wps5;n)'9 POLICY RESEARCH WORKING PAPER 2014 Land Institutions and Secure property rights to land and well-functioning land Land M arkets rental and sales markets are essential for creating investment incentives, Klaus Deininger improving the allocation of Gershon Feder land, and developing financial markets. Yet regulatory restrictions on land rental and sales and regulatory frameworks providing inadequate tenure security are common. This paper looks at the impact of imperfections in other factor markets and the costs and benefits of government intervention to improve the security of property rights and the functioning of land markets and draws conclusions about band policy issues. The World Bank Development Research Group Rural Development November 1998 I POLICY RESEARCH WORKING PAPER 2014 Summary findings In agrarian societies land serves as the main means not review factors affecting participation in land sales and only for generating a livelihood but often also for rental markets, particularly the characteristics of the accumulating wealth and transferring it between agricultural production process, labor supervision cost, generations. How land rights are assigned therefore credit access, the risk characteristics of an individual's determines households' ability to generate subsistence asset portfolio, and the transaction costs associated with and income, their social and economic status (and in market participation. These factors will affect land sales many cases their collective identity), their incentive to and rental markets differently. Removing obstacles to the exert nonobservable effort and make investments, and smooth functioning of tand rentat markets and taking often their ability to access financial markets or to make measures to enhance potential tenants' endowments and arrangements for smoothing consumption and income. bargaining power can significantly increase both the With imperfections in other markets, the institutions welfare of the poor and the overall efficiency of resource governing the allocation of land rights and the allocation. functioning of land markets will have implications for Drawing on their conceptual discussion, the authors overall efficiency as well as equity. draw policy conclusions about the transition from Deininger and Feder examine how property rights in communal to individual and more formal land rights, land evolve from a situation of land abundance. They steps that might be taken to improve the functioning of discuss factors affecting the costs and benefits of land sales and rental markets, and the scope for individual land rights and highlight the implications of redistributive land reform. tenure security for investment incentives. They also This paper - a product of Rural Development, Development Research Group -was prepared as background for the forthcoming Handbook on Agricultural Economics. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Maria C. Fernandez, room MC3-542, telephone 202-473-3766, fax 202-522-1151, Internet address mfernandez2@worldbank.org. Theauthorsmaybecontactedatkdeininger@worldbank.org or gfeder@worldbank.org. November 1998. (44 pages) The Policy Research Working Paper Series dissemb ates the findings of work in progress to encourage the exchange of ideas aboPt development issues. An objective of the series is to get the findings out quickly, even if the presenitations are less than fuilly polished. The papers carry the names of the authors and sbould be cited accordingly. The fiiidings, interpretations, and conclusions expressed in this paper are entirely those of the authtors. They do not necessarily represent thge view of the World Bank, its Executive Directors, or the Lcountries they represent. Produced by the Policy Researcb Dissemination Center Land Institutions and Land Markets Klaus Deininger and Gershon Feder Land Institutions and Land Markets Klaus Deininger and Gershon Feder 1 INTRODUCTION In agrarian societies land is not only the main means for generating a livelihood but often also to accumulate wealth and transfer it between generations. The way in which land rights are assigned therefore determines households' ability to produce their subsistence and generate marketable surplus, their social and economic status (and in many cases their collective identity), their incentive to exert non- observable effort and make investments, and often also their ability to access financial markets or to arrange for smoothing of consumption and income. Given this context, markets in which to exchange rights to land can provide a low-cost means to effect transactions that would bring this factor of production to its most productive use. The institutions governing the functioning of land markets will affect the transaction cost associated with such exchanges, the magnitude and distribution of the benefits generated by them, and the incentives for rational economic agents to undertake efficiency-enhancing transfers and land-improving investments. Furthermore, since land is one of the best collateral assets available, clearer property rights and greater ease of their exchange are likely to affect the emergence and efficiency of financial markets. This implies that land markets have an essential role in the broader process of economic development. In this paper we first examine the way in which property rights in land evolve in an ideal and undistorted environment. We view the emergence of land rights as an endogenous response to increased scarcity of land and the associated incentives for land-related investment and then discuss other factors such as further increases in population density, better access to markets, or the introduction of improved technology to exploit the land, that would lead to increased productivity of agricultural cultivation, as well as endogenous technical change. We note that, historically, there have been few cases where such an uninterrupted evolution has been followed and briefly sketch the conditions required as well as the deviations from this ideal path. With this in mind, we discuss factors affecting the costs and benefits from individualized land rights and examine empirical evidence for their magnitude. In particular, the implications of tenure security for investment incentives are highlighted. Having clarified the concept of property rights in land, we turn to land sales and rental markets. To do so, we consider the main factors affecting participation in those markets, in particular characteristics of the agricultural production process, labor supervision cost, credit access, the risk characteristics of an individuals' asset portfolio, and the transaction costs associated with market participation. These factors will affect land sales and rental markets differently; in particular, even if owner-operated farms are more productive than wage-labor operated ones, the sales market will not necessarily shift land to them. This implies that, in environments where financial markets are imperfect, land market operation needs to be considered within a broader perspective focusing on access to other markets and the availability of alternative assets. We note that, in general, land rental markets would be less affected by these problems because renting out does not preclude the landlord from utilizing land as a collateral to access credit which could then be passed on to the tenant in an interlinked contract. Removing obstacles -often government regulations or imperfections in other markets- that prevent smooth functioning of land rental markets and taking measures that enhance potential tenants' endowments and bargaining power, can considerably increase both the welfare of the poor and overall efficiency of resource allocation. There are also many instances where sales markets are regulated in a manner which hamnpers incentives for socially optimal behavior. In addition to reducing tenure insecurity, governments can in these situations improve the efficiency of resources use by avoiding interventions limiting rental and sales market. Finally, in a number of countries, a highly unequal land ownership distribution implies inefficient and inequitable resource use which the land sales or rental markets are not able to smoothly transform into a more efficient and equitable allocation. Based on these issues we draw policy conclusions concerning the transition from communal to individualized land rights and the award of titles, steps that might be used to improve the functioning of land sales and rental markets, and the scope for redistributive land reform. 2 PROPERTY RIGHT TO LAND 2.1 THE EMERGENCE OF LAND RIGHTS The process of gradual individualization of property rights in land can be conceived as an induced institutional response to higher shadow prices of land to encourage longer term investments in land, as in the pioneering analysis by Boserup. I At the earliest stages of development, even before the establishment of sedentary agriculture, tribes of hunters and gatherers assert control over certain locations wlhere they collect food and engage in hunting. As population density increases, forest fallow systems, and then communal property right systems emerge. Under these arrangements, the general right to cultivation of a It is well understood that this idealized process has rarely been followed in actual history (Boserup herself devotes more than one chapter to the issue of coercion and the description of feudal systems). It is, nonetheless, useful to illustrate the main underlying factors. 2 piece of land is an inseparable and in principle inalienable element of tribal membership. GCltivation rights are assigned to individuals on a temporary basis, normally as long as the cleared plot is cultivated. Once cultivation has ended (due to exhaustion of soil fertility), the plot falls back to the community and the family either selects a new plot (if land is abundant) or has a plot allocated by the chief of the tribe.The fact that land is held by the community rather than the individual facilitates periodic redistribution of at least part of the land among community members based on population growth, serving as a social safety net and preventing the emergence of a class of permanently landless individuals. Tenure security in a general sense is very high, i.e. individual members enjoy secure and inheritable general rights to cultivable land which can be reactivated even after a period of absence. As the relative scarcity of land increases, the pledging or intra-community rental of land emerges. This practice, whereby land that is not used can temporarily be pledged to another family, with the stipulation that it has to be returned upon request, facilitates the productive use of land in case the original owner is unable to undertake cultivation. It is distinctly different from permanent land transactions and is generally not allowed to involve people from outside the community? It also does not uniformly apply to all land - unimproved land lying fallow at any given time continues to be at the free disposal of the community for example for grazing by domestic animals owned by any family with cultivation rights. Variations of such communal tenure systems where parcels are re-allocated from time to time in order to accommodate population growth and grazing land is left for communal use are common in many parts of the developing world such as China, large parts of Africa, and Mexico. What are the factors driving this process of successively increasing precision in the definition of property rights to land? The most frequent explanation is that a virtuous cycle of technical change and investment is set in motion by a combination of increasing population density, technical progress, commercial integration, and reduction of risk. Boserup was the first to point out the fact that, historically, higher population density was the driving force behind an endogenous process of better definition and enforcement of property rights, changing arrangements for the organization of production, and higher levels of investment. The Boserupian framework of changes in the relative scarcity and of land use of labor-saving technology can, for example, explain systematic changes in the strength of womens' land rights (Platteau 1998b). Under land abundance and predominance of shifting cultivation, agriculture tends to be female- 2 Indeed, the distinguishing characteristic of communal tenure systems is not a lack of general tenure security but the fact that property rights are not pernanently linked to a specific plot, implying the existence of restrictions on the transferability of land rights (especially to individuals who are not members of the community). 3 dominated, polygyny is widespread, and women enjoy high status as workers as well as childbearers. Marriage is accompanied by the transfer of bridewealth to the bride's family and, in case of the husband's death, women retain land rights either in their native or in their new village. With greater increased land scarcity and adoption of the plough the importance of women in agricultural production tends to decline and bridewealth, as well as customary safeguards to protect widowed and isolated women disappear. Instead, upon marriage the women receives a pre-mortem inheritance in the form of dowry which -as it remains the property of the wife- establishes a threat point in intra-household bargaining and provides economic security in the case of divorce or death of their husband. The diffusion of exogenous technical change and/or expansion of trade generally have an investment- increasing effect similar to the one caused by increased population density. By increasing the stream of incomes that can be derived from a uniit of land, they augment incentives for better definition of property rights in land. Indeed, establishment of tree crops, and the associated heavy investment in clearing and leveling of land, was generally only undertaken where institutional innovations had enhanced tenure security adequately so that individuals could be sure to reap the benefits from such investments. Similarly, the transport revolution caused by the steamship in the late 19th century led not only to the incorporation of hitherto unexplored countries and states into global trade but also increased the demand for individualized ownership of land. For example the opening of Thailand to international rice trade through the Bowering treaty of 1826 induced a quantum increase in the demand for rice land in the Thailand plains, and brought about the introduction of a formal land registration system (Feeny, 1982), induced a quantum increase in the demand for rice land in the Thailand plains, and brought about the introduction of a formal land registration system. Another important factor furthering the evolution of individual property rights to land is the reduction of risk to income and consumption. The three major avenues for this to come about are (i) the development of markets for output, capital, and insurance; (ii) technical progress that allows diversification, reduction of the covariance of yields, and the probability of crop failure, and (iii) the emergence of access to non- covariate streams of off-farm income. It has long been noted that group ownership of land (or joint communal production) can be viewed as an "insurance policy" to eliminate the threat of permanent assets loss or to reduce vulnerability to idiosyncratic consumption shocks. However, the scope for using communal land ownership to insure against non-idiosyncratic shocks is limited by the weather-induced covariance of agricultural production. Especially when it entails collective production,3 the costs of l Group ownership has often been prevalent where risk is high and where factors such as remoteness, environmental hazard, or presence of external enemies, imply that superior insurance mechanism are not available (Ellickson 1993). 4 collective production are high and individuals to prefer individual ownership once alternative and less costly mechanisms to insure against covariate risks become available (see Key et al. 1998 for the case of Mexican ejidos).4 It is well known that, because monitoring of effort in agricultural production is very difficult and costly, collectives where individuals are not residual claimants to profits are highly inefficient forms of agricultural production (Deininger, 1993). However, contrary to widespread misconceptions, communal tenure systems are generally not based on collective production. Instead production on arable plots is normally undertaken by individuals who are residual claimants to output implying that, on arable plots, incentives for effort supply by individual cultivators are likely to be appropriate. Inefficiencies may persist with regard to either the longer-term sustainability of decisions concerning the use of communal areas such as forests and pasture or the disincentive to invest deriving from the inability to claim ownership rights to specific plots. In an analysis of Mexican farming communities (ejidos), McCarthy et al. (1 997) provide empirical evidence for the existence of collective action problems regarding the use of pasture and forest, but not of individually managed plots. Communal resource ownership is often motivated by the ability to provide benefits in the form of easier provision of public goods, arrangements to enhance equity, or the ability to take advantage of synergies that would be difficult to realize under fully individualized ownership. Examples include risk-reduction through diversification advantages in highly variable environments (Nugent and Sanchez 1994), the utilization of economies of scale to break seasonal labor bottlenecks (Beard 1996; Mearns 1996), or investment in community-level infrastructure (Boserup 1965; Dong 1996).5 As long as effective means of governance and accountable institutions at the local level are available, these systems can -especially in situations when there is need for community-level investment- be very effective. For example, under the medieval open field system, cultivation decisions were made collectively but monitoring-sensitive tasks were carried out on an individual basis. This allowed utilizing economies of scale in fencing, harvesting, shepherding, and risk diversification through strip-cropping without compromising the advantages of individual effort supply (McCloskey 1975 and 1991;Townsend 1993; Blarel et al. 1992). The usufructuary rights given under communal tenure systems do not impose large losses as long as population density is low and land is relatively abundant, pay off for making long-term investments is 4 The potential usefulness of communal land ownership as a device for consumption- smoothing is inversely related to the incidence of locally covariate climatic shocks. It is thus not surprising that, at comparable levels of population density, communal tenure systems have proven to be more durable in environments where such risks are lower. SAn interesting case to illustrate this is made by Ellickson (1993) who compares different settlements (Jamestown, Plymouth, Salt lake City, and the Bennudas) to suggest that, while many frontier settlements started out with group ownership and production to utilize economies of scale in 5 low, and definition of individual property rights in land is costly. However, even though individuals have the right to cultivate specific plots (a measure that avoids the efficiency losses due to collective production), the lack of permanent rights that is implied by the periodic re-distribution of plots may decrease incentives to make long term land-related investments under communal arrangements. A similar effect comes thorough the limitation of land transfers to members of the community and the inability to utilize land as a collateral for credit. Improved access to markets, infrastructure, and financial intermediation, are alternative ways to provide the benefits -in terms of insurance, diversification, and access to funds for investment- associated with communal forms of land ownership. At the same time these exogenous factors increase the costs -in terms of investment disincentives and foregone land transactions with outsiders- associated with traditional land ownership systems. This implies that, with economic development, the relative attractiveness of communal systems will decrease and, at some point, it would be economically rational for a community to allocate permanent and fully tradable ownership rights to individuals (seeThompson and Wilson 1993 for Mexico), completing the transition from a communal to an individualized tenure system. However, instead of following an smooth endogenous evolution along the lines outlined above, the transition to individual property rights has in the large majority of historical cases been affected by exogenous interventions. As population growth increases the relative scarcity of land, one observes a general increase in boundary conflicts and social tension.6 In the absence of strong and representative community-level institutions, this often leads to appropriation of property rights to the communal resources by powerful individuals, by and abuses of power and land grabbing through local chiefs and headmen.7 These phenomena are often seen as a major course of environmental degradation and increased social tension and inequality that leaves out the poor and vulnerable. History demonstrates that regions with potential for agricultural or non-agricultural exports were generally characterized by the appropriation of large tracts of lands through imperialist, colonial, or other overlords who either replaced local chiefs and elders, or tried to co-opt them to enforce their rule. These changes undermined traditional tenure systems, the associated structures of accountability, and thus the institutional underpinning of the organic evolution of such systems (Downs and Reyna, 1993; Feder and defense and other activities, the length of maintaining group ownership can be related to the riskiness of the environment, the frequency of social interaction and the hierarchy structure of decision-making. 6 Zimmerman and Carter (1996) show that incorporating agent heterogeneity, risk, and subsistence constraints can facilitate a more differentiated assessment of the welfare- and productivity-impact of a given institutional innovation (e.g. the adoption of marketable land rights) on different groups of producers. 7 For example, despite extremely low levels of population density in Zambia, almost 50% of small producers feel that their security of tenure is insufficient and are willing to pay (a mean amount of US $ 40) for getting secure ownership rights (Deininger et al. 1998). Low-cost means of 6 Noronha 1987).8 Furthermore, once thiey discovered that access to labor rather than land was the most limiting factor, overloads generally introduced distortions in other markets to reduce the reservation utility of independent farmers and to assure a supply of labor for export production in mines or for the newly established estates. In addition to reducing the reservation utility that cultivators could obtain from independent cultivation, such restrictions have contributed to widespread rural poverty and retarded development of competitive markets in rural areas, often laying the basis for continued rural-urban dualism. In more recent times, Governments have, through implicit and explicit taxation, drained the rural sector of resources that could have fueled a process of increased markets integration and technology development while at the same time higher rates of population growth vastly increased the need for new technology and better infrastructure (Schiff and Valdes 1995). The associated lack of markets and technological opportunities has, in a number of cases, contributed to a situation akin to the "involution" that had earlier been diagnosed for Asian systems (Geertz 1960), with far-reaching implications for the structure of resource ownership rights. For example in Rwanda, with very high population density (787 persons per km2), traditional systems of land allocation have become defunct and fail to provide even the most basic services they were designed for (Andre and Platteau .1996). As traditional limitations on land sales have been discarded, speculative land purchases by individuals with access to non-covariate off- farm income lead to a rapid dis-equalization of landholdings.9 While costly land disputes consume productive energy, environmental degradation continues unabated and the return to an idealized notion of "communal property rights" is unlikely to be a feasible option. To judge, however, what alternative arrangements would be feasible, it is necessary to consider in more detail the costs and benefits associated with different tenurial arrangements. Drawing together the evidence on costs and benefits associated with more secure and fuller property rights arrangements, three conclusions emerge. First, where population density is sufficiently high, increased tenure security -not necessarily equivalent to formal title- has an important impact on increased investment. Second, there is some evidence that a higher degree of transfer rights provides additional incentives for investments and for more efficient use of family labor. Finally, the ability to use land as collateral to increase access to medium and long term formal credit markets is of importance if increasing tenure security and reducing encroachment from outside through better accountability and issuance of community-titles could possibly increase welfare and tenure security. ' This was independent of whether the intervention was associated with the elimination of traditional tenure systems in favor of individualized rights to the selected group, as in many parts of Central and Latin America, or the use of local chiefs and dignitaries as intermediaries for the central power as in African countries. ' It is of interest to note that about 65% of sales are classified as distress sales -the incidence of which is not restricted to the lowest landholding group- and an additional 17% of lands are sold to cover litigation expenses, often arising from land disputes. 7 foreclosure is feasible. Studies that compared the financial costs and economic benefits of titling program suggest that high rate of return are possible but that, unless measures to reduce the transaction costs associated with administering credit to smallholders are undertaken, the benefits associated with titles may not accrue equally to all types of farmers. 2.2 BENEFITS AND COSTS OF INDIVIDUALIZED PROPERTY RIGHTS The main benefits from well-defined and secure individual property rights relate to (i) greater incentives for (and lower costs of) long-term resource conservation and the associated increased demand for investment; (ii) improved transferability (temporary or permanent) of land to cultivators who have the resources to make better use of it and (iii) the ability to use land as collateral in formal credit markets, a benefit that is more significant where formal title exists and land transactions are actually feasible. These benefits need to be weighed against two main types of costs: the administrative and logistical expense associated with definition of boundaries, enforcement of rights, and resolution of disputes among claimants, and the increased risk of losing a safety net provided by communal control of land. 2.2.1 Benefits from individual land rights Improved tenure security brought about by individualized land rights will be associated with static and dynamic benefits. Even without having full long-term security of tenure, individual cultivation rights that entitle an individual to residual claimancy of profits generated on a plot mark the difference between collective and private forms of cultivation. The transition from collective to private cultivation has historically been associated with large increases in productivity, as for example in the case of China (McMillan et al. 1989, Lin 1992, Lin et al. 1994) However, equally important benefits from better defined long-term property rights would come about in an inter-temporal setting where higher security of tenure would increase the incentives for long-term investments, the incidence of productivity-enhancing transfers, and the supply of credit to make such investments. These aspects are elaborated upon below. Tenure security Conceptually, insecurity associated with the lack of well-defined property rights can be understood as a random probability of loss of future income due to conflicting challenges. Eliminating such a threat through informal institutions (customary tenure) or formal institutions (land titles) will clearly increase the subjective payoff from productivity-enhancing long-term investments and thus the owner's 8 willingness to undertake them. While the theoretical expectation is straightforward and easily formalized (see for example Feder et al. 1988; Besley 1995), the critical question, and much empirical debate, has focused on the magnitudes of such effects in a different settings. The analysis of different types of land rights in Africa is complicated by the need to take in account for the potential endogeneity of investment (Besley, 1995). The reason is that there may be certain types of investments -from marking of boundaries to planting of trees and hedges, and building of houses or sheds- that may be undertaken with the primary purpose of establishing implicit property rights to land rather than to increase productivity (Brasselle et al. 1997). Depending on how such actions affect the probability of land loss and whether or not there are community rules to provide (partial) compensation for such investments when a plot reverts to the community, it is easy to construct scenarios where communal tenure systems may increase rather than decrease the amount of land-related investment undertaken (Sjaastad and Bromley 1997).1O The key result from a number of studies that have investigated the investment-enhancing effect of tenure security is that, under formal as well as informal regimes, tenure security -as measured by the extent of rights possessed by the owner- significantly affects farmers' investment decisions. Especially where investments are labor-intensive but involve few cash outlays, the unambiguous conclusion is that higher levels of tenure security -even if they are not associated with high levels of transferability and defined only at an informal level- do indeed provide an important incentive for increased investment.1 I Evidence from one of three study areas in Ghana indicates that greater tenure security at the plot level significantly increases the probability that individuals will plant trees, and undertake a wide range of other investments such as draining, irrigating, mulching, etc. (Besley 1995). The fact that field-specific rights but not mean household rights can be shown to be significant, suggests that plot-level tenure security, rather than credit supply effects accruing to the household as a whole, are likely to be at the root of this relationship between tenure security and investment. Results from China confirm the importance of tenure security for investment. Comparing plots planted to the same crop within the same household but under different tenure regimes, it is found that farmers tend to apply more manure and labor, and obtain significantly higher yields, on plots that are privately owned and therefore more secure (Rozelle et al. 1996). This is the case even though the possible impact of "' Using comparative statics from a simple model it can be shown that communal as compared to individual tenure is more desirable from the individual's point of view as the discount rate increases; the productivity increase generated by investment is smaller compared to rent; the initial probability of eviction is low; and the probability of recovering investment even after eviction is high. A combination of these factors may cause individuals under indigenous tenure to commit resources to land improvement beyond what would be the case under individual resource ownership. 9 greater tenure security on crop choice (shifting to orchards instead of growing maize) is not accounted for. Similarly, Yao (1996) finds that higher levels of tenure security in Chinese villages have a strong and very significant investment-enhancing impact (application of green manure).12 Analysis of the impact of higher tenure security through land titling in the Brazilian Amazon yields similar results (Alston et al. 1995 and 1996) and there is considerable amount of more anecdotal evidence on a positive association between availability of title and farm output or investment (see Binswanger et al. 1995 for references). On the other hand in Niger, a more land-abundant setting, different degrees of tenure security between plots with full private ownership and plots held under usufruct, do not give rise to large statistically significant differences in application of manure, a medium term yield improving investment (Gavian and Fafchamps 1996). In this context, farmers apply significantly lower amounts of manure on rented as compared to owned plots, but there is no significant difference between parcels held under full private ownership and those held under "traditional" usufruct. The conclusion is that apparently tenure security on the latter is high enough for farmers to expect to be able to reap the benefits from their (medium term) investment. At a more general level, it indicates that, in order to determine whether specific property right arrangements are conducive to higher levels of investment, more detailed study is necessary and generalizations are unlikely to be helpful. What is instead required is a more differentiated judgment that takes account of the time horizon of the investment, the opportunity cost of the resources used, and the size and distribution over time of the expected payoff associated with the investment. Transferability Land markets tend to be highly localized. As a consequence, the ability to transfer land between users may be of limited importance in early stages of development when there is little heterogeneity of skills across the population and non-agricultural opportunities are limited. However, the importance and value of being able to transfer use or ownership rights to land increase with economic development, specialization, and better development of other markets. In this case, the transfer of land from those who have lower productivity to those who are able to make more productive use of the land improves the overall resource allocation. The demand for such exchanges increases further as the rural economy becomes more integrated geographically, facilitating transactions between individuals who are not members of the same community. Such situations generally involve larger problems of asymmetric This does not necessarily imply that actions to increase tenure security are warranted or even needed (Platteau, 1998). 12 At first glance this would seem to be at variance with the finding by Feder et al. (1992) who find that, for a similar sample from four Chinese provinces, neither short-term nor long-term tenure security -captured by farmers' perception about the possibility that their land may be reallocated before the expiration of the current 15-year contract.One can reconcile the two findings by noting thatFeder et al.'s study considers non-attached investment (machinery, livestock, and construction) which should not be aftected by individual plots' tenure security, that access to 10 information and greater benefits from more formal systems of land ownership recording. If the ability to liquidate investments (through land transfers) increases the incentive to undertake such investments, higher levels of transfer rights, and the greater ability to affect transfers which is entailed in formal land rights systems, will not only improve resource allocation but will also be associated with higher levels of investment and labor use by individual cultivators. The only data that allow testing of this hypothesis come from China where one observes variability in systems of transfer rights in different communities (Carter and Yao, 1998). Results suggest that higher levels of transfer rights increase investment (application of green manure). In addition, evidence from China indicates that higher levels of transfer rights also induce a better allocation of the households labor endowments in response, for example, to outside employment opportunities. Households with higher levels of transfer rights apply less labor on their farm and devote more time to more remunerative off- farm activities (Yao 1998), thereby contributing to equalization of factor ratios within a village and increasing overall efficiency. More indirect support for an important efficiency-increasing (but not investment enhancing) impact of higher transfer rights is provided by Rozelle et al. (1997) who find that an increase in off-farm opportunities narrowed the difference between labor spent on (transferable) private and (non-transferable) communal plots.13 Credit access In addition to inducing investment, secure land ownership is likely to increase the supply of credit from the formal credit system to undertake such investment. The reason is that, because of its immobility and virtual indestructibility, land with secure, clearly defined, and easily transferable ownership rights is an ideal collateral. The provision of a collateral -facilitated by possession of formal land title- is generally a necessary condition for participation in formal credit markets for medium and long-term credit. In fact, there is evidence of titles facilitating access to informal (but impersonal) credit markets as well (Siamwalla et al., 1993). Existence of well documented and transferable property rights and of institutional arrangements to facilitate the low-cost transfer of land are likely to make an important contribution to the development of financial markets. However, while use of titled land as collateral can, under the condition that foreclosure is socially feasible, reduce a bank's default risk and thereby enhance credit supply, it will have little impact on the working capital emerged as being an important determinant, and that the subjective measure of tenure security used by them is more noisy than the index of land tenure security used by Yao. '" Evidence is not uniform: for Ghana, the hypothesis that sales and rental rights do not have a significant impact on investnent decisions can not be rejected (Besley 1995). This suggests that the prospect of being able to transfer land more easily through sales and rental markets in the future is, in this environment, not an important consideration in individuals' decision to effect land-related investment. 11 transaction costs associated with administering credit to small producers in rural areas. In environments where these costs are high, the improved credit-worthiness brought about by possession of land title may therefore not be enough to facilitate access to formal credit by small farmers. Unless complementary measures to reduce transaction costs and ensure access to credit by this group are undertaken alongside with individualized property rights through titling, the benefits from titling programs may accrue only to medium and large landowners. The importance of the credit supply effect associated with provision of land title is supported by evidence from Feder et al.'s (1988) study in Thailand, where farmers' opinions and econometric evidence point towards improved credit supply as the main benefit from titling. Land ownership titles induced higher investment in farming capital (attached investments and other capital),14 titled land had signiificantly higher market values and higher productivity per unit. In three of the four provinces covered, households' credit supply had been significantly enhanced by the availability of title. By contrast, and in line with the above, title was found to have little impact on either investment or farm income where formal credit markets were not available (Atwood 1990, Carter and Wiebe, 1994; Migot Adholla et al. 1994).15 Additional evidence from a study based on panel date from Paraguay indicate that titling had a positive income or productivity enhancing effect through credit market benefits for at least some groups of farmers. Due to a strong impact of formal title on both credit supply and investment demand, the benefits from title are relatively large (about 10% of farm income), and significantly higher than the cost of titling,. However, the impact of awarding titles was strongly size-differentiated. Estimates indicate that producers below 20 hectares remained rationed out of the credit market and therefore did not benefit at all from the credit-supply effect of title (Carter and Olinto, 1996). This differential impact suggests that, in environments where other markets (such as credit markets) entail distortions which put smaller and poorer farmers at a disadvantage, individual property rights on equity, and -in the medium to longer term- on the direction and nature of land transfers between different size classes of producers, could imply greater inequity. Whether, in the presence of heterogeneity in endowments, small producers will benefit from such policies depends critically on the ability to reduce, together with titling, transaction costs and policy induced distortions that limit access to credit markets. 4 Problems of endogeneity and self-selection are circumvented by drawing samples from squatter villages in areas nominally under public ownership (where titles could not be awarded) and private areas where all residents already had obtained titles. " Pender and Kerr (1996) show that for India land ownership has little impact on credit supply, a fact that is attributed to severe non-price rationing. Nonetheless, land values for titled land are on average about 15% higher than for untitled land, suggesting that possession of formal title reduces the probability of land loss for potential buyers. 12 2.2.2 Costs The most obvious cost associated with formal definition of property rights in land is the expenditure needed to physically demarcate and delineate plots, to establish and maintain accurate records of land ownership, and to enforce these rights and resolve whatever disputes might arise.16 These costs are borne by individual land owners in situations, for example frontier settlement, where public enforcement of property rights is absent and individuals make defensive investments such as guards, fences, and other demarcation devices to demonstrate the legitimacy of their claims to property and to defend such rights against possible intruders (Muller, 1997). It has been shown that the privately "optimal" amount of spending by individuals on means of protection will be inefficient from a social point of view (Feder and Feeny; de Meza and Gould 1992). Furthermore, the defensive activities undertaken often have little social value and may generate negative externalities, an issue that has been emphasized with respect to the Brazilian Amazon where the need to demonstrate "productive" land use to establish ownership claims has been linked to increased deforestation (Binswanger 1988, Southgate et al. 1991). Even where they are not associated with externalities, defensive activities that are often undertaken in speculative attempts to secure "ownership" of large tracts of land, can lead to complete dissipation of the rents to be had (Allen 1991).17 Given the undesirable impacts of private rights enforcement, public provision -in the form of land records, police, and a judiciary- would therefore be preferable in all situations except ones characterized by very low levels of population density (Malik and Schwab, 1992). This is indeed observed throughout human history. The specific form in which land records are established will still depend on the relative costs and benefits from such an activity - something that depends partially on the technology and infrastructure available for record keeping. 1 8 At initial stages of human development assignments of property rights appear to have been handled orally by the community (with community functionaries holding public sessions at the gate, for example). However, the benefits from keeping written records seem to have been so great that across a large number of cultures, officially validated land records are among the first documents to appear once a written language had been developed (Ellickson 1993).19 In addition to establishing unambiguous ownership rights, written records do allow to verify the ownership status of land at low cost, thus reducing the scope for asymmetric information about ownership and quality of land and making land sales and 16 Note that the number of disputes is itself endogenous, depending on the type of property rights system chosen. '' Spontaneous collective action to limit the dissipation of resource rents associated with individualized defense of property rights has been observed in a number of cases where group sizes were small (Umbeck, 1977). ' Ellickson (1993) notes that historically the establishment of formal land rights is closely related to the emergence and widespread use of written language; in many cultures records of land transactions were among the first texts that have been officially recorded. i See, for example the Indian Arthsastra from the 4th Century BC as well as references in the Bible relating to the period 600 BC. 13 rentals cheaper to implement. This reduction of transaction costs would increase the liquidity of the land market and could bring the number of efficiency-enhancing transactions closer to the optimum, i.e. helping to transfer more land from less productive to more productive individuals. A second type of social cost associated with fully individualized property rights relates to the fact that, at low levels of development, communal land ownership may perform an important insurance function that would be eliminated by establishing fully individualized property rights in land.20 Furthermore, it has long been known that in cases where other markets are highly incomplete, land sales markets may not automatically transfer land to more productive users. In such situations, individualization of land rights could be doubly disadvantageous (see Platteau, 1998 for references). On the one hand it could pave the way for the emergence of sales markets that deprive traditional communities of their source of livelihood (often without adequate compensation), thus generating social unrest and violence and eliminating an important form of insurance. On the other hand, where land rights are introduced in such an environment, productivity will not necessarily increase as availability of land rights could induce concentration of landholdings by a privileged minority of wealthy individuals who -for example by having access to non- covariate sources of income- are in a position to accumulate land for speculative purposes without making productive use of this asset.21 Historical evidence does indeed suggest that, in situations where other markets are not well developed or where policy-induced distortions affect the functioning of land markets, increased transferability of land may deprive the poor of an important social safety net. The importance of the insurance aspect is confirmed by the fact that, even where societies have made the transition to individualized land rights, they have often maintained land-related social safety nets to provide insurance for the poor. One example for a mechanism to do so is to allow continuing uses of communal pastures and forest areas of low productive value as well as an universal right to collect leftovers after the harvest or graze animals on harvested fields. Another one is the provision for periodic redistribution of at least part of the land available to the community - for example the Bible mandates land redistribution every 50 years.22 Such redistribution of cultivation rights is likely to decrease productive efficiency by attenuating individuals' 2' Jodha (1990) provides evidence on the importance of access to the commons as a safety net for the poor. Based on panel data from China, Burgess (1997) finds that the equitable allocation of land use rights under communal tenure has an effect similar to a lump sum transfer that provide insurance against low nutritional outcomes in a way that is more incentive-compatible than arex post redistribution. The fact that land ownership has a more significant impact on improving nutrition than on income can be explained by the fact that, with imperfect rural grain markets, considerable cash outlays would be required to achieve a similar effect through market purchases of grain. The presence of equity benefits from periodic redistribution of land rights in China would be consistent with peasants' strong support for the system of periodic redistribution (Kung, 1995). 21 Note that this is historically well-founded, as the many examples in Binswanger et al. (1995) demonstrate. 22 If incentive structures and enforcement mechanisms to ensure that such provisions are actually implemented at the local level are non-existent, the provision for regular redistribution can actually give way to arbitrary behavior and rent extraction by local leaders. For a theoretical and empirical discussion of these issues see Tumer et al. (1998). 14 incentive to make plot-specific investments. The fact that societies have been willing to incur these efficiency losses suggests that the subjective valuation of the benefits in terms of avoiding widespread landlessness, social destitution, and discontent, has been high. This implies that where land is an important asset for poor and marginal groups, both social and efficiency aspects associated with land rights need to be accounted for in assessing the potential benefits from individualizing land tenure arrangements. 3 LAND MARKETS: FUNCTIONING AND EFFICIENCY IMPLICATIONS If there are differences in individuals' skills and endowments of different factors of production, markets should help in optimizing factor proportions employed and thus increase overall efficiency of resource allocation. This section aims to outline the main determinants that would affect participation in the land sales or the land rental market, and based on this to elaborate on links and differences between these two markets, in terms of their impact on equity and efficiency of resource allocation. The productivity advantage of small farmers who rely predominantly on family labor rather than on less motivated hired workers who have to be supervised would imply that, in the absence of imperfections in other markets, a functioning land market should facilitate efficiency- and equity-enhancing transfers from large to small producers - or from ones with lower management skills to better operators. However, land sales transactions could be efficiency decreasing if, due to policy induced credit market distortions, large owners' advantage in accessing credit would offset the productivity advantage of owner operators or if, due to the inability to insure, significant land holdings are not part of poor people's optimal asset portfolioZ3 Thus, before actions to activate the land market are undertaken, careful empirical investigation of the functioning of financial markets and insurance mechanism, and possibly steps to improve their functioning, might be in order. Even if imperfections in markets for credit and insurance reduce the scope for the land sales market to bring about improved land allocation through land transfers from large to small producers, such allocation should- in a frictionless world- be facilitated through the land rental market. One possibility would be an interlinked contract whereby the landlord uses the credit access provided by land ownership to provide the tenant with working capital as part of the rental contract. High transaction costs -part of them related to government regulation- reduce the extent of land rental transactions in a number of countries. Examining the 23 Indeed, there is descriptive evidence indicating that in environments with imperfect credit market access, e.g. in Africa, land sales markets result in an efficiency-reducing transfer of land from small to large producers (Collier 1989). 15 implication of regulations in more detail would be of importance as removing unjustified interventions is likely to go a long way towards improving resource allocation in agricultural systems characterized by very unequal land distribution. Most rental markets in developing countries involve some form of share tenancy. While this arrangement does not lead to full efficiency, it is a second best solution given risk and imperfect capital markets. The sections below elaborate these points and review relevant evidence. 3.1 KEY DETERMINANTS OF LAND MARKET PARTICIPATION The shadow price of land for different types of heterogeneous agents is determined by the agricultural production function, the households' inherent managerial ability, and by a number of imperfections in labor, credit, and land markets that are common in rural areas. If credit and land rental markets were perfect, the supervision costs associated with the use of hired labor would make smaller farms more productive, and would lead households to lease in or lease out the amount of land required to maintain a uniform ratio of family labor endowment to operated area, irrespectively of the land ownership distribution (Feder 1985). However, imperfections in other markets may change this, with implications for the functioning of land rental and sales markets. If, in the presence of credit market imperfections, 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. While the magnitude (and direction) of this effect would depend on the elasticity of output with respect to effective labor and of labor effort with respect to supervision, it can overwhelm the productivity advantage of family farmers and give rise to a positive relationship between owned farm size and productivity. In addition to this, capital and insurance market imperfections may also affect the production activities of poor producers - possibly leading them to pursue less risky but also less productive activities. Below we review the factors which affect the productivity of farmers, and thus determine their demand for land. Economies of scale: The presence or absence of economies of scale would systematically affect the shadow price of land for different farm-size classes. Possible economies of scale could arise from the presence of indivisible factors of production or cost elements leadingto an initial range of farm size where the average cost of production declines with farm size. In cases where other markets function reasonably well, optimal farm sizes tend not to exceed the scale at which family labor is fully occupied (utilizing seasonal hired labor for specific tasks).24 Agricultural activities where significant economies of scale in the production process 24 A large number of empirical studies (e.g. Olson-Lanjouw 1995 for India, Feder et al. 1992 and Burgess 1997 for China, Olinto 1995 for Paraguay) are indeed unable to reject the hypothesis of constant retums to scale in agricultural production. 16 exist are few.25 Some economies of scale are associated with the processing and marketing of many agricultural products, but this does not have important implications for the unit cost of farming operations as long as competitive markets for outputs and inputs exist. Alternatively, access to such markets is sometimes arranged through cooperatives. Only for a few "plantation crops" such as sugarcane, bananas, or tea could the need for immediate large-scale processing or marketing transmit economies of scale from the processing stage to production. To reap the economies of scale associated with the former, production of these crops is generally organized on a scale that corresponds to the optimum scale of the processing factory.26 Labor supervision cost: Constant returns to scale would imply that the size of agricultural operations has little impact on productivity. However, the need to supervise hired labor would confer a productivity advantage on owner-operated farm units. The fundamental reason for this is the presence of agency costs (Jensen and Meckling, 1976), which result from the need to manage wage labor and enforce effort in large- scale operations. The lack of incentives for wage workers to exert effort, and the consequent need to supervise labor or to offer incentive contracts, has received considerable attention in industrial organization literature (Jensen and Meckling 1976), and is recognized to have profound implications for the organization of production and for the optimal size of the firm (Calvo and Wellisz, 1978; Eswaran and Kotwal 1985a and 1985b). The cost of supervision is particularly large in agricutural production due to spatial dispersion of the production process and the need to constantly adjust to micro-variations of the natural environment. Family members are residual claimants to profits and thus have higher incentives to provide effort than hired labor. 27 They share in farm risk, and can be employed without incurring hiring or search costs. These attributes underlie the general superiority of family farming over large-scale wage operations, manifested empirically in an inverse relationship between farm size and productivity. A large number of studies based on aggregate, or cross-sectional, and panel data have confirmed the existence of the inverse farm-size productivity relationship for all but the smallest farm size classes (Berry and Cline 1973, Carter 1984, Benjamin 1995, Newell et al 1997, Kutcher and Scandizzo 1981, Olinto 1995, Burgess 1997, Udry 1997).28 Thus, unless there are other countervailing forces, one would expect land markets to transfer land from large to small producers. We turn now to discuss these countervailing effects. 2 Exceptions are limited to cases of highly specialized machinery, specialized livestock production, or plantation crops where economies of scale are transmitted from the marketing to the production stage. 26 However, the supervision advantages of pwner-operators have in many cases motivated large processors to contract production out to smallholders under outgrower or contract farming schemes, often providing credit in kind as well as technical assistance (Glover 1990). 27 Empirical evidence confirms that family labor is more productive than hired labor, and that the intensity of supervision by family members affects the performance of hired labor (Frisvold 1994). 28 Bhalla and Roy (1989) and Benjamin (1992) have shown that from cross-section analyses (e.g. Berry and Cline 1973, Carter 1984, Newell et al 1997, Kutcher and Scandizzo 1979) tend to overestimate the productivity advantage of smaller farms if soil quality is not specifically accounted for. 17 Credit market access: A key reason for land markets' transfer from large to small producers to be rarely observed is that it is very difficult for small farmers to access markets for credit and insurance.29 This implies that, on the one hand, the value of accessing credit markets is capitalized in land prices, making it very difficult to acquire land in the sales market with the expectation of paying off the debt from agricultural profits alone without recourse to equity. On the other hand, credit market imperfections that increase the shadow price of credit for small producers would reduce small farmers' competitiveness in the land sales market and at the same time outweigh the supervision cost advantage they enjoy. Asymmetric information and moral hazard lead generally to quantity rationing in credit markets (Stiglitz and Weiss, 1981). Formal credit markets can overcome the problem of asymmetric information by utilizing a collateral requirement. However, the costs of and political impediments to foreclosure on smallholders land are often quite significant. This is part of the generally high transaction costs associated with providing credit to small producers. In informal credit markets, close familiarity and social control is used to select promising clients or projects. This is quite costly as the scope for effective supervision is limited. Furthermore, informal lenders have only limited scope to diversify covariate risks, and they typically do not provide much long term credit. Interest rates on informal loans are thus high. Thus, both limited availability of credit and high cost of borrowing of credit and high cost of borrowing would prevent those whio do not have accumulated savings from acquiring land. 30 Credit market imperfections can thus offset small farmers' supervision cost advantage. For the case of Sudan, for example, yields for virtually all crops are lower for poor (small) farmers and higher for rich (large) farmers, thus turning the farm-size productivity relationship upside down. Furthermore, the lanid rental market leads to land transfers from poor and labor-abundant small holders, to rich and relatively labor scarce households (Kevane 1996). The reason is that capital market imperfections combined with reasonably functioning land and labor markets and a technology that is not supervision intensive make it more attractive for small credit-constrained households to rent out land and work for a wage than to engage in owner-cultivation without capital inputs. By contrast, in panel data from Burkina Faso an inverse farm size-productivity relationship was observed even though a positive presence of correlation between yields and cash inflows from non-agricultural employment suggests the presence of capital market imperfections (Udry 1996). The conclusion is that imperfections in land, labor, credit, and 29 Due to the covariance of production risks, crop insurance is very difficult and forward markets to insure against price risk are often unavailable to small producers due to high transaction costs. The difficulty of land acquisition through borrowing by would-be smallholders, in spite of their productivity advantage, has been highlighted by Binswanger and Elgin (1998) and Carter and Mesbah (1995). Furthermore, they point out that by exhausting access to credit for land acquisition, the ability to borrow for working capital is eliminated. 18 insurance markets have to be analyzed together. Efforts at land redistribution that do not simultaneously address credit market imperfections may be costly and ineffective. Portfolio composition: Small producers' inability to access formal markets for credit and insurance often forces them to adopt costly insurance substitutes, one of which is the adjustment of crop and asset portfolios to a low return -low risk combination.31 In order to ensure satisfaction of a minimum subsistence requirement during periods of distress, credit-constrained producers could hold a portfolio of less risky but also less productive assets than unconstrained producers. In particular, small holders may demonstrate a lower demand for land than that which would seem to be justified by their potential productive advantage. Zimmerman and Carter (1996) use parameters from Burkina Faso to show that, starting from an egalitarian distribution of land, production risk together with covariance of land prices, leads to successive concentration of land via sales from more productive small producers to relatively less productive large farmers. This illustrates that improving the functioning of land sales markets will not necessarily lead to better resource allocation if other markets' distortions are not tackled.32 Transaction costs: A further factor that might prevent land markets from achieving a first-best allocation is the transaction cost associated with land sales and rentals. It has often been observed (see, for example, Balcazar 1993, Carter and Zegarra, 1995) that, especially in countries with a dualistic distribution of land ownership, land sales markets are highly segmented in the sense that, despite a considerable frequency of land transactions within farm size groups, land sales across farm-size class-boundaries are virtually absent. One explanation is that transaction cost entailed in subdividing farms so as to enable to a multitude of small holders is high. Similarly, the fact that certain costs (e.g. formal registration) associated with land transactions are independent of the size of the purchase creates indivisibilities that would either discourage small land transactions or drive them into informality where such costs are not incurred. While the discussion of costs associated with land rentals in the literature is less extensive, government regulations appear to have reduced the amount of land leasing below what would take place otherwise. Even in countries that avoided the imposition of explicit restrictions on tenancy, which (as discussed below) were associated with significant efficiency losses, the threat of expropriative land reform in many countries implied that renting out land to more productive smaller producers exposed the landlord to a considerable risk of losing ownership rights in the course of land reform. To prevent this from happening, many landlords Examples are provided by Rosenzweig andWolpin (1995), Dercon (1996), Dercon and Krishnan (1996), and Rosenzweig and Binswanger (1993). 32 The fact that study of land markets can not be divorced from the functioning of other markets has beeremphasize by Basu (1986) in a model of "interim" land transactions which explicitly serve as a credit substitute. In this context, the supply of land for sale would increases with the probability of being able to buy back the land, the attractiveness of other (financial) assets as compared to land, and the need for liquidity. Sengupta (1997) draws out the implications of limited liability on contract choice within a more general set of contractual options. 19 appear to have evicted tenants altogether, resorting instead to mechanization, cattle ranching, or cultivation using a hired labor force (de Janvry and Sadoulet, 1993). The implications for land rentals, although they have not been rigorously quantified in any of the cases, appear to have been considerable. 3.2 LAND SALES MARKETS The discussion of the previous section implies that credit market imperfections tend to cause the equilibrium price of land to be above the capitalized value of the income stream that can be generated by an individual who has to rely on full mortgage-based land acquisition. In addition, policy distortions will tend to increase the wedge between the price of land and the capitalized value of the income from agricultural production. Use of land as an inflation hedge, as well as credit subsidies and tax advantages that allow the use agricultural activities as a tax shelter are examples (Gunjal et al. 1996; Brandao and Rezende 1992; Just and Miranowski 1989; Feldstein 1980). To the degree that such distortions confer disproportionate benefits to larger landholders (as in the case of tax advantages which are generally of no relevance to the poor), this would further bias the operation of the land sales market against redistributing land to landless or marginal landowning households who could have a productivity advantage as family farmers. Analysis of land market transactions and offer and asking prices in Paraguay indicates the presence of a large gap between willingness to sell and willingness to pay,33 significant differences in such prices across farm sizes, and very distinct regional patterns of land market performance depending whether or not other markets exist and how well they function (Carter and Zegarra 1996). Both reported and actual sales prices are downward sloping over the 1-40 hectare range (indicating superior productivity of small farmers) in traditionally settled zones in the country's interior, but not at the frontier where the labor cost advantage of family farms was overshadowed by capital market imperfections. Simulations using regression parameters suggest that the productivity advantage of small farmers would manifest itself in the land purchase market only if land market reform was combined with improved access to capital markets. A similar conclusion is implied by the observation that the degree to which financial markets were accessible to small producers was (together with the initial distribution of assets and the characteristics of the production system) one of the key factors that determined the response of land accumulation patterns to agro-export booms in Guatemala and Chile (Carter and Barham 1996). Willingness to sell was significantly higher than was willing to pay to purchase land but the gap decreases with farm size (from 50% for the smallest farms to 20% for medium sized units). This could be in indication for labor market imperfections, i.e. the value given to land as a source for self-employment, in addition to small farmers' unwillingness to be bought out. 20 Exposure to undiversifiable residual risk causes farmers to resort to liquidation of their assets during periods of severe crisis, a phenomenon commonly referred to as distress sales. This is implies that the covariance of weather risks for the farming population causes land prices to be low (due to insignificant effective demand and high supply) during bad crop years, with the consequence that individuals who had to sell off land during crises may not be able to repurchase land during subsequent periods of recovery (Bidinger et al. 1994). Distress sales have not only played a major role historically in shaping more concentrated land ownership pattems, but are also linked to the elimination of traditional mechanisms for coping with risk in the literature (Kranton and Swamy 1997; Brockett 1984)34 The link between unmitigated production risk and distress sales is highlighted by Cain (1981) who examines the implications of different insurance mechanisms on distress sales and the land ownership distribution between 1960 to 1980 for predominantly agricultural villages in India and Bangladesh. These villages faced very high production risks but were characterized by distinct differences in mechanisms of risk-insurance: In Maharashtra, 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. Thus, 60 percent of land sales in Bangladesh were undertaken to obtain food and medicine, undoubtedly due at least in part to the lack of other insurance mechanisms. About 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 to finance consumption expenditures accounted only for 14 percent of sales and were incurred mainly by the rich to meet social obligations. On the other hand, 64 percent of land sales were undertaken in order to generate capital for productive investment (digging of wells, purchase of pump sets, and children's education), and the land sales market actually contributed to a slight equalization of the land-ownership distribution. This suggests that in this case the poor were not only able to avoid distress sales, but were able, through access to cash-generating employment, to acquire some land as rich households liquidated agricultural assets to be able to pursue non-agricultural investment. Survey data on land transactions from India indicate that purchases of land are almost all undertaken by individuals with access to sources of income which are not correlated with agricultural production and that borrowing to finance agricultural land acquisition is virtually non-existent (Sarap 1995). 34 Distress sales have been important in China (Shih 1992) in early Japan (Takekoshi 1967) in the Indian Punjab (Hamid 1983), and in Latin America following the abolition of communal tenure (Brockett 1984). 21 3.3 LAND RENTAL MARKETS As the discussion above illustrated, land sales markets will not necessarily lead to an optimal allocation of land in the presence of credit and insurance market imperfections. However, improved resource allocation can be achieved through land tenancy contracts even when other markets are incomplete. Analysis shows clearly that land rental markets serve an important function in equalizing returns to non-tradable factors of production such as family labor and bullocks (Skoufias 1991).35 Given the huge diversity of tenancy arrangements, we need first to explain the wide range of tenancy contracts that is empirically observed in developing countries. This gives rise to the second issue, namely to investigate in more detail the implications of these contracts for the efficiency of resource allocation. Assume a constant returns to scale production function Q O OF(e,h) where Q is output, e is effort, h is number of tenants, and 0 is a stochastic element. Then the range of contracts can be summarized as follows. The landlord's income is y = h[(1-a)Q - f3], and the representative tenant's income is Y=aQ + P. The fixed rent contract is given by {(a-1,f3<0}, the pure wage contract is represented by {7=0,P>0}; and the share contract is given by {O