, emerged (Pearse 1975).
Slavery was extensive after 1580 in the coastal valleys for the production of sugar, cotton, and
wine (Davies 1984). When slavery was abolished, sugar plantations resorted to indentured labor from
China and Japan, which comprised more than 90 percent of the work force on some estates (Gonzales
94



1985). Oth&r crops, predominantly cotton were, however produced under tenancy contracts (Gonzales
1991) after slavery was no longer availble, suggesting that this form of labor was more profitable
thar. farmnng the area under large farms.
Africa
Algeria
Land market Interventions. With the French occupation, all state, religious, and tribal land
became state property; uncultivated and waste land was subject to titling which allowed settlers to
acquire land at no price and 'amounted to little short than robbery' (Ageron 1991). In some cases,
such titling left the Muslims with slightly more than 5% of the land area and much of the land
declared waste included land grazed by nomads in the course of their migrations. Since the number of
setlers remained limited, various forms of settlement (mcludin, establishment of native villages) weie
tried to make the colony economically viable.
Ihe desire to impose French rule in Algeria after the 1870171 rebellion led to initiation of a
large colonization and setlement program between 1871 and 1882. At a huge cost to government,
settlers were provided free ?and and infrastructure but either sold out or farmed their land with native
sharecroppers (Ageron 1991). The so-called 'settlers' law' from 1873 allowed Europeans to acquire
rights to vast amounts of community land by purchasing a small share thereof and led to the
accumulation of vast estates at little cost (Ageron 1991).
D,fferenIa taxes and labor levies. Beginning In 1849, all Arabs had to pay head taxes from
which those working as sharecroppers or wage laborers on European farms were exempt (Bennoune
1988). Still, while 'they had always been willing to cultivate for the French as khammes or
sharecroppers", at the beginning of the 20th century only about 12% of Arabs were working as farm
laborers. French viticulturalists relied on foreign, immigrant labor from mediterranean countries.
Differential provision of credit to Europeans, led to rapid growth of vine cultivation. Market
fluctuations, together with additional land grants to te.a newly-rich settlers, led to the consolidation of
large estates of between 4000 and 5000 ha.
Angola
Land market lnterw'ons. In 1838 and again in 1865 all "unoccupied' land could be given as
concessions to Europeans. "The settlers were given lands, seAs, tools, and slaves by the government,
and measures were taken to ensure that their products could be sold" (Clarence-Smith 1979, 15).
From 1907 to 1932, 98 square miles were set aside for native reserves, 4 square miles were given to
Africans along with land titles, and about 1,800 square miles of the best land was given to Portuguese
settlers and other foreigners (Bender 1978).
Dfferentla taxation and labor leves. After the abolition of domestic slavery in 1875, slavery
continued in a variety of forms but due to tremendous demand for labor from the cocoa plantations of
Sao Tome, prices for slaves increased steadily, making it more profitable to export worker: than to
use them on inefficien' settler farms (Clarence-Smith 1979). Vagrancy laws passed in 1875 subjected
all "nonproductive" Africans to nonrmunerated labor contracts (Bender 1978). The laws were
replaced in 1926 by native laws, which provided for payments of wages but retained the provision
95



that all Africans bh-' to work for European landlords or could be contracted oy the state (Henderson
1980).
Egypt
Land market interventons. Land grants of the 1840s gave some 40 percent of the land to
Turko-Egyptian landlords and facilitated the formation of large estates (Richards 1982). Expropriation
of communal lands which took place in 1850-70, exacerbated this trend. Land taxes in 1856 (per
acre) were four to six times higher for smallholders than for the large land holdings (Richards 1982)
and in many cases large landowners did not pay taxes at all (Owen 1986).
Dtfferential taxation and labor levies. In contrast to their usual practice, the Ottomans in the
sixteenth century did not distribute Egypdan lands to military leaders but assessed collective tribute.
They wished to avoid disrupting agricultural production in Egypt, 'the granary of the Otte-man
Empire" (Richards 1983:7). Corv6e laborers were recruited initially for public works to set up an
extensive irrigation system and later for cotton production on the ruler's home farm. Following the
large land grants made in the 1840s, 'large landowners arranged to have corvee !aborers work on
their estates and to get their peasants exempted from the corv6e* (Richards 1982:23), thus closely
paralleling events on the Latin American hacienda.
Large landowners obtained coniiderable direct government subsidies for cotton-price
stabilization programs in the early 1920s and 1930s, supplemented by an official limitation of the
amount to be planted to cotton and financial support to lower interest rates for large landowners
which, by the 1930s, were heavily indebted. Similarly, imposition of tariffs on imported flour in 1932
and 1934 and protection of the market for domestically produced sugar, directly supported large
landowners (Owen 1986).
Kenya
Land market Intervendotos. With the arrival of Europeans, all vacant land was declared to be
Crown land and sold to European setders at extremely favorable conditions. Much of the land
continued to be farmed by African tenants, which were called squatters (Mosley 1983). Africans' land
rights were limited to reserves and a formal prohibition of African land purchases outside the reserves
was codified in 1926.
Djfferentl taxation and labor levies. The British introduced a number of regressive hut and
poll taxes in order to 'increase the native's cost of living (Bernan 1990:509). To pay these taxes,
Africans initially did not seek wage labor but increased production, mainly on tenanted land. Despite
repeated requests from settlers to grant tax-exempt status to Africans working on European farms,
such taxes had to be paid by workers as well, thus large estates based on wage labor remained
relatively unprofitable as compa. 4 to tenancy.
The squatter law from 1918 required tenants to provide at least 180 days a year in labor
services to their landlord at a wage not to exceed two-thirds of the wage for unskilled labor. This
ordinance was amended twice (in 1926 and 1939), both times increasing the minimum amount of
3F For more detail on Kenya, South Afiica, and Zimbabwe, see Deminiger and Binsa%nger (1992).
96



labor services (to 270 days por year in 1939), limiting the area sllowed to be cultivated as well as the
amount of stock owned per tenant, and making eviction of tenants easier. Labor passes, which had
beet introduced in 1908, limited the mobility of Africans; leaving without the employer's consent was
a criminal offense (Berman 1990).
Input and Qowput market lrervendons. A dual price system for maize, adopted in the 1930s,
reduced the returns African farmers could obtain for the same produce as supplied by their European
counterparts and, in addition, unloaded most of the price risk on Africans (Mosley 1983).
Grower associations that excluded Africans were formed for most of the important cash crops.
High licensing fees kept Africans out of pyrethrum production, and they were prohibited outright
from cultivating coffee (Berman 1990).
During World War II, European farmers received direct subsidies to mechanize their farms
(Cone and Lipscomb 1972).
Sokotho-Callphate (present day Burkina Faso, Cameroon, Niger, and northeri Nig,ria)
Land market Imeventons. After 1804, land was granted to settlers by tile caliphate
government in the areas around defensive centers, the amount of land depending on the number of
slaves owned. Thus "anyone with slaves could obtain enough land to start a plantation' (Lovejoy
1980). ITere were about 100-200 slaves per plantation, although there are reports of officials who
managed to obtain holdings of morr than 1,000 slaves (Lovejoy 1978).
Dterenti taxaton and labor leves. The pattern of 'slavery' in the area, which was
populated by Hausa and Fulani, was characteristic of many parts of Africa in the nineteenth century
(Lovejoy 1980).21 Slaves which made up some 50 to 75 percent of the local population were
acquired by warfare, direct seizure, or as tribute from subjected trtbes. Limited export markets and
the relatively low price of slaves Oandowners could replenish their bonded work force through
independent raids; Lovejoy 1980) allowed relatively lenient treatment of slaves who enjoyed more
rights e.g. the possession of heritable house-plots (Hogendorn 1977) and the right to self-redemption
often using fimds acquired by cultivating surplus land (Hill 1978) than the slaves acquired for cash by
market-oriented plantations in the Americas. Land and the absence of economies of scale meant,
however, that slave owners had to take measures to prevent slaves from escaping and establishing
their own operations (Hogendorn 1977). Eventually, these factors led to the demise of the large
holdings (Hopkins 1973).
Malawi
Land market lntervendons. In 1894, Europeans were allotted more than 1.5 million hectares, or
about 15 percent of total arable land.
"h IS is some diswsion in the littur an the approprite nomelatue for this system, which combines
e _mnt of avauy ud sefdom.
97



Dferenial taxadon and labor levies. Attempts to introduce labor tenancy on European-owned
cotton lands were unsuccessful as farmers abandoned the land and fled to uncultivated crown land.
The situation improved only as a law was introduced in 1908 which allowed Africans to gain a
significant reduction in the head tax they had to pay by working for European cotton growers for at
least one month a year. Africans' possibility to gain a similar reduction of the head tax by producing
cotton on tenanted land, was, due to landowners' pressure, eliminated (Mandala 1990).
Mozambique
Land market Intervendons. Exclusive property rights in land and quasi-governmental authority,
were in the early 19th century, granted to lessees (often companies) for a period of three generations
under the institution of pra.o. The prazo-holder had to piovide minimal public services, cultivate part
of the property, pay quitrent and tithe, but could levy annual tributes (in cash, kind, or labor) on the
local population and (see below) was endowed with a complete monopoly on all trade within and
outside the area (Vail and White 1980).
DigerentIa taxation and labor levies. HIut taxes were established in 1854. After 1880, at least
ha!f of die tax hid to be paid to the local prazo-holder in the form of labor services (Vail and White
1980).
Under the vagrancy law of 1899, all male Africans between fourteen and sixty years old were
legally obliged to work. The area of crops to be grown or the wage-employment required to satisfy
this obligation could be varied by local prazo-holders, providing them with ample instruments to
increase the supply of labor. Contingents of migratory labor were often 'sold" to other areas (such as
South Africa) where labor was relatively scarce (Vail and White 1980). Vagrancy laws were repealed
in 1926 - at about the time many prazos were expiring- and the use of forced labor for 'private
purposes" (i.e. non-quota production) was banned. The labor code of 1942 instituted an obligatory
labor requirement of six months for all African men.
Input and output market Inerventions. In 1892 all itinerant African trade within prazos was
abolished, conferring a monopoly onprazo-holders of all commerce in their prazw (Vail and White
1980:132). Prazos turned into a kind of mini-state, each with its own closed economy and unlimited
freedom for the prazo-holder to determine the terms of trade. Deprived traders to provide outlets for
their produce 'that had made peasaat production so attractive to the local people' Africans almost
completely withdrew from cash-crop productions and the prazos became 'private labor pools from
which the companies, by direct force or by indirect manipulation of the economy, could compel the
labor they required" (Vail and White 1980:132). Following their expiration about 1930, prazos were
replaced by a 'concession system". Concession holders received monopoly rights to purchase cotton
and rice at state-administered low prices from African growers in return for enforcing Africans' work
obligations and providing inputs and supervision (Isaacman 1992). Although exactions from Africans
were still high, (forced) cultivation of all but sugar reverted to smaller scale units rather than large
scale farms.
South Africa
Land market Intervendons. Native reserves were firmly established at the end of the 19th
century although they were legally defined only in 1912. For example in Transvaal in 1870, the area
allocated to African reserves was less than a hundredth of the area available to whites (Bundy 1985).
98



The Glen Grey Act (1894) restricted African land ownership in the reserves to a parcel of no more
than about 3 hectares and instituted a perverted form of "communal tenure' which banned the sale,
rental, and subdivision of land in order to prevent the emergence of a class of independent African
smallholders (Hendricks 1990). The inability to sell land in the reserves, which persists up to this
day, is recognized to be major reason for the low productivity of agriculture in the homelands (Lyno
and Nieuwodt 1991).
Various legal measures to discourage tenancy on European farms such as a limit on the amount
of tenants per farm in 1895 and assessment of license fees for tenants in 1896 4id not lead to the
desired results. The Native Lands Act (1912), circumscribed the extent of African reserves and
declared real tenancy on European farms illegal, forcing all African tenants to either become wage
laborers or labor tenants on European farms or to move to the reserves.
D,ferentda taxes and labor levies. Prior to state intervention on their behalf, very limited
market produc.aon by European farmers was based on slaves or, after the prohibition of slavery in
1834, indentured labor.
Masters and Servants Laws and the Mines and Workers Au (1911) restricted Africans'
occupational mobility and excluded them from skilled occupations in all sectors except agriculture
(Lipton 1985). Restrictions on mobility were reinforced and tightened by pass laws (influx controls)
from 1922 and the establishment of labor bureaus to enforce the legislation from 1951 (Lipton 1985).
I addition to restricting Africans' ability the obtain jobs outside agriculture, more rigid pass
laws and rigorous enforcement of such laws also provided a flow of cheap labor for white
agriculturalists. It is estimated that, in 1949, about 40 000 pass-law offenders were supplied to fas
as prison laborers (Wilson 1971).
Input and output market nterventions. European farmers were assisted by a large array of
monopolistic commodity marketing boards and direct credit subsidies. In 1967, the amount spent on
subsidizing about 100,000 white farms was almost double the amount spent on education for more
than 10 million Africans (Wilson 1971).
Tanganyika (part of present day Tanzania)
Land market Inervendons. From the late 1890s until 1904 it was common practice to allocate
several villages apiece to incoming German settlers.
Dfferendal taaton and labor levies. A hut tax, to be paid in cash or labor services, was
imposed in 1896 'not so much for the revenue which resulted but as a means of propelling them into
the labor market" (Rodney 1979, 131) although half of the hut-tax income went direcdy to settlers'
District Councils. Vi'llage headmen were required to provide a fixed number of workers each day to
provide labor for the settlers to cultivate their rubber and sisal plantations. Every African was issued
a work card that obligated him to render services to an employer for 120 days a year at a fixed wage
or else to work on public projects (Illife 1979). In 1902, the Germans introduced compulsory cotton
production in certain coastal areas; it is widely accepted that this scheme was one of the main causes
leading to the outbreak of the Maji Maji revolt in 1905 (Coulson 1982).
99



Africans were excluded frtm credit by the Credit to Natives Ordinance of 1931 which required
that an Afican have specific government permission before he could even request a bank to lend him
money (Coulson 1982). Attempts by Africans to set up a marketing cooperative for coffee led to the
attempt to outlaw traditional practices of coffee growing in 1937, which led to riots. Settler-dominated
marketng monopolies for African-grown crops were set up in the 1940s and creamed off most of the
profits from those crops (Coulson 1982).
Zimbabwe
Land market Interventons. Reserves for Africans in remote areas of often low fertility were
established in 1896 although their boundaries underwent some changes until 1931 (Palmer 1977),
when African land purchases outside the reserves mnd specifically designed 'African Purchase Areas"
were declared illegal.
D&erendad taxadon and labor levies While all Africans were subject to poll and hut taxes,
specific taxes discriminated against cash rental an;' share tenancy contracts fom 1909 (Palmer 1979).
The prospect of (temporarily) easing the tax load led to large-scale migration of Africans into the
reserves when commodity prices were extremely low in the early 1920s (Arrighi 1970).
Input and output market Interwntions. Volatility and downturns in output markets were
smoothed by government interventions such as increased land bank loans, debt moratoria (especially
during the depression in 1930) and, after protracted lobbying by European producers, the
establishment of monopoly marketdng boards (for tobacco, dairy, pigs, and cotton)in selected crops
and the establishment of export subsidies.
African maize and livestock producers were discriminated against by dual price systems.
Pressure by European miners who were interested in cheap supplies of maize limited the extent of
price discrimination against African producers in maize. Quarantine-based restrictions on Ailcan
livestock sales initially led to the buildup of large herds and the associated soil degradation in the
reserves. To ease this problem, in 1939, compulsory destocking was mandated; prices paid fer
African cattle were between one third anWi one sith of the prices fetched for comparable European
stock (Mosley 1983).
100



Appendix 2
Now Market Imperfections Affect the Farm Size - Productivity Relation
Conrider a region wh  ewh farm houehold conmsits of E fmiy membaer capable of
conducting farm opertons as wel as supevising hie work of hired laborers." The household owns y acres
of land, but the size of farm it actually operates, denoted by A,, is determined through renting in or
renting out land at the going rental rate & Output depends on effective labor lid and land M.
Effective labor is defined as the product of the number of individuals employ -d and the effort i1 they
exert. While family members can be expected to perform farm tasks with maximum effort, say i,
hired laborers' work effort depends on the intensity of supervision. The intensity of supervision is
represented by the ratio of household members to operational farm size (M/0). It is assumed that the
marginal returns to supervision intensity are diminishing,
e = e(FIA), e'> 0, e' < 0, lim e=1                                              (1)
FIA -_oo
With N hired laborers per operated acre and a total of E household members, the
effective labor input is given by
L = F.e+A N-e(FIA).                                                             (2)
Output is determined by a neoclassical production function that depends on effective
labor and land,
Q=Q(L, A).                                                                      (3)
Assuming constant rets to scale, and substituting equation 2 in eciation 3, output
per operated acre is given by
q=Q[1.(FIA)+N*e(FIA);lJIq[I- (FIA)+N  e(FIA)J,                                  (4)
1/ This ppeadix is based an Feder (1985).
101



where q-Q/A and q'>0, q' <0.
A simple but realhitc way to introduce a credit market Imperfecdon to the present
model is to assume that the supply of credit depends on the amount of land owned by the household
denoted E&
S mS(VeS'Jr.
(5)
With $* wage rate denoted by w, Intermediate input costs per acre by c, and cash
consumption expenditures per family member during the season by 0, the cash requirements of a
family wih an opwational holding of size A are w.N.A+c -A+R.(A - V)+ U*F, and the working
capital constaint fced by the farm Is:
w.N.A+c.A+R (A - v)+eOF-sF v).                                                   (6)
T7he aumr's objective is to maximize end-of-season profits (accounting for interest
charges  per dollar borrowed), subject to the working capital constn. Formally,
max 1 =-q[e U(F/A) + Ne(F/A)l
A,N
4w.N.A + c.A i R.(A-VJ(l+O,
subject to inequality (6) and  20, N20.
Detingto lU an tficda   [+k* S(1       w*N.A-c.A-R.-(AV)4.F1, whA   is
ft shaow przm of ft creit cons_ant, t  Kuhn-Tcice cod      for opmid,on iply:
a^  -q-q 'fF/eA) +N(FIA) e    -(w-N+c+A (1 +L +A) A,
at
(74)
ad
102



A!q-w(.+ks.(82)
A  cIN   -W li§ )O
N N4                                                                                       (8b)
a8N
(9a)
"1-S(v)-w *AJf -c'A -Rv-P)-B&                                                           (9a)
kA 1 -0,                                                                              (9b)
AkO,>N20,120,                                                                              (10)
We stai with Iho co  in which the credit coaint is not binding( X-0); solving st-order
conditon (7a) ad (8.) for th optmal value of A and iF and diffe  tating yie:
_ A
ad4
dF F
(11)
-m0.
dF
(12)
Equat  (11) implies tot in to absc of biwng redit constraints, te elucity of fth
opta opeaonal uize with repect to houhold size s unity, i.e., thr is a fixet     tiop nl holding to
householdsize rtio. Mm amount of owd land dos not affect the optimal ratio. This outocom is intuitively
expected in a situation of constt rturn to scale with prfect ntal  d capital marts.
103



t
Equation (12) implies that the optimal number of hirnd laborers per acre is not affected by
household size (neither is it affected by the size of the owned holding). Since the earlier roLults imply that the
operational holding is proportional to household size, it folows that the number of hired leborers per acre is
identical on ull farms, whatever the size of the operational hutding (a. 4 that the ratio of family to hired labor
declines with operational holding size). A trivial extension of thes rslts is the observation that the level of
effective labor per acre is identical on all farms (since the ratioZ/A is fixed and A is the sam  on all farms),
assuming all other farm and farmer attributes are identical. It therefre follows that output per unit of land
operated is uot affected by the size of the operational fann or by the amount of land owned.
The analysis and the presentation in the case where the credit constraint is binding Q\>0) are
greatly simplified by assuming that the finctions q( * ) and e( * ) are of lxed elasticity with respect to their
arguments, that is, that (q'he,) * (LIA) * il, the elasticity of outut with rspect to effective labor, and (e'/e)
(FIA) *  i, the elasicity of effort with respect to upesion, and where
dA                   WR   <-
dV                    [             w              e
a and V are parmet  within the intew'l (0,1). Th& stadard treatment of labor in the literaure - the
assuption that hired labor is not affected by family supervision - is then the special case n0 in the presnt
model.
Differeiation of equations (7a), (8a), (9a), under the assnuption of an intmal solution,
yields after some manipulaton
(13)
d4  -(1 -1  -11  wF.
_                                  w
dV      Ils    u  (c + R)    Is -(  ZcF
w                   eA
The denominator ow be dwsw  to be positive if second-ord  conditions hold. It follows that
th sig of equation (13) is detmined by the sip of (I-" , j), which is the limit value of total output
elasticity with Lrespect to land as the share of family labor tends to zero.
To demonstrate that the relation between per-hectars yields and openrional holding size can
follow different pattns within the frmnework of the preset Laodel, we use te definition of effective labor and
the fist-order conditions to calculate the optimal per-ectare input of labor:
104



(IJA)*-v f[c+R) ./wa JIaJS    F/AJ(1(I-"1 *  .).                                             (14)
Diffrentiation of quaton 14 with respect to owned holding sie y yields
d(14A) ;*                    e di
dV           e A  w  At(1I1iL)
(1S)
Clarlny, if the labor mart is pedfect (a.0), labor per hectao of land dos not vay with
fam style. Ibspection of equaton 13 verifies that a and the sig of equaton 1S thus depends an th term in
.squr brwket.
In the cas where 17-7i * I  0, the relafion between the offecive labor iput per hecta  and
owned holding size can be negative or positive. Consider, for instance, tho cas wher the output Iasticitv i
equals '. Firt-orde conditions imply [(1- *     (Ig)**A)-[( I  <1Jc0, hec, in t  ca  where  i-'A, it
follows tdat d(/)Idi/<O, i.e., tho effective labor input (and yieds) declin  with owned holding size. Mhm
ame result can be obtaned for aiwll <. By an avWment of coStinuity since in the cuo (l-1p-0 it holds
that dl&4/)I_>0 (in that ca  the  is a final c.mional fam sa of weth), ther mustexis
some low (but positive) values of the tem (1-""- it) for which d(ld)l/d_>0 holds. The conclusion is,
therefore, that one may observe a posiiive or a negative reaton betwee opetional holding siz and per-
hect  yields, dep&udig on the relafive magnitudes of ti and p. In tho case (1-t  *p)-0 ther wiUl be no
oetion betw    operational holding size and per-hwetac yields.
105



BlbHogaphy
Abdullah, A. (1974) 'Land reform and agrarian change in Bangladesh', Bangladesh Develogment Studies, ?:
67-99.
Aberlombie, K.C. (1972) 'Agricultural mechanzation and employment in Latin America', International Labour
RBev&. 106: 11-45.
Alston, L.J., Datta S.K., and Nugent, J.B. (1984) 'Tcnancy choice in r competitive framewcrk with transaction
costs', Journal of Political Economy 92: 1121-1133.
Asako, K. (1991) 'The land price bubble in Japan', Ricerthe Economiche, 45: 167-184.
Aston, T.H., and Philpin, C.H.E. eds (1985), 'he Brenner debate: agrarian class structure and economic
deelgDMent in greijndustrial Europe. New Yv.k: Cambridge University Press.
Atwoou, D.A. (1990) 'Land registhation in Africa: the impact on agricultural production', World Development,
18: 659-671.
Barham, B.L., and Childress, M. (1992) Membership desertion as an adjustment process on Honduran agrarian
reform enterprises,  enonoMic .Daftment and Cultural Change,:S87-613.
Barraclough, S, and Collarte, J.C. (1973) Ama :  A reume of the CIA I
tenure studies of Argenti. Brazl. Chile ColoMbia, Ecuador, Guatemala. and Peru. Lexington Books
Barrows, R., and Neuman, M. (1990) A review of experience with land zoning. Land Tenure Center mimeo.
Barrows, R., and Roth, M. (1990), Land tenure and investment in African agriculture: theory and evidence,
Jounal of Modern African Studies, 28: 265-297.
ell, C. (1977) 'Alterative theories of sharecropping: some tests using evidence from northeast India, Lrmal
ofDeveloWmen Stdi , 13: 317-346.
Bell, C. (1988) 'Credit markets and interlind transactions, in: Chenery, H., and Srinivasan, T.N. eds-
Handbook of develo=meeonomics Vol I. Amsterdam, North-Holland.
Ben-Ner, A. (1984) 'On the stability of the cooperative type of organization, Journal of Conmarative
E.uD.mis, 8:247-260.
Benjamin, D. (1992) Household composition, labor markets, and labor demand: testing for separation in
agricultural household models, Ecnomerc 60:287-322.
Bentley, J. (1987 Economic and ecological approaches to land fragmentation: in defense of a much-maligned
enomenon', Ann. Rev. Anthronoj, 16:31-67.
erry, RA., and Clue, W.R (1979) Airan structure and producvit in developing countries. Geneva:
ternatonal Labor Ornizaon.
106



Bhaduri, A. (1976) 'The evolution of land relations in eastern India under British rie', Indian EconoMic and
Socia Hisgt Revie-w-, 13:45-58.
Bhaduri, A. (1986) 'Forced commerce and agrarian growth', Q[rld Deve.agmet,14:267-272 .
Bhagwati, J.N., and Chakravarty, S. (1969) 'Contributions to Indian economic analysis: a survey', American
EconomiRevie,59:
Bhalla, S.S., and Roy, P. (1988) 'Mis-specification in farm productivity analysis: the role of land quality,
Oxford Economic Papers, 40:55-73.
Bidinger, P.D., Walker, T.S., Sarkar, B., Murty, A.R., and Babu, P. (1991) 'Consequences of mid-1980s
drought- longitudinal evidence from Mahbubnagar', Econoi and Poitical_YWlfy, A105-A114.
Binswanger, H.P. (1989) 'Brazilian policies that encourage deforestation in the Amazon', World Development,
19:
Binswanger, H.P., and El in M. (1988) 'What are the prospects for land reform', in: Maunder, A., and Valdes,
A. eds, Agriculture and eoverments in an interdependent world. Proceedings of the Twentieth International
Conference of Agricultural Economists, 1988.
Binswanger, H.P., and McIntire, J. (1987) 'Behavioral and material determinants of production retations in land
abundant tropical agriculture', Economic Develogment and Cultural Change, 36:75-99.
Binswanger, H.P., and Rosenzweig, M.R. (1986), 'Behavioral and material determinants of production relations
in agriculture', Journal of Development Studies, 22:503-539
Bird, R. (1974) Taxing agricultural land in develoging countries. Cambridge, Massachusetts, Harvard University
Press.
BIirel, B., Hazell, P., Place, F., and Quiggin, J. (1992) 'The economics of farm fragmentation: evidence from
Ghana and Rwanda', World Bank Economic Reviq, 6:233-254
Bliss, CJ., and Stern, N.H. (1982) MAMrnpnthe econoMy of an Indian villaj. Oxford, Clarendon Presr:
Blum, J. (1977) lhe end of the old order in rura Europe. Princeton, NJ.: Princuon University Press.
Boldt, R.A. (1989) 'The ffects of land dtle on small farm production in the highlands of E:'uador', Land
Tenre Center Research Paper. 30
Bonin, J.P. (1985) 'Labor management and capital maintenance: investment decisions in the socialist
labor-managed firm, in: Jones, D.C., and Svejnar, J. eds, Advances in the Economic Analysis of Participatory
and Labor Managed Firms. Vol I. Greenwich, Conn.
Bonin, J.P., and Putma, L. (1986) Economics of cooperation and the labor-managed economy,
Fundamentals of Pure and Applied Ecnomics 14. Harwood Academic Publishers.
Boserup, E. (1965) Coditions of Agriultural Gro.
107



Brading, D.A. (1977) *The transition from traditional bacienda to capitalist estate, in: Duncan and Rutledge.
Brandgo, A.S.P., and Rezende, G.C.de (1992) Credit subsidies, inflation and the land market in Brazil:
theemetical and empirical Analysis.
Braverman, A., and Stiglitz J.E. (1982) 'Sharecropping and the interlinking of agrarian markets, Ameri
Economic Review, 72:695-715.
Braverman,A., and Srinivasan, T.N. (1981) 'Credit and sharecropping in agrarian societies, lournalf
Development Economics,9: 289-312
Brenner, R. (1976) 'The origins of capitalist development: a critique of neo-Smithsian Marxism', New
Review, 104:25-92.
Br-  Zt, C.D. (1984) 'Malnutrition, public policy, and agrarian change in Guatemala', Journal of Interamerican
St1jies and World Affairs, 26:477497.
Brown, M.R. (1989) 'Radical reformism in Chile: 1964-1973, in: Thiesenhusen, W. ed., SearchingforAgri
Reform in Latin America.
Bruce, J.W. (1988) 'A perspective on indigenous land tenure systems and land concentration, in: Downs and
Reyna, eds.
Burt, 0. (1986) 'Econometric modeling of the capitalization formula for farmland prices', American Journal
nf Agricultural Economics, 68:10-26.
Caballero, J.M. (1983) 'Casual labor in Peruvian agrarian cooperatives, in: F. Steward, ed; Work, ince
Cain, M. (1981) 'Risk and insurance: perspectives on fertility and agrarian change in India and Bangladesh',
Population and Development Review, 7:435474.
Callison, C.S. (1983) 'Land to the tiller in the Melkng Delta: economic, social and political of land reform in
four villages of South Vietnam. University Press of America.
Cambranes, J.C. (1985) Coffee and peasants in Guatemala. Stockholm.
Carter, M.R. (1984) 'Identification of the inverse relationship between farm size and productivity: an empirical
analysis of peasant agricultural production' Oxford Ecunomic Papers, 36:131-145.
Carter, M.R. (1987) 'Risk sharing and incentives in the decollectivization of agriculture', Oxford Econoin
Eawers, 39:577-595.
Carter, M.R. (1989) 'The impact of credit on peasant productivity and differentiation in Nicaragua', kJounl
of Development Economics, 31:13-36.
Carter, M.L, and Kalfayan, J. (1989) 'A general equilibrium exploration of the agrarian question, Madison,
Wisconsin.
108



Carter, M.R., and Wiebe, K.D. (1990) 'Access to capital and its impact on agrarian structure and productivity
in Kenya', American Journal of Agricultural Economics, 72:1146-1150.
Castillo, L., and Lehman, D. (1983) 'Agrarian reform and structural change in Chile, 1965-79', in: Ghose,
A.K. ed., Agrarian reform in contmporaa develoging countries. New York; St.Martins Press.
Chavas, J.P., and Aliber, M. (1992) 'An analysis of economic efficiency in agriculture: a nonparametric
approach. mimeo.
Chayanov, A.V. (1991) The theory of peasant cooperati   Ohio State University Press.
Cheung, N.S. (1979) The theory of share tenancy. University of Chicago Press.
Chevalier, F. (1963) Land and societ in colonial Mexico: the ereat hacienda. University of California Press.
Christensen, G.N. (1989) Determinants of private investment in rural Burkina Faso. Dissertation, Cornell
University, Ithaca, New York.
Christodoulou, C. (1990) The unpromised land: agrarian reform and conflict worldwide. London: Zed Books.
Chuma, H., Otsuka, K., and Hayami, Y. (1990) 'On the dominance of land tenancy over permanent labor
contract in agrarian economies', Joural of the Japanese and International Economies, 4:101-120
Cline, W.R. (1970) Economic consequences of a land reform in Brazil. Amsterdam: North-Holland.
Colburn, F.D. (1990) Managing the commanding heights: Nicaragua's state enterprises. University of California
Press
Collier, P. (1989) 'Contractual constraints on labour exchange in rural Kenya', International Labour Review,
128: 745-768.
Cornia, G.A. (1985) Farm size, land yields and the agricultural production function: an analysis of fifteen
developing countries, World Development 13: 513-534.
Courtenay, P.P. (1980)      nacultur, Bouler, Colorado: Westview Press.
Davis, S.H. (1983) 'State violence and agraian crisis in Guatemala', in: Diskin, M. ed., Trouble in our
backlAr& Central America and the United States in the eighties. New York, Pantheon Books.
Delanvry, A (1981) The Agrarian Ouestion and Reformism in Latin America. Baltimore, M.D.; Johns Hopkins
University Press.
DeJanvry, A., and Sadoulet, E. (1989) 'A study in resistance to institutional change: the lost game of Latin
American land reform', World evelpumnt, 17:1397-1407.
Tanvry, A., Sadoulet, E., and Wilcox Young, L. (1989) 'Land and labour in Latin American agriculture from
Se 1950s to the 1980s', Journal of Peasant Studia16:396-424.
Deaton, A. (1989) 'Savings in developing countries: theory and review', Woodrow Wilson School, Princeton
109



University Discussion Paper 144.
Deaton, A. (1991) 'Household savings in LDC's: credit markets, insurance, and welfare', Woodrow Wile
School, Princeton University, Discussion Paper 153.
Deininger, K., and Binswanger, H.P. (1993) Rent seeking and the development of agriculture in Kenya, South
Africa, and Zimbabwe. World Bank Po'icy Working Paper.
Deolalikar, A.B. (1981) 'The inverse relationship between productivity and farm size: a test using regional data
from India', American Journal of Agricultural Economics, 275-279
Diba, B.T., and Grossman, H.I. (1988) 'The theory of rational bubbles in stock prices, Economic Journal,
98:746-54.
Dobb, M. (1963) Studies in the development of capitalism. London: Routledge and Keegan.
Domar, E.D. (1970) 'The causes of slavery or serfdom: a hypothesis', Journal of Economic History, 30:18-3
Don, Y. (1985) 'The economics of transformation from agricultural to agro-industrial production cooperatives:
the case of the Israeli Kibbutz' in: Duelfer, E., and Hamm, W. eds., Cooperatives in the Clash between
Member Participation, Organizational Development
Dorner, P., and Thiesenhusen, W.C. (1990) 'Selected land reforms in East and Southeast Asia: their origins
and impacts, Asian Pacific Economic Literature, 4:69-95.
Downs, R.W., and Reyna, S.P., eds., Land and society in contemnorary Africa. Hanover and Londo
Published for University of New Hampshire by University Press of New England.
Elder, J.W. (1962) 'Land consolidation in an Indian village: a case study of the consolidation of holdings act
in Uttar Pradesh, Economic Development and Cultural Change, 11:16-40.
Ellis, F. (1983) 'Las transnacionales del banano en centroamerica', Editorial Universitaria Centroamerican
San Jose, Costa Rica.
Eswaran, M., and Kotwal, A. (1985) 'A theory of contractual structure in agriculture', American Economic
Review, 75:352-367.
Evans, R.J., and Lee, W.R. eds. (1985), The German peasantry: conflict and community in rural society from
the eighteenth century to the present. London.
Falk, B. (1991) 'Formally testing the present value model of farmland prices', American Iournal of Agricultural
Economics. 73:1-10.
Fan, S. (1991) 'Effects of technological change and institutional reform on production and growth in Chinese
agriculture', American Joumal of Agrcultural Economics,:266-275.
Feder, G, Onchan, T., Chalamwong, Y., and Hangladoran C. (1986) Land policies and farm productivity in
nailand. Baltimore, M.D.; John Hopkins University Press.
110



Feder, G. (1985) 'To relation between farm size and farm productivity: the role of family labor, supervision,
and credit constraints', JIW of Develggment Economics18:297-313
Feder, G., and Feeney, D. (1991) 'Land tenure and property rights: theory and implications for development
policy', World Bank Economic Review, 5:135-155.
Feder, G., Onchan, T., and Raparla, T. (1988) 'Collateral, guaranties ard rural credit in developing countries:
evidence from Asia', &riculgal Economics, 2:231-245
Feeney, D. (1988) 'The development of property rights in land: a comparative study', in: Bates, R.H. ed.,
Toward a political economy of development: a rational choice persDective. University of California Press.
Feldstein, M. (1980) 'Inflation, portfolio choice, and the prices of land and corporate stock', American Joumal
of AgrclMWral Economics, 62:910-916.
Fenoaltea, S. (1976) 'Risk, transaction costs, and the organization of medieval agriculture', Exploratons in
Econominc Hisry, 13:129-175.
Finkder, K. (1978) 'From sharecroppers to entrepreneurs: peasant household production strategies under the
ejido system of Mexico, Economic Development and Cultural Change.
Florescano, E. (1969) 'Precios del maiz y crisis agricolas en Mexico (1708-18 10): ensayo sobre el movimiento
de los precios y sus consecuencias economicas y sociales,' Mexico.
Fogel, R.E. (1971) 'The reinterpretation of American economic history' New York, Harper and Row.
Forster, N.R. (1992) 'Protecting fragile lands: new reasons to tackle old problems', World Development,
20:571-585.
Fujimoto, A. (1988) 'The economics of land tenure and rice production in a double-cropping village in southern
bhailand', Develg,ing Economi , 26:189-211.
Gerschenkron, A. (1965) 'Agrarian politics and industrialization in Russia, 1861-1917i', Cambridge Economic
Hig= of E=u    . Vol 6.
Ghai, D., Kay, C., and Peek, P. (1988) 'Labour and development in rural Cuba', Geneva ILO: and Macmillan
Press.
Glover, D. (1990) 'Contract farming and outgrower schemes in East and Southern Africa', JLurnal of
Agdeu=tural Economics, 41: 303-315.
radin, B. (1989) 'Land tenure, subdivision, and residential change on a Maasai group ranch', Institute fo
mQoDment AntW1gyg Newsletter, 9-13.
Grigg, D.B. (1974) '"he agricultural systems of the world. Cambridge University Press.
Grindle, M.S. (1990) 'Agrarian reform in Mexico: a cautionary tale', in: Prosterman, R.L.,'et al., Agrarian
refor  and gasroots development: Ten Case Studies. Lyenne: Rienner Publishers.
111



Hagen, W.W. (1985a) 'How mighty the Junkers? Peasant rents and seigneurial profits in sixteenth-cetr
Brandenburg', ast and Presnt, 108:80-116.
Hagen, W.W. (1985b) 'The Junkers' faithless servants: peasant insubordination and the breakdown of serfdom
in Brandenburg-Prussia, 1763-1811', in: Evans and Lee.
Hamid, N. (1982) 'Dispossession and differentiation of the peasantry in the Punjab during colonial rule, lourna
of Peasant Studie:52-72.
Hayami, Y., and Kikuchi, M. (1984) Asian village economy at the crossroads: an economic aproach to
institutional change. University of Tokyo Press.
Hayami, Y., Quisumbing, M.A.R., and Adriano, L.S. (1990) 'Toward an alternative land reform Raradigm:
a Philippine perspective: Ateneo de Manila University Press.
Hazell, P., Pomareda, C., and Valdes, A. (1986) Crop insurance for agricultural development: issues and
experience. Baltimore, M.D., John Hopkins University Press.
Heath, J.R. (1992) 'Evaluating the impact of Mexico's land reform on agricultural productivity', Word
Development, 20:695-711.
Henneberry, D.M., and Barrows, R.L. (1990 'Capitalization of exclusive agricultural zoning into farmland
prices', Land Economics, 66:249-258
Heston, A., Kumar, D. (1983) 'The persistence of land fragmentation in peasant agriculture: South Asia,
Exnlorations in Economic History, 20:199-220.
Hilton, R. ed. (1978) 'The transition fom feudalism to capitalism' London: New Left Books.
Hoff, K. (1991) 'Land taxes, output taxes, and sharecropping: was Henry George right? World Bank Economic
Review, 5:93-111.
Holton, RJ. (1985) Ihe transition from feudalism to cpitalism. London: Macmillan.
Horton, D.E. (1976) 'Haciendas and cooperatives: a study of estate organization, land reform and new reform
enterprises in Peru Ph.D. dissertation. Cornell University, Ithaca, N.Y.
Huizer, G. (1972) The reyolutionary potential of peasants in Latin America: Lexington, M.A., Lexington
Books.
Inter-American Development Bank (IDB), (1986) 'Jamaica land tiding project: feasibility report', Washington,
D.C.
Isaacman, A., and Isaacman, B. (1983) Mozanbigue: from colonialism to revolution, 1900-1982. Boulder,
C.O.: Westview Press.
Jarvis, L.S. (1985) Chilean agriculture under military rule: from reform to reaction, 1973-1980. Institute of
International Studies, University of California, Berkeley.
112



Jarvis, L.S. (1989) 'The unraveling of Chile's agrarian reform, 1973-1986', in: Thiesenhusen, W. ed.
Searching for airalan reform In Latin America.
Jodha, N.S. (1975) 'Famine and famine policies: some empirical evidence', Economic and Political Weeldy
1609-1623.
Jodha, N.S. (1986) 'Common property resources and rural poor in dry regions of India', Economic and Political
Weekly, 21:1169-1181.
Jodha, N.S. (1990) 'Rural common property resources: contributions and crisis, EcQnomicand politica Weeldy,
25: A65-A78.
Just, R.E., and Miranowski, J.A. (1989) 'U.S. land prices: trends and determinants', in: Maunder, A. and
Valdes, A. eds., 755-768.
Kay, C. (1983) 'The agrarian reform in Peru: an assessment', in: Ohose, A.K., ed., Affarian reform
contemporary deveggirng countries. London: St. Martin's Press.
Kay, C., and Silva, P. eds. (1992) Development and social change in the Chilean countryside: from the pre-land
reform period to the democratic transition. Amsterdam: CEDLA.
King, R. (1977) Land reform: a world survey. London: G.Bell and Sons.
Klein, H.S., and Engerman, S.L. (1985) 'lhe transition from slave to free labor: notes on a comparative
economic model, in: Fraginals et al.
Koo, A.Y.C. (1968) Land reform and economic development: a case study of Taiwan. New York: Praeger.
Kriger, N.J. (1992) Zimbabwe's guerrilla war: easant voices. Cambridge University Press.
.euger, Schiff, Valdes ?
Kutcher, G.P., and Scandizzo, P.L. (1981) The acultura gecm    of Norteast Bai. Washington, D.C.:
World Bank.
Laffont, JJ., and Matoussi,-M.S. (1988) 'Moral hazard, fnancial constraints and sharecropping in El Oulja',
Californian Institute of Technology and Social Science WP 667.
Larson, B.A., and Bromley, D.W. (1990) 'Property rights, externalities, and resource degradation: locating the
qgedy', Journal of Develogment Economics, 33:235-262.
Lau, LI., and Yotopoulos, P.A. (1971), A test for relative efficiency and application to Indian agriculure,
American Ec mic Revie, 61:94-109.
Lau, LJ., and Yotopoulos, P.A. (1979) eds., 'Resource use in agriculture: applications of the profit function
selected countries, EQod Research Institute Studies, 17:1-115.
Lawry, S.W. (1990), Tenure policy towards common property natura resources in Sub-Saharan Africa', Natual
113



R20=  JouMal, 30:403-422.
Le Roy Ladurie, E. (1974) The peasants of Languedoc. Urbana: University of Illinois Press.
Lemann, D. (1986) 'Sharecropping and the capitalist transition in agriculture: some evidence f*om the
highlands of Ecuador', Journal of Development Studies, 23:333-354.
Leo, C. (1978) 'he failure of the 'progressive farmer' in Kenya's million-acre settlement scheme' Journal of
Modern Afican Studies, 16:619-678.
Leys, C. (1974) Underdevelojment in Kenya: The political economy of neo-colonialism. 1964-71. University
of California Press.
Libecap, G.D. (1986) 'Property rights in economic history: implications for research, &Worations in Economic
Histoary, 23:227-252.
Lin, J.Y. (1990) 'Collectivization and China's agricultural crisis in 1959-1961', lournal of Political Economy,
98:1228-1249.
Lin, J.Y. (1991) 'The household responsibility system reform and the adoption of hybrid rice in China', okumal
f Develoment Economics. 36:353-372.
Lin, J.Y. (1992) 'Rural reforms and agricultural growth in China', Amedican Economic Rviw, 82:34-51.
Lindo-Fuentes, H. (1990) Weak foundations, the economy of El Salvador in the nineteenth century. University
Of California Press.
Lipsey, R.O., Lancaster, K. (1956157) 'The general theory of the second best', Review of Economic Studies,
24:11-32.
Loveman, B. (1976) Strugle In the countyside: politics and rural labor in Chile. 1919-1973. Indiana University
Press.
Ludden, D. (1985) Peasant history in souLh India. Princeton, NJ.: Princeton University Press.
Luetge, F. (1979) Deutsche sozial und wirtschaftsgeschichte:. ein ueberblick, Berlin: Springer.
Mahmood, M. (1990) 'The hanmge in land distribution in the Punjab: empirical application of an exogenous-end-
ogenous model for agrarian sector analysis', Pakistan Develggpment Review, 29:149-289.
Mililc, A., and Schwab, R.M. (1991) 'Optimal investnents to establish property rights in land', Journalf
Utbn Economis, 29: 295-309.
Mason T.D. (1986) 'Land reform and the breakdown of clientelist polidcs in El Salvador', C£MDarative
Political Studi, 18:487-516.
Masters, W.A. (1991) 'Comparative advantage and government policy in Zimbabwean agriculture', Ph.D.
dissertation, Standord University, Palo Alto, C.A.
114



Mc Clintock, M. (1985) 'be Americconnectign. Vol. 1, State terror and opoular resistance in El Salvador.
London: Zed Books.
McClintock (1981) C. M.?? Peasant cooperatives and political change in Peru. Princeton, N.J.: Princeton
University Press.
McClintock, C. (1984) 'Why peasants rebel: the case of Peru's sender luminoso', WAorld Politics, 37:48-84.
McCloskey, D.N. (1975) 'The persistence of English common fields, in: Parker, W., and Jones, E. eds.
European Peasants and their Marke=. Princeton, N.J., Princeton University Press.
McGregor A. (1977) 'Rent extraction and the survival of the agricultural production cooperative', American
Journal of Agricultural Economics, 59:478-488.
McMilan, J, Whalley, J., and Zhu, L. (1989) 'The impact of China's economic reforms on agricultural
productivity growth, Journal of Political Economy, 97:781-807.
Meillassoux, C. (1981) Maidens. meal and money: capitalism and the. iomestic community. Cambridge
University Press.
Melmed-Sanjak, J., and Carter, M.R. (1991) The economic viability and stability of capitalized family farming:
an analysis of agricultural decollectivization in Peru', Journal of Development Studies, 190-210.
Meyer, C.A. (1989) 'Land reform in Latin America: the Dominican case. Praeger.
Migdal, J.S. (1974) 'Peasants. politics, and revolution: pressure toward political and social change in the third
worW. Princeton, N.J., Princeton University Press.
Migot-Adholla, S., Hazell, P., Blarel, B., and Place, F. (1991) 'Indigenous land rights systems in Sub-Saharan
Africa: a constraint on productivity?' World Bank Economic Review, 5:155-175.
Mills, D.E. (1989) Is zoning a negative sum game?', Land Economics, 65:1-12.
Mitchell, J. (1990) 'Perfect equilibrium and intergenerational conflict in a model of cooperative enterprise
growth', Journal of Economic Theory, 51:48-76.
Mitra, P.K. (1983) 'A theory of interlinked rural transactions', Journal of Public Economics, 20:167-191.
Moerner, M. (1973) 'The Spanish American hacienda: a survey of recent research and debate', Hisanic
American Hisry Review, 53:183-216.
Moll, P.G. (1988) 'Transition to freehold in the South African reserves', World Development 16:349-360.
Moore, B. (1966) Social origimns of dictatorship and democracy: lord and peasant in the making of the modern
_gla. Boston: Beacon Press.
Morooka, Y., and Hayami, Y. (1990) 'Contract choice and enforcement in an agrarian community: agricultural
tenancy in upland Java', Journal of Development Studies, 2842.
115



Mosley, P. (1983) The settler economies: studies in the economic history of Kenya and Southern Rhodesia.
1900I1963. Cambridge University Press.
Murrell, P. (1983) 'The economics of sharing: a transactions cost analysis of contractual choice in farming',
Bell Journal of Economics, 14:283-293.
Nabi, I. (1986) 'Contracts, resource use and productivity in sharecropping', Journal of Development Studies,
22:429-441.
Nagarajan, G., Quisumbing, M.A., and Otsuka, K. (1991) 'Land pawning in the Philippines: an exploration
into the consequences of land reform regulations', Developing Economies, 29:125-144.
Newbery, D.M.G., and Stiglitz, J.E. (1979) 'Sharecropping, risk sharing and the importance of imperfect
information', in: Roumasset, J.A. et al., eds., Risk. uncertainty. and agricultural development. Agricultural
Development Council.
Nolan, P. (1988) The political economy of collective farms: an analysis of China's post-Mao rural reforms.
Boulder, C.O.: Westview Press.
Noronha, R. (1985) 'A review of the literature on land tenure systems in Sub-Saharan Africa', World Bank
Discussion Paper, Washington, D.C.
North, D.C., and Thomas, R.P. (1971), 'The rise and fall of the manorial system: a theoretical mode!', Jumrnal
of Economic History, 31:777-803.
Oldenburg, P. (1990) 'Land consolidation as land reform in India, World Development, 18:183-195.
Ortega, E. (1990) 'De la reforma agraria a las empresas asociativas', Revista de la CEPAL, 40:105-122.
Otsuka, K. (1989) 'Determinants and consequences of land reform implementation in the Philippines'. mimeo.
Otsuka, K., Chuma, H., and Hayami Y. (1992) 'Toward a general theory of land and labor contracts in agrarian
economies', Journal of E.mQmic Literature.
Otsuka, K., and Hayami, Y. (1988) 'Theories of share tenancy: a critical survey', Economic Development and
ultural Chang, 37: 31-68.
almer, R. (1977) Land and racial domination in Rhodei, University of California Press.
Pant, C. (1983) 'Tenancy and family resources: a model and some empirical analysis', Journal of Development
EcQnomics, 12:27-39.
:arse, A.C. (1975) The Latfn American peasant London: Cass.
Pfeiffer, K. (1985) Agrarian refbrm under state capitalism in Algeria, Boulder, C.O., Westview Press.
Phimister, I. (1988) An economic and social history of Zimbabwe 1890-1948: capital accumulation and class
stuggI . LongmanL
116



Pickett, L.E. (1988) Qrganizing d m t giicipation, C-eratve organization and servicesf
land sedmnt. A study prepared for the ILO. London: Croom Helm.
Pingali, P., Bigot, Y., and Binswanger H.P. (1987) Agricultural mechanization and the  of farmin
systems in Sub-Saharan Africa. Baltimore, M.D. Johns Hopkins University Press.
tingali, P.L., and Xuan, V.T. (1992) 'Vietnam: decollectivization and rice productivity growth', Economic
Development and Cultural Change. 697-717.
Pitt, M.M., and Rosenzweig, M.R. (1986) 'Agricultural prices, food consumption, and the health and
productivity of Indonesian farmers', in: Singh et al.
Platteau, J.P. (1992) 'Formalization and privatization of land rights in Sub-Saharan Africa: a critique of current
orthodoxies and structural adjustment programmes. London School of Development Economics Paper 34.
Pogodzinski, J.M., and Sass, T.R. (1990) 'The economic theory of zoning: a critical review, Land Economics,
66:294-314.
Prosterman, R.L., and Hanstad, T.M. (1990) 'China: afieldwork-based appraisal of thehousehold responsibility
system, in: Prosterman, R.L., Temple M.N., Monstad, T.M., Agrarian Reform and Grassroots Develogment.
Boulder, C.O., Leyenne Rienner Publishers.
Prosterman, R.L., Temple, M.N., and Hanstad, T.M. (1990) Agrarian reform and grassroots development: Ten
Case Studies. Boulder, C.O., Lyenne Rienner Publishers, Boulder, C.O., Leyenne Reinner Publishers.
Pryor, F.L. (1992) 'Problems of decollectivization with special attention to East Germany', in: L. Somogyi
eds., Problems on Transition to a Market Economy.
Puterman, L. (1989) 'Agricultural producer cooperatives, in: Bardhan, P.K. ed., The economic theoryof
agmdaran bsittins 319-339.
Putterman, L., and DiGiorgio, M. (1985) 'Choice and efficiency in a model of democratic semi-collective
agriculture', Oxford Economic Papers, 37:1-21.
Quibria, M.G., and Rashid, S. (1984) 'The puzzle of sharecropping: a survey of theories', World D)velgpment
12:103-114.
Randall, A., and Castle, E.N. (1985) 'Land resources and land markets. HNRE, Vol. II, 571-620.
Ranger, T. (1985) Peasant consciousness and guerrilla war in Zimbabwe: a comparative study. London: James
Currey.
Rao, C.H.H. (1975) Technological change and distribution of gains in Indian agriculture. Delhi: Macmillan.
Ravallion, M. (1991) 'Reaching the rural poor through public employment: arguments, evidence, and lessons
from South Asia', World Bank Research Obse , 6:153-176.
Ray, R. (1975) 'The Bengal Zamindars: local magnates and the state before the permanent settlement, Indio
117



Economic and Social History Review, 12:263-2292.
Reid, J.D. (1976) 'Sharecropping and agricultural uncertainty', Economic Development and Cultural Change,
24:549-576.
Robertson, A.F. (1982) 'Abusa: the structural history of an economic contract', journal of Development
Stuie, 447-478.
Robison, L.J., Lins, D.A., and VenKatraman, R. (1985) 'Cash rents and land values in U.S. agriculture',
American Journal of Agricultural Economics, 67:795-805,
Roemer, J.E. (1982) A general theory of exploitation and class. Cambridge: Harvard University Press
Rosenzweig, M.R. (1978) 'Rural wages, labor supply and land reform: a theoretical and empirical analysis',
American Economic Review, 67:847-861.
senzweig, M.R. (1988) 'Risk, implicit contracts and the family in rural areas of low-income countries',
-nomic Jggurnal, 98:1148-1170.
Rosenzweig, M.R., and Binswanger, H.P. (1993) 'Wealth, weather risk and the composition and profitability
agricultural investments', EgonQmic Journal.
Rosenzweig, M.R., and Stark, O., (1989) 'Consumption smoothing, migration, and marriage: evidence from
rural India, Journal of Political Economy, 97: 905-926.
Rosenzweig, M.R., and Wolpin, K.I. (1985) 'Specific experience, household structure, and intergenerational
transfers: farm family land and labor arrangements in developing countries', Quarerly Joumnal of Economics,
100-961-987.
Rosenzweig, M.R., and Wolpin, K.I. (1989) 'Credit market constraints, consumption smoothing and the
accumulation of durable production assets in low-income countries: investments in bullocks in India. University
of Minnesota.
Ruthenberg, H. (1980) Farming systems in he tropics. Oxford University Press.
Sadoulet, E. (1992) 'Labor-service tenancy contracts in a Latin American context', A n  Economic Rey'
1031-1042.
Scaritt, J.R. (1991) 'Zimbabwe: revolutionary violence resulting in reform', in: Goldstone, J.A., Gurr, T.R.,
Moshiri, F., eds., Revolutions of the Late Twenthieth Century. Boulder, C.O.: Westview Press.
Schwartz, S.B. (1984) 'Colonial Brazil, c.1580-c.1750: plantations and peripheries', in: Cambridge history of
Latin Amerida Vol I.
:cott, J.C. (1976) The moral economy of the peasant: rebellion and subsistence in Southeast Asia. New Haven,
C.T.: Yale University Press.
en, A.K (1964) 'Size of holdings and productivity', Economic Weeldy A=al Number.
118



Sen, A.K. (1981) 'Market failure and control of labour power: towards an explanation of 'structure' and chang
in Indian agriculture', parts 1 and 2, Cimbridge Journal of Economics, 5:201-228 and 327-350.
Shaban, R.A. (1987) 'Testing between competing models of sharecropping', JouMal of Political Economy,
95:893-920.
Shaban, R.A. (1991) 'Does the land tenancy market equalize holdings?' Working Paper, University o
Pennsylvania.
Shalit, H., Schmitz, A. (1982) 'Farmland accumulation and prices', American Journal of Agmicultural
Economics, 64:710-719.
Shih, H. (1992) Chinese rural society in transition: a case study of the Lake Tai area, 1368-1800, Berkeley:
University of California Press.
Siamwalla, A. et al. (1990) 'The Thai rural credit system: public subsidies, private information, and segmented
markets, World Bank Economic Review. 4:271-295.
Simons, S. (1987) 'Land fragmentation and consolidation: a theoretical model of land configuration with an
empirical analysis of fragmentation in Thailand: Ph.D. thesis, University of Maryland, College Park.
Singh, B., Bal, H.S., and Kumar, N. (1991) 'A spatio-temporal analysis of land-lease markets in Punjab',
Indian Journal of Agricultural Economics, 46:355-360.
Skinner, J. (1991) 'If agricultural land taxation is so efficient, why is it so rarely used?' World Bank Economic
Review, 5:113-133.
Skinner, J. (1991) 'Prospects for agricultural land taxation in developing countries', World Bank Economic
Review, 5:493-512.
Skocpol, T. (1982) 'What makes peasants revolutionary?' CMarative Eoitcs, 14:351-375.
Skoufias, E. (1991) 'Land tenancy and rural factor market imperfections revisited', Journal of Economic
Development, 16:37-55.
Southgate, D. (1990) 'The causes of land degradation along 'spontaneously' expanding agricultural frontiers
in the third world', Land Economic, 66:93-101.
Southgate, E., Sierra, R., and Brown, L. (1991) 'The causes of tropical deforestation in Ecuador: a statistical
analysis', World Development, 19:1145-1151.
Srinivasan, T.N. (1973) 'Farm size and productivity. Implications of choice under uncertainty, Sankhya, Tbe
Indian Journal of Statistics, Series B.
Srinivasan, T.N. (1979) 'Agricultural backwardness under semi-feudalism: comment', Eonomic Joumal,
89:416419.
Stanfield, D. (1990) Rural land titing and registration in Latin America and the Caribbean: implications for
119



rural deyelopment rrograms. Land Tenure Center.
Stiglitz, J.E. (1986), 'The new development economics', Wrld Development, 14:257-265.
Stiglitz, J.E., and Weiss, A. (1981) 'Credit rationing in markets with imperfect xnrmation', American
Economic Review, 71:393 409.
Strasma, I., Alsm, J., Shearer, E., and Waldstein, A. (1987) 'Impact of agricultural land revenue systems on
agricultural land usage'.
Sweezy, P. (1978) 'A critique', in R.Hilton ed., The Transition from Feudalism to Capitalism.
Takekoshi, Y. (1967) The economic aspects of the history of the civilization of Japan. London.
Taslim, M.A. (1988) 'Tenancy and interlocking markets: iss**s and some evidence', World Development,
16:655-6.
Thiesenhusen, W.C. (1987) 'Incomes on some agrarian reform asentamientos in Panama', Ecnomic
P-miDpment and Cultural Change, 809-831.
Thiesenhusen, W.C., and Melmed-Sanjak, J. (1990) 'Brazil's agrarian structure: changes from 1970 through
1980', World Development, 18:393-415.
Trautman, W. (1985) 'Rural development in Algeria: the system of state-directed cooperatives', Ouarterl
Journal of International Agriculture, 24:258-267.
Udry, C. (1990) 'Credit markets in northern Nigeria: credit as insurance in a rural economy', World Bank
Economic Review, 4:251-270.
Vaillancourt, F., and Monty, L. (1985) 'The effect of agricultural zoning on land prices', Quebec, 1975-81,
Land EonQomics, 61:36-42.
Walker, T.S., and Ryan, J.G. (1990) Village and household economies in India's semi-arid tropics, Baltimore,
M.D.: Johns Hopkins University Press.
Weiner, D. (1988) 'Land and agricultural development', in: Stoneman, C. Ed., Zimbabwe's prospects: issuea
of race. class, state, and cagital in southern Afria. London: Macmillan.
Wenfang, Z. and Makeham, J. (1992) 'Recent developments in the market for rural land use in China', Lnd
Economics, 68:139-162.
Wilson, F. (1971) 'Farming 1866-1966', in: Oxford History of South Africa, Vol.1l. Oxford University Press.
'olf, E. (1968) Peasant wars of the twentieth century. New York: Harper and Row.
World Bank (1984) 'Agricultural mechanization: a comparative historical perspective, 673. Washington, D.C.
Wuyts, M. (1985) 'Money, planning, and rural transformation in Mozambique', Journal of Development
120



Sudies, 22:180-207.
Zamosc, L. (1989) 'Peasant struggles and agrarian reform'. Washington.
Zimmerman, F., and Carter, M. x1992) 'A dynamic simulatior. of endogenous structural evolution in an
imperfect-market, agrarian, two asses market economy., mimeo.
121



Policy Research Working Paper Series
Contact
Title                            Author                  Date               for paper
WPS1 151 Is Growth Bad for the Environment?  Charles van Marrewijk    July 1993      J. Verbeek
Pollution, Abatement, and        Federick van der Ploog                     33935
Endogenous Growth               Jos Verbeek
WPS1152 Population, Health, and Nutrition:    Denise Vaillancourt  July 1993         0. Nadora
Annual Operational Review for Fiscal Stacye Brown                           31091
1992                            and Others
WPS1 153 North American Free Trade        Alberto Musalem         July 1993          P. Infante
Agreement: Issues on Trade in    Dimitri Vittas                             37664
Financial Services for Mexico   Ashl DemirgOq-Kunt
WPS1 154 Options for Pension Reform in Tunisia Dimitri Vittas     July 1993          P. Infante
37664
WPS1155 The Regulation and Structure of   Martin F. Grace         July 1993          P. Infante
Nonlife Insurance in the United  Michael M. Barth                           37664
States
WPS1 156 Tropical Timber Trade Policies: What Panayotis N. Varangis    July 1993     D. Gustafson
Impact Will Eco-Labeling Have?  Carlos A. Primo Braga                       33714
Kenji Takeuchi
WPS1 157 Intertemporal and Interspatial   Sultan Ahmad            July 1993          E. O-Reilly-
Comparisons of Income: The Meaning                                           Campbell
of Relative Prices                                                          33707
WPS1158 Population Growth, Externalities,    Nancy Birdpall       July 1993          E. Homsby
and Poverty                     Charles Griffin                             35742
WPS1 159 Stock Market Development and     Asli DemirDg8-Kunt      July 1993          P. Sintim-
Financial Intermediary Growth:  Ross Levine                                  Aboagye
A Research Agenda                                                           38526
WPS1 160 Equity and Bond Flows to Asia and   Punam Chuhan         July 1993          Rose Vo
and Latin America: The Role of Global Stijn Claessens                       31047
and Country Factors              Nlandu Mamingi
WPS1 161 Increasing Women's Participation in  Molly Maguire Teas  July 1993          L. Maningas
the Primary School Teaching Force and                                       80380
Teacher Training in Nepal
WPS1 162 The Slovenian Labor Market in    Milan Vodopivec         July 1993          S. Moussa
Transition: Issues and Lessons   Samo Hribar-Milic                          39019
Learned
WPS1 163 Domestic Distortions and         James E. Anderson       July 1993          D. Gustafson
International Trade             J. Peter Neary                              33714



Policy Research Working Paper Series
Contact
THle                             Author                  Date               for paper
WPS1164 Power, Distortions, Revolt. and   Hans P. Binswanger      July 1993          H. Binswange.
Reform in Agricultural Land Relations Klaus Deininger                       31871
Gershon Feder