Designing land and property rights reform for poverty alleviation and food security^[2] - M.R. Carter^[3]

Department of Agricultural and Applied Economics, University of Wisconsin-Madison, Madison, United States of America Land and property rights policy reforms are receiving new and renewed attention as instruments to redress rural poverty by enhancing the access to land of poor households. This article evaluates what we know - and what we do not know - about these policy instruments. It argues that land is disproportionately valuable to poor households and that enhancing the land access of poor rural households thus represents good social and economic policy. Moreover, assigning land rights to women enhances the land policy effectiveness, and land policy should thus proactively pursue this. However, there are a number of constraints to the effectiveness of land policy, in particular wealth-biased financial markets that dampen the impact of land access on poverty and hamper the competitiveness of small-scale producers. Finally, the article argues that now is the time for carefully monitored piloting of new modalities of land redistribution and land reform organization that hold out the promise of resolving these constraints and making land access work as an important element in the battle against rural poverty.

INTRODUCTION

Land and property rights reforms are receiving new and renewed attention as instruments to redress rural poverty by enhancing the land access and tenure security of poor households. The World Bank's draft policy research report on land, Land policy for growth and poverty reduction (World Bank, 2003) is perhaps the sharpest signal of the prominence of these reforms in contemporary policy debates and agenda. Several conceptual and empirical advances have motivated the search for what Stiglitz (1998) calls a "post-Washington Consensus" about economic development policies that explicitly redress poverty and inequality. As part of this search, land access and distribution have re-entered the debate. However, the land policy tools now under consideration are rather different from those utilized in an earlier land reform era. In contrast to direct state-mandated land redistribution, the World Bank's policy research report on land and contemporary land access policy is built around a primary and secondary set of policy instruments:

1. Liberal land market fundamentals: The primary set of policy instruments is built around the idea that property rights reform that assigns legally secure, usually marketable land rights to individuals should enhance efficiency and the land access (via rentals or sales) and incomes of the poor.

2. Direct redistribution of landownership: A secondary set of instruments is built around the idea that constructive engagement with the land market (through market-assisted land reform or other mechanisms) is possible and may be necessary to overcome deep-seated inequality and improve land access.

The goal of this paper is to evaluate what we know - and what we do not know - about land and property rights reform policies. Specifically, the paper poses the following questions:

• When and why will enhanced land access alleviate poverty, i.e. is promoting land access a good poverty alleviation strategy?

• When and why will liberal land market fundamentalism work?

• When and how can the land purchase market be engaged constructively to achieve redistributive reform?

After discussing each of these questions, the paper asks what can be done to design land and property rights reform policy that will realize poverty alleviation and food security goals more effectively.

WILL ENHANCED LAND ACCESS ALLEVIATE POVERTY?

How important is access to land for the economic well-being of poor rural households? The history of land reform in Latin America and elsewhere would seem to suggest that it is very important. However, if we strip away considerations of identity and political power that have often been mixed into land reform debates, we are left with an important question without an obvious answer.

This section considers four avenues by which land access may contribute to livelihoods of poor households and to the alleviation of poverty in the short and longer terms. These are:

• Excess income gains that accrue when improved land access enhances family income above and beyond the pure rental value of land.

• Food security effects that occur when enhanced land access cheapens the effective or shadow price of food for households.

• Safety net and investment effects that occur if better land access permits households to weather economic shocks more effectively, and thereby pursue and defend more effective wealth-building strategies by sending children to school and keeping them there.

• Dynamic income distribution effects that occur if a more egalitarian land distribution in the present lays the foundation for a broadly-based income growth in the future.

After reviewing each of these effects, this section closes by considering whether or not the assignment of rights specifically to women further enhances the livelihood value of land to poor households.

Income effects of land access

It is not obvious that improved land access will have any extraordinarily high or excess income effects for poor households. If it does not, then, as Lopez and Valdes (2000) have argued, poverty alleviation policy might better emphasize the transfer of other forms of wealth or opportunity to poor rural households. The notion that land has no special value for poor households can be made easily by reference to a standard economic model of the agricultural household. Assuming that: (i) it takes only land and labour to produce agricultural goods; (ii) there are no economies or diseconomies of scale in agricultural production; (iii) households can buy or sell any amount of labour they want at the market wage rate; and (iv) hired labour is as productive as family labour; then the household can adjust fully to restricted land access by using the labour market to sell and obtain the full benefit from any extra labour it might otherwise be able to employ on farm. In this case, enhancing a household's access to land by giving it ownership rights to a piece of land would simply increase its income by the rental value of land. The same effect could be achieved by giving the household a savings bond or any other asset that has the same return as land - there would be nothing special about enhancing the household's land access.

Correspondingly, increasing the household's access to land by permitting it to rent-in additional units of land would have no effect whatsoever on its income. Given optimizing behaviour by the household both before and after the increase in land access, the increment in income produced with the land would be displaced exactly by the rental price of land (and the reduction in off-farm earnings). If access to land is to have special livelihood value for households, then it must somehow have a value in excess of its rental price. So what is missing from this model, and is it important?

Whether or not access to land will generate excess value for poor households depends on the nature of the local labour market. If we modify assumption (iv) above and assume that thin labour markets constrain the amount of labour that households can sell off-farm, then enhanced land access (either owned or rented) will give an extra boost to household income and livelihood beyond the rental value of land. In this case, for households that have constrained access to land, an increase in land access boosts their income by the rental value of land plus by increasing returns to family labour. This second effect, which we might call a "shadow wage effect", results from the household's ability to self-employ its labour and escape dependence on an imperfect labour market.

The very same logic that underlies this shadow wage or livelihood effect of land access also creates the inverse farm size - farm productivity relationship that has been much studied in the empirical literature. While the voluminous literature on the inverse relationship tends to confirm that output per unit area declines with farm size, direct explorations of the income effects of land are rather scarce. Another recent approach to estimating the income value of land for poor households begins by observing that land access will be especially valuable for households that are: (i) constrained in the labour market; and (ii) constrained in the land market. We would not expect land access to be especially valuable for households for which these conditions do not both hold. Simply examining data on a cross-sectional sample of households may be misleading if it mixes doubly constrained and unconstrained households. This is especially likely to be true where households that are fortunate enough to be unconstrained in the labour market also happen to cultivate less land.

In summary, the evidence on the extra income effects of land access is less complete than we might hope given the importance of the issue. In principle, it would be informative to take a random sample of rural households and give improved land access to some and not to others. Observation of differences between these groups over time should permit identification of the average livelihood value of land. Efforts to monitor and evaluate new land access programmes do offer the opportunity to approach this question in this quasi-experimental fashion.

Food security effect of land access

Burgess (1999) puts forward the hypothesis that, beyond its income effects, improved land access cheapens the relative price of food for families, and, controlling for income, results in better nutritional outcomes. When coupled with the wellknown linkages between nutrition and educational outcomes, and the notion that a well-educated population is a basis for endogenous economic growth, the potential importance of this hypothesis is quite large.

Conceptually, Burgess rests his argument on the idea that, for many smallholders, marginal increments of food production cannot be sold economically because of transactions costs in output markets. Thus, marginal increments in farm income will be consumed in-kind, shifting the balance of household expenditure towards better nutrition.

To buttress this claim, Burgess compares data from India with that from China, showing that for similar absolute income levels, rural households are better nourished in China than in India. This finding is at least consistent with the food security hypothesis, given that China's relatively egalitarian land distribution means that poorer deciles in rural areas enjoy much greater land access than do similarly poor households in India. Burgess goes on to investigate the food security hypothesis more directly using household data from China. Consistent with the cross-national data, his econometric analysis shows that better land access has an expenditure switching effect as households with more land have higher food expenditures even after controlling for land and non-land based sources of income.

Safety net and investment effects of land access

In addition to its static income or livelihood effect in the presence of imperfect labour markets, land access may further affect household well-being by stabilizing income and thereby enhancing intertemporal accumulation, especially in the form of children's education. The latter is especially important because even the occasional withdrawal of children from school (perhaps as the family copes with an income shortfall) damages effective human capital accumulation. However, direct evidence on the impact of land access on schooling attainment is scarce. An important exception is a recent study of El Salvador by Conning, Olinto and Trigueros (2001). This study provides evidence that land access plays an important role in stabilizing the incomes of rural households in the face of economic shocks and downturns. In their study, land access seemed to have little effect on boosting incomes in an average year. However, their study also covered a year in which rural El Salvador suffered a weather-related economic shock. Here they found that households with better land access were better able to maintain returns for their labour. In addition, land access also promoted better school enrolments. This evidence from El Salvador indicates that the full livelihood benefits of land access run deeper, influencing the stability of incomes and the intergenerational accumulation of human capital.

Dynamic income distribution effects

The prior three subsections have considered micro- or household-level mechanisms that may link improved land access to poverty alleviation in both the short and the long term. Latin America has had long experience with what Cox, Niño de Zepeda and Rojas (1990) and others have labelled "exclusionary growth" - a process that displaces peasants and tenants, prematurely mechanizes the agrarian economy, and reproduces or deepens rural inequality. This raises the issue as to whether the income distribution and poverty alleviation consequences are themselves shaped by the economy's underlying pattern of land access.

Data on rural income distribution are scarce, making it difficult to test directly for exclusionary growth. However, from a theoretical perspective there are reasons for thinking that the loose empirical association between exclusionary growth and land inequality is not accidental. As a whole, these theoretical considerations suggest that the process of agrarian growth and transformation in market economies can be an unsteady one, and one in which initial levels of inequality are reproduced and deepened by a growth process. Such possibilities introduce the question of the degree to which the trajectory of agrarian growth itself becomes a function of the initial asset distribution in the agrarian economy. Can the relatively egalitarian growth paths of East Asian economies be understood partially as the result of their relatively equal initial distribution of land, as opposed to the continuing reproduction of inequality that has characterized agrarian growth throughout much of Latin America?

To investigate these questions, Carter (2001) uses a Gini decomposition framework to analyse econometrically four linkages or pathways between initial agrarian land access inequality and subsequent income inequality. Using panel data to estimate a mixed effects model of income inequality, Carter finds significant evidence that initial landownership inequality has persistent and long-lasting effects on income inequality. The evidence is ambiguous about whether or not these effects reflect a continuing set of effects or simply a holdover effect of the initial inequality. However, it does suggest that there may be broader social gains from policies that enhance the equality of land access. While precise identification of these effects and their quantitative significance will have to depend on future research efforts, the available theoretical evidence offers insights on the policy implications of these effects. From a theoretical perspective, both the exclusionary agrarian growth effect and the unequal human capital accumulation effect are rooted in missing financial markets. Put differently, agrarian structure can condition the income distribution consequences of growth because missing financial markets create a linkage between the assets a household already has and the new investments it can undertake.

Women's land access and rights and the livelihood value of land

The analysis to this point has been cast in terms of the effect of land access on the incomes and livelihood of disembodied "households." However, does it matter who within the household enjoys improved access to land? Among others, Deere and León (2001) argue that it matters whether women or men enjoy enhanced land access. They distinguish two principal avenues by which enhanced land rights for women can be expected to have an impact on the rural household economy and on gender roles within that economy. The first avenue, which Deere and León call "women's well-being and the family", focuses on property rights as a form of economic access to the key markets that constitute the rural economy - the markets for agricultural goods and services, for capital, for labour and for land itself. The second avenue, which Deere and León call "the empowerment of women", emphasizes the role that land rights can play in strengthening indirectly women's ability to participate effectively in important economic decision-making processes at the household, community and broader levels of society.

From this perspective, increasing women's claims to land, whether as joint or individual owners, can be expected to have positive income and welfare effects both for women and for their children. The literature on intrahousehold expenditure patterns suggests strongly that mothers dedicate greater proportions of their incomes to household public goods, including food, child health and educational expenses. Empowerment of women should in turn increase their influence over household expenditure patterns. In addition to the short- and medium-term economic gains generated by greater access to product, capital and land markets, women with stronger property rights in land are also less likely to become economically vulnerable in their old age, or in the event of the death of or divorce from their spouse.

While additional work in this area is needed, the strong suggestion of recent studies is that it is impossible to discuss the livelihood impacts of land independently of how rights over that land are distributed within the household.

HOW CAN THE LAND ACCESS OF POOR HOUSEHOLDS BE IMPROVED?

The evidence discussed above suggests that efforts to improve the land access of poor rural households warrant inclusion in policies designed to alleviate rural poverty. There are at least three distinct perspectives on how to achieve this goal. The first, termed structuralist orthodoxy, underwrote the major agrarian reforms of the twentieth century, stretching from the Cardenas redistribution in Mexico in the 1930s through to the Sandinista agrarian reform in Nicaragua in the 1980s. The structuralist perspective sees agricultural productivity and income distribution as influenced inexorably by the distribution of landownership. Land access - and the overall performance of the agrarian economy - is inseparable from the agrarian ownership structure from this perspective, and state-mandated land redistribution, which operates directly on the ownership distribution, is the only way to improve the land access of poor households.

Contrasting with this structuralist orthodoxy is what might be termed liberal land market fundamentalism. This liberal perspective relies on market-mediated redistribution of land and land access to achieve the productivity and income distribution desiderata of land reform. Interestingly, the strongest drives to use markets to separate agrarian performance from structure have occurred in countries, notably Mexico, Nicaragua and Peru, that initially carried out some of the most far-reaching redistributive reforms and extensively promoted collective forms of rural organization.

Sounding a cautionary note to this liberal perspective about the ability of markets to separate performance from structure is what might be best termed neo-structuralist scepticism. In contrast to structuralist orthodoxy, the neo-structural perspective argues that although liberalized land markets could in principle contribute to improved land access for the rural poor and raise agricultural productivity, such a virtuous outcome is very unlikely unless the intrinsic flaws in other rural factor markets, especially financial ones, are addressed.

More specifically, the neo-structuralist perspective is rooted in the following propositions:

• The liberal claim that land-poor households enjoy a fundamental productivity and competitiveness advantage over larger-scale producers is correct.

• The liberal claim that "agricultural modernization" and policy reforms that complete and secure individual private property rights to land and "ensure the right price" will suffice to activate the competitiveness advantage of smallholders is flawed and rests on a naïve view about rural financial markets.

• Agricultural performance (and subsequent evolution of the ownership structure itself) will remain tightly linked to ownership structure even in the wake of policy reform that ensures that land market fundamentals are "right".

Without repair of complementary rural markets, this neo-structural perspective indicates that land markets will tend to be biased in their operation towards wealthier agents and larger farms. In turn, this will obviate efforts to create a level playing field, especially in what are already highly inegalitarian agrarian economies. More direct interventions, aimed at either reshaping how financial markets work or shifting the ownership distribution of land directly, are thus required from the neostructural view. Distinguishing between this sceptical, neo-structural perspective and the land market fundamentalist perspective is the task of the next section.

WILL LAND MARKET FUNDAMENTALISM IMPROVE THE LAND ACCESS OF POOR HOUSEHOLDS?

While often discussed in terms of its impact on investment incentives, the secure and marketable property rights that compromise the core of the land market fundamentalist perspective can also activate land rental markets. First, they may affect the rental market directly by increasing the transferability of land. Second, they may indirectly make owners more willing to rentout land when their individual ownership rights are made more secure.

From a conceptual perspective, the ability of activated land rental markets to enhance the land access of the poor and break the link between agrarian ownership and performance depends critically on the functioning of rural financial markets. Unequal, wealth-biased access to capital can distort the functioning of land markets in two ways. First, where poorer households cannot access the liquidity needed to cultivate higher value crops (and manage the risk associated with them), then unequal access to capital will impinge directly on the operation of the land market by depressing the relative competitiveness of poor households. Second, even if it does not blunt small farm competitiveness completely, wealth-biased access to capital (in conjunction with transactions costs) may break the land rental market into small-farm and large-farm segments, with little exchange between the two.^[4] In this case, the land market would fail to enhance the land access of the poor in a broad way and would fail to break the linkage between ownership structure and performance. Given the centrality of financial markets, we first turn to examine their operation before considering the direct evidence on the operation of land markets.

Access to capital

From a conceptual perspective, the provision of secure and marketable land rights could enhance the capital access of land-poor households by transforming whatever land they might have into a collateralizable asset. Generally, some argue that economic liberalization will open the way to new and innovative forms of financial intermediation that will better serve the needs of low wealth producers. Contrasting with this argument is the claim of other analysts that even liberalized rural financial markets will remain wealth-biased, with low wealth producers more likely to be subjected to forms of non-price rationing.

Which of these two views is correct is ultimately an empirical question. Unfortunately, this empirical question is not easy to answer. The difficulty stems from the fact that access to capital and having a loan are not the same thing. More specifically, a household has poor or rationed access to capital if it has excess demand for credit at the going rate of interest. A household that does not desire a loan because the interest rate is too expensive will not have any excess demand for credit and will have adequate access to capital despite the fact that it may borrow nothing. The empirical challenge is to identify properly those who are rationed in their access to capital and those who are not.

One approach to this problem is the econometrics of unobserved regime switching. Using this method, Carter and Olinto (2003) estimate the patterns of capital access using panel data from Paraguay for the early to mid-1990s period. They estimate that: (i) low wealth producers are quite likely to have rationed access to capital; and (ii) legal tenure security has little influence on capital access for low wealth producers, though it does have a significant impact for larger (mid-sized) producers. The key question is whether the more recent round of economic reforms has broken this pattern of wealth-biased capital access.

Several recent studies have tried to answer this question based on a data collection strategy that tries to determine directly whether or not a household has excess demand for capital. In these studies, respondents are asked a sequence of questions about whether they received loans, whether they applied for loans, whether or not they could have obtained a loan had they applied, and why they did not apply for loans. Results suggest that households with low wealth levels are estimated to have a very low chance of gaining access to the capital they demand, and in a very real sense must work their way out of a deep economic hole. Households that have sufficient own liquidity or large enough masses of collateralizable wealth are able to have their credit needs met (the former because they do not demand loans, the latter because the financial system is wealth-biased and works better for them).

The findings reflect the severe wealth bias that is consistent with the sceptical neo-structuralist view of the ability of land markets to enhance the land access of poor households. While it could be argued that wealth-biased credit rationing will be eliminated once more time passes and new private entrants shift the terms of credit access, there have yet to be clear indications that this is occurring.

Land rental markets in practice

There is a growing literature that examines the operation of land rental markets in practice, asking if well-functioning rental markets indeed enhance the land access of poor households. More details on this literature can be found in de Janvry et al. (2001) and in World Bank (2003). More simply, this section attempts to summarize some of the key lessons that are emerging from this literature.

Rental markets for agricultural land are quite active in both developed and developing countries, with as much as two-thirds of agricultural land operated by renters in some European countries. Rental markets are also active in many parts of Asia. For example, the lifting of legal restrictions to land rental in China has been shown to improve the land access of less well-off households and to enhance overall economic efficiency. Similarly, several African economies (e.g. Ethiopia and Uganda) that have relaxed rental restrictions have seen land markets enhance the land access of land-scarce households. A similar pattern has also been found in Pakistan where land rental markets indeed tend to shift land to lower wealth households. However, in anticipation of findings on land rental in Latin America, land rental markets in the more commercialized areas of Pakistan tend not to improve the land access of the poor. This is apparently because the lack of access to adequate working capital blocks the access of poorer households to land.

A small but growing number of studies have tried to explore these same issues in Latin America, especially in the wake of the economic liberalization policies of the 1990s. Honduras, Mexico and Nicaragua all underwent substantial institutional and policy reforms in the early 1990s that affected both land and credit markets. The best evidence to date gives a rather uncertain picture of the effects of this liberalization. In Mexico, there is no evidence that the reforms enhanced the land access of poorer households on average. However, the opposite is true for poorer households that had adequate access to capital. For these households, the reforms did create a better functioning land market where they can rent-in land from those who have more. However, because of the post-reform pattern of credit access, mid-sized farmers (those with more than 13 ha) seem to benefit most in terms of increased land access from the reforms. In the case of Mexico, the message seems to be that land markets can be made to function better for the less well-off, but that policy needs to be attentive to the design and proper sequencing of reforms so that wealth-biased credit access does not undercut the prospects for improved land access.

Broadly similar patterns are discernible in Honduras and Nicaragua. Economic reforms in these countries show little effect on the overall relationship between area operated and area owned. In both cases, the large majority of rentals take place between family members and individuals who describe themselves as long-term acquaintances and friends. In Nicaragua, there is also some evidence of land market segmentation as most land rentals appear to take place between smallholders (e.g. a household owning 3 ha rents another hectare from a household that owns only a few hectares). Rentals from large to small farmers are relatively rare, at least in Nicaragua. This observation raises additional issues about whether land rental markets are still constrained by legal uncertainty and transactions costs that inhibit their fluid operation, something that has been observed in the Dominican Republic.

Land rental markets in Honduras appear to operate in a different fashion. Landless and near landless households are more likely to rent from larger landowners (those who own more than 35 ha) than they are from other ownership strata. However, the amounts actually rented-in by these households are quite small, consistent with the notion that capital access problems block rental activity. Further analysis of both the Honduran and Nicaraguan experiences is needed to determine more precisely the interaction between credit constraints and land rental activity. However, the evidence to date matches the Mexican research by suggesting that weak access to capital has a major depressing effect on the ability of poor households to use even liberalized land markets to improve their access to land.

CAN THE LAND PURCHASE MARKET BE ENGAGED FOR REDISTRIBUTIVE REFORM?

Contrary to the liberal, land market fundamentalist view, land rental markets may be unable to improve the land access of the poor and de-link agricultural performance from the landownership structure, especially in economies with highly unequal landownership. These observations shift the focus back onto forces that shape the distribution of landownership, including the land sales market. This section asks whether or not that market can be engaged constructively through market-assisted land reform programmes to enhance the land access of poor households. As with questions concerning the livelihood value of land and land rental markets as a vehicle of land access, careful empirical analysis is required to understand the functioning of land sales markets in any particular environment.

Land purchase markets in practice

The theoretical literature is ambiguous about the impact of asset markets on the distribution of land. However, reality neither waits for nor requires theoretical clarity. Carter and Zegarra (2000) discuss in some detail the empirical evidence on land purchase markets. We here summarize some of that literature in an effort to gain inference about the prospects for engaging the land market constructively as a vehicle to enhance the land access and ownership of less well-off households.

Changes in the patterns of land distribution by class strata over time give an indication of the ability of the poor to gain access to land through land markets. In highland Ecuador, the expropriation and breaking up of large haciendas during reform, and the freeing up of the land market, increased access to land for the traditionally marginalized indigenous population. In 1954, this population owned 4 percent of the land, by 1961 they owned about 61 percent of it, with no indication of any reconcentration after the reform. In Venezuela, the process of land regularization and public land transfers by the government discriminated against poor agents in favour of medium-sized holders, although this seems not to have resulted in a more skewed distribution of land.

A number of more recent studies have examined the changing patterns of land access in areas of Latin America that have experienced agro-export booms. In Chile, the land reform that began in 1966 ultimately created a class of some 48 000 smallholders (called parceleros), many of whom gained access to some of the country's most productive land. The period since then has provided an excellent opportunity to study the operation of the land market in the context of rapid agroexport growth for a sizeable smallholder population. In the fertile Central Valley, data indicate that nearly half of the parceleros sold their farms between the late 1970s and 1991, leading to a modest reconcentration of land in the hands of medium- and large-scale farmers. Land transactions were most prevalent in the northern Central Valley (where export 54 production has been most profitable), but also spread to the southern portion of the Central Valley in the late 1980s.

Mobility analysis of the Chilean data, which recorded the movement of households between different farm-size categories between 1977 and 1990, confirms the pattern of relative smallholders squeezed out of land access by the export boom. Large farms maintained their position in the period, with some large landowners accumulating substantial parcels of additional land. Forty-seven percent of parcelero households in the main fruitproducing region had become landless by 1990. The mobility analysis reveals no significant upward mobility by any parcelero household but does show substantial upward mobility by a new class of well-financed, often non-agricultural professionals and business people who have purchased land from the parceleros.

This restructuring has led some to characterize Chile's agro-export growth boom as exclusionary, pushing out the uncompetitive smallholder sector (Cox, Niño de Zepeda and Rojas, 1990). This interpretation is bolstered by the fact that real wages in Chile were depressed during the most intense period of sales by parceleros, and that only 20 - 30 percent of those who sold their farms did so because of lack of interest in farming, or because of old age^[5]. Similar patterns of land concentration triggered by export booms were observed in the 1970s and 1980s in several Central American countries.

In sharp contrast to Chile, an export boom in winter vegetable products in Guatemala induced a transfer of land from larger to smaller farms. Land accumulation by the small-scale producer of non-traditional exports was significantly higher in the boom period than pre-boom period for the same farm unit. Farms that began with relatively large holdings (3 ha) did not increase their land significantly in the post-boom era, while those households that had less than 1 ha prior to the boom, and who began producing "boom" crops, expanded their landholdings significantly. Recent changes in the boom sector may be affecting these outcomes.

In the 1970s and 1980s, an agro-export boom in Paraguay was associated with sharply rising real land prices along its eastern frontier, and increased land access by the largest farm-size class, while land access by other size classes of farms was dampened. This pattern of unequal land accumulation is reflected in national agricultural census figures. These show an increase in landholdings by large farms in the period 1980 - 1990. In contrast to this national pattern, in the two more traditional areas of agriculture, not much affected by the boom and characterized by minifundia and peasant agriculture, near landless farmers have continued to accumulate landholdings even as land has become scarcer in real terms.

Overall, the evidence from Latin America suggests that the land rental and land sales markets work in fundamentally different ways. Whereas the land rental market appears to function as a mechanism of land access for labour-abundant, capitalconstrained households, the latter does not. One would in fact expect that capital and insurance constraints would display their influence most heavily in the sale market. Not expected, but of potentially great policy import, is the finding that only when agents are capital unconstrained does their relative technical efficiency appear able to express itself as effective demand for more land. Hence, while part of the interest in land titling and other tenure-oriented policies is in activating a market that it is hoped will shift land to the more efficient, there seems to be little indication that land markets work that way, at least in Latin America. Evidence from other world regions is much rarer, but research on African land purchase markets yields a similar picture regarding the dominating effect of capital constraints.

Market-assisted or negotiated land reform

While the preceding discussion should make us sanguine about the prospects for constructive engagement with the land market, Deininger (2001) describes a set of negotiated or market-assisted land reform initiatives that are predicated on the hope that land markets can be made to work for the rural poor. Brazil, Colombia and South Africa are among countries providing capital grants to landless or land-poor individuals as part of an effort to let these individuals use the market as the medium for asset redistribution.

The case for relying on market-mediated as opposed to state-mandated land redistribution rests on several propositions:

• Small-scale, family-labour farms are fundamentally competitive in the sense that their economic ability to pay for land out of net production revenues is no less than that of large farmers.

• The initial capital grants suffice to overcome any capital access problems that might otherwise dampen smallfarm competitiveness.

• Unlike conventional redistributive programmes that grant beneficiaries a standard allocation of land and perhaps capital, market-assisted programmes permit individuals to choose the farm size flexibly (and the land - capital mix) that matches their other resource constraints and needs.

• Market-assisted programmes rely on a greater degree of beneficiary selfselection, raising the prospect that benefits will be channelled to those most interested and able to use them.

• Administratively, market-assisted programmes are cheaper than conventional redistribution programmes.

• Because they operate on a willing-buyer, willing-seller basis, market-assisted programmes are politically feasible in contemporary environments where expropriation is not.

Market-assisted land reform programmes built on these principles have proved to be controversial, evoking strong criticism at a number of fora, including the conferences organized by the World Bank to discuss its policy research report on land (above). It is useful to disentangle the various criticisms that have been directed at market-assisted reform efforts.

One line of criticism emanates from a social justice perspective. This argues that making land reform beneficiaries buy their land is an unseemly legitimization of past injustices (e.g. why should the victims of apartheid have to buy back lands taken from their ancestors). This argument would seem to apply with particular force when beneficiaries need to match capital grants with some of their own savings in order to purchase land. However, unless reform land is to be seized without compensation, it seems that this criticism could be accommodated through an appropriate grant structure.

A second line of criticism revolves around the perceived competitiveness problems of new beneficiaries. Here the concern is that smallholders will not be able to afford to buy land at the prices that largescale, commercially-driven producers will demand, or that they will not be able to generate the revenue stream needed to pay off any land debt or otherwise make the land purchase economically worthwhile.

This competitiveness critique appears to have two components. One, which Carter and Mesbah (1993) articulate, is that reform beneficiaries will not be able to leverage the working capital needed to produce high-valued products on their land. This argument says that market-assisted land reform needs to be matched with institutional innovation in rural financial markets. The second component of this competitiveness critique is that reform beneficiaries are generally ill-prepared to function as independent, commercially oriented producers.

It is important to note that this competitiveness critique ultimately applies to any land reform programme irrespective of the modality of redistribution (state grant versus market-assisted). The solution would seem to lie in rediscovering some old, mixed land reform models. Rather than establishing all beneficiaries as independent farmers, a mixed model might rely initially on a partially collectivized production process. Nicaragua in the 1960s and 1970s experimented with "dead furrow" cooperatives. In these cooperatives, land was planned and planted collectively, while individual families managed later cultivation and harvest stages on subplots demarcated by dead furrows (i.e. an unplanted furrow was used to mark one family's property).

Such semi-collective organizations can ease access to capital. In the early stages of the household responsibility system reforms in China, a similar semi-collective approach was taken. An important lesson from China (and from the more general failure of fully collectivized agriculture) is not to make the collective arrangements permanent. Instead, they can provide a transition period while reform beneficiaries learn the business of farming, establish credit reputations, etc. The key here would seem to be to establish individual property rights firmly while still easing the transition to individual farming through the establishment of an initial semi-collective model that itself has a definite time at which it will disappear and from which individuals can exit.

In summary, criticism of market-assisted land reform programmes has crystallized competitiveness issues that any modality of land redistribution must ultimately address. Whether market-assisted programmes can actually work by providing land to intended beneficiary populations on a sustainable basis has yet to be determined. Given current interest in the topic, the moment seems right for experimentation and serious monitoring and evaluation of marketassisted and other land reform modalities designed to shift the ownership structure of land.

DESIGNING LAND POLICY AND PROPERTY RIGHTS REFORM FOR POVERTY ALLEVIATION

The renewed attention in policy circles to land and property rights reform presents an opportunity to make economic growth work better for the less well-off members of society. It also brings a responsibility to understand the complexity of rural markets well enough to design those programmes well. This paper has couched much of the overview of the issues in ambiguity and uncertainty. However, countries such as Brazil, the Philippines and South Africa that are implementing land and property rights reform do not have the luxury of being able to wait for the full resolution of that ambiguity. In that spirit, here then are some key conclusions that can be drawn from the literature:

• Land is disproportionately valuable to poor households. Providing a hectare of land to a landless rural household will boost the family's net income significantly, and significantly more than it will boost the income of a wealthier household. This latter finding is especially important because it indicates that land permits the poor household to make better use of its labour and other endowments, something that is good for the family and for the overall economy. Thus, enhancing the land access of poor rural households is good social and economic policy.

• Assigning land rights to women enhances the land policy effectiveness. Controlling for household income and other factors, and assigning property rights over land to women, will have modest but significant effects on both household food consumption and educational attainments. Thus, land policy should pursue the assignment of rights to women proactively.

• Wealth-biased rural financial markets dampen the impact of land access on poverty. Given the ways in which rural financial markets actually work, most poor rural households are creditconstrained. These constraints reduce the abilities of poor families to extract the most income from land and other resources. Therefore, a truly effective land programme must also address constraints in rural financial markets.

• Land markets by themselves will do little to transfer land to the rural poor. While measures to increase tenure security and improve the fluidity of land markets have modestly improved the ability of poor households to rent land, for the most part the land market by itself has very little impact on the overall structure of landownership and access. Thus, proactive policies such as land banks are necessary if the benefits of improved land access are to be enjoyed.

• There is more to learn (re-learn) about models to secure beneficiary competitiveness. The ample evidence that incentives attached to individual farm ownership are powerful and important does not mean that completely individualized farming is the optimal initial institutional arrangement for land reform beneficiaries. Temporary, semi-collective farming may be a useful first step towards securing beneficiary competitiveness. Experimentation with and careful evaluation of reform models and modalities of redistribution (market-assisted versus state-mandated) must be priorities.

REFERENCES

Burgess, R. 1999. Market incompleteness and nutritional status in rural China. Working Paper. London, London School of Economics. (mimeo)

Carter, M.R. 2001. Land ownership inequality and the income distribution consequences of growth. WIDER Working Paper No. 201. Helsinki, UNU/WIDER.

Carter, M.R. & Mesbah, D. 1993. Can land market reform mitigate the exclusionary aspects of rapid agro-export growth? World Dev., 21(7): 1085 - 1100.

Carter, M.R. & Olinto, P. 2003. Getting institutions right for whom? Credit constraints and the impact of property rights on the quantity and composition of investment. Am. J. Ag. Econ., 85(1): 173 - 186.

Carter, M.R. & Zegarra, E. 2000. Land markets and the persistence of rural poverty in Latin America: post-liberalization policy options. In R. Lopez & A. Valdes, eds. Rural poverty in Latin America: analytics, new empirical evidence and policy. Basingstoke, UK, MacMillan Press.

Conning, J., Olinto, P. & Trigueros, A. 2001. Managing economic insecurity in El Salvador: the role of asset ownership and labor market adjustment. Williamstown, USA, Williams College.

Cox, M., Niño de Zepeda, A. & Rojas, A. 1990. Politica agraria en Chile: del crecimiento excluyente al desarrollo equitativo. Santiago, CEDRA.

de Janvry, A., Gordillo, G., Platteau, J.P. & Sadoulet, E., eds. 2001. Access to land, rural poverty, and public action. Oxford, UK, Oxford University Press.

Deere, C.D. & León, M. 2001. Empowering women: land and property rights in Latin America. Pittsburgh, USA, University of Pittsburgh Press.

Deininger, K. 2001. Market assisted land reform. In A. de Janvry, E. Sadoulet & G. Gordillo, eds. Land reform and public policy. Oxford, UK, Oxford University Press.

Lopez, R. & Valdes, A. 2000. Rural poverty in Latin America: analytics, new empirical evidence and policy. Basingstoke, UK, MacMillan Press.

Stiglitz, J.E. 1998. More instruments and broader goals: moving toward a post-Washington Consensus. UNU/WIDER Annual Lecture. Helsinki, UNU/WIDER.

World Bank. 2003. Land policy for growth and poverty reduction: a policy research report. Washington, DC, World Bank, and Oxford, UK, Oxford University Press.