Günther Schönleitner, Discussing Brazil's agrarian question

While some theorists have already written the obituary on agrarian reform, the issue has remained quite alive in Brazil and it seems that it is not going to disappear from the political agenda in the near future. Too deeply rooted are the social problems that fuel demands for redistribution of land. However, the nature of the debate on agrarian reform has changed quite clearly, as has its social, political and economic context. In the 1950s and 1960s the debate focused on the role of agrarian reform in the process of capitalist development. By eliminating semi-feudal relations of production, agrarian reform was perceived to be a precondition for capitalist transformation in the countryside. This has clearly been proven wrong, given the subsequent ”conservative” modernisation of agriculture. In the 1970s and early 1980s agrarian reform was advocated for its support of industrialisation by providing more and cheaper wage goods for the urban masses. Even this case has been weakened by the declining weight of small producers in the supply of food vis-à-vis the modern large-farm sector (Abramovay, 1994: 96). In the 1990s, some authors argue that by modernising its agriculture first, Brazil has lost the historical opportunities of reforming its agrarian structure in any fundamental way. Others emphasise the crucial role of agrarian reform for solving the country's social dilemma. This line of ”equity” arguments focuses on the need for poverty alleviation, income generation, employment creation and extension of ”citizenship” to the rural poor.

Given the recent ”strategic turnaround” associated with the abandonment of Import-Substituting Industrialisation (ISI) and the shift to a new economic model based on neo-liberal policies, both the macroeconomic framework and the policy approach to agrarian reform are undergoing fundamental transformations. The theoretical framework underlying the new economic model, neo-classical theory, faces a dilemma with respect to agrarian reform. On the one hand it takes institutions like the land tenure system as ”given” (Lipton, 1993: 641) and is reluctant to interfere with existing property rights. On the other, it aims at the most efficient possible allocation of resources which is hindered by the prevailing agrarian structure. Thus, neo-classicals are ”drawn toward land redistribution, partly so as to optimise farm size” (ibid.). It is common place to point out the inherent contradiction and inevitable trade-off between equity and efficiency. However, in a country where agricultural modernisation was characterised by massive state intervention in the sense of a heavy regressive policy-bias in favour of large farms, policies aiming at efficiency are also likely to work to some degree in the direction of equity. Thus, by turning the incentive structure against land concentration and by reducing distortions encouraging the use of capital rather than labour, the new economic model has created more favourable conditions for substantial changes in Brazil's agrarian structure than the previous developmentalist epoch. However, in terms of its actual policy approach towards agrarian reform, the present government appears to be caught in similar ambiguities as its ideological underpinnings. Although it aims at redistributing land to 280,000 families until 1998, prominent Brazilian politicians view agrarian reform merely as a means of compensatory social policy, denying its economic rationale or even declaring it economically irrelevant, as did the Minister of Planning, Antônio Kandir, in a seminar on the issue earlier this year. This kind of reasoning is clearly reflected by the organisational separation of agrarian (=social) and agricultural (=economic) policies in two different ministries. Moreover, while some ministers declare agrarian reform dead (or irrelevant), the government has discovered the virtues of family farming adopting an ambitious rural development programme (Pronaf) to support it, and currently designs alternative mechanisms of land transfer to the poor.

Without intending to give a complete coverage, this paper attempts to critically review and analyse these recent trends in Brazilian policy and debate with regard to the contemporary meaning, economic possibility, and potential scope of agrarian reform in the context of a ”conservatively” modernised agriculture and the new neo-liberal economic model. Agrarian reform is understood in this paper in a broad sense as state policies designed to transfer land rights from large holdings to poorer rural groups. The concept of agrarian structure is used in the sense of Thiesenhusen's definition of the term as the institutional framework of agriculture including, for instance, the patterns of land ownership, production system, social hierarchy, and class groupings (1990: 394). The essay is divided into five sections. In the first part I will review and discuss the nature of Brazil's agrarian structure and point out the reasons that recently have given rise to increased demands for land reform under social and political considerations. The second section deals with the economic rationale of agrarian reform by reviewing the international debate on the ”Inverse Relationship” and its fate in the context of technological change. The third and fourth chapters examine empirical evidence from diverse data sources and studies on the performance of small and large farms in Brazil. Moreover, they analyse the effects of state policy on the agrarian structure and their implications for the prospects of agrarian reform. In the final section I shall outline the policy options made available by the new economic model and assess the recent government policies in the tension between ”social compensation” and a genuine family farm strategy.

Brazil's extreme concentration of land ownership has deep historical roots. Going back to large land grants by the Portuguese crown, the latifúndio-minifúndio system has become the dominant characteristic of the rural economy over the centuries. In the absence of agrarian reform, the patterns of land ownership and income inequality in the countryside have changed little over time. The agrarian structure has remained essentially exclusive, constantly reproducing inequality, poverty and political domination. In 1980 the Gini-coefficient for family land ownership was 0.86, and when landless families working in agriculture are included, it was 0.9 (Maddison, 1992: 91). In 1985 the top two percent of landowners controlled 57 percent of total farm land, while the bottom 30 percent of farmers owned just one percent of the farm land (Hall, 1990: 206).

Brazil's agrarian structure: bimodal, heterogeneous, partly modernised, and socially inefficient

Brazil's agrarian structure can be categorised as predominantly bimodal since it is essentially based on a duality of large estates relying on hired labour (with tenancy and sharecropping diminishing) and a small farm sector being marginalised. To put it in the terms of Kay (1980) and de Janvry (1981), Brazil has ”resolved” the conflict between landlords and peasants in favour of the landlord enterprise, leading to the proletarianisation or semi-proletarianisation of the peasants. It has chosen the so called ”Junker” path to capitalist agriculture characterised by the transition from pre-capitalist Grundherrschaft to capitalist Gutswirtschaft. It is widely accepted that the capitalist mode of production has come to be predominant in Brazilian agriculture. Capitalist modernisation has taken place without having had to previously re-establish an independent peasantry through agrarian reform, as suggested by Ghose (1983). Thus the modernisation of agriculture since the 1960s has been conservative, i.e. based upon largely unchanged patterns of land ownership. However, this process has not yet become comprehensive. It has remained partial and resulted in an agrarian structure that is in various ways heterogeneous: (1) concerning farm size (latifúndio, minifúndio, and medium-sized farms), (2) regarding scale of production (intensive/extensive; modern/traditional) and (3) with respect to the mode of production (capitalist and semi-feudal remnants).

Thiesenhusen's analysis of the changes of Brazil's agrarian structure during the 1970s has pointed out the major characteristics and tendencies associated with the modernisation process: a growing concentration of agricultural land, additional land on the frontier being absorbed mainly by large farms rather than by landless or land poor people; the persistence of a labour-absorptive small-farm sector with a substantial agricultural production despite declining land availability and sparing use of capital; a medium and large farm sector characterised by extensive use of land and capital as well as sparing use of labour; an inverse relationship between farm size and agricultural receipts per hectare, though yields tended to be higher on larger farms; large farms tended to be the most dynamic in terms of productivity change; capital was being substituted for labour, especially on medium and large farms in commercial areas; the role of family labour was diminishing and sharecropping nearly disappearing, both being replaced by wage labour (1990: 408).

The results of this study (based on 1970 and 1980 census data) underline both the inequity and social inefficiency of Brazil's agrarian structure. The data show a significant (though declining) proportion of totally idle agricultural land (33 million hectares in 1980, i.e. almost the size of Germany).[2]The wasteful use of land and capital on large farms is illustrated by a strong inverse relationship between the share of cropland and farm size; a converse relationship between farm size and the share of natural pasture; and an inverse relationship between farm size and both capital use per hectare and capital productivity. At the same time, Thiesenhusen estimates, there are 9 million landless or partially landless (sic) farmers (1990:400). This number may be exaggerated, given the more recent calculation by FAO/Incra (1996) identifying some 2.5 million landless farmers (tenants, sharecroppers, and occupants) within the still existing ”family farm sector”. The latter study points out that capitalist large farms need 60 hectares to create one job, as against only 9 hectares on family farms. Thus, it becomes clear that the patterns of factor utilisation are inefficient in terms of total factor productivity, in an economy that is capital-scarce but labour-abundant. Moreover, even in Brazil land must be treated as scarce since its productive capacity can be used to generate income for and alleviate the poverty of millions of poor Brazilians.

Poverty, violence and clientelistic domination: continued backwardness in the countryside

According to Eclac (Cepal),[3] in 1992 43 percent of Brazil's households lived in poverty, 56 in rural areas, 39 percent in urban zones. Ipea (1994) suggests that in 1990 41.9 million Brazilians were subjected to poverty, while living in the world's tenth largest economy in terms of GDP. In 1989 the richest 20 percent concentrated 51.3 percent of total income whereas the poorest 20 percent saw their share restricted to 2.1 percent of total income.[4] The fact that the country has registered a slight decline in rural poverty in recent decades is probably a result more of mass migration from rural to urban areas than of improved income distribution. Stédile points out that some 30 million rural dwellers migrated to the cities within 20 years, transforming the urban centres into ”disorganised megalopolis” (1993: 22) and this is probably put mildly.

Brazil's agrarian structure has been a source of social tensions, political domination and violence. Highly unequal distribution of wealth, income and human capital as well as extreme poverty in the large majority of the rural population have frequently caused ”social unrest”, leading eventually to political instability and violent repression (e.g. the 1964 coup). Moreover, poverty and inequality have forced the rural poor ”into clientelistic bonds with elites who offer minimal benefits and protection in exchange for obedience and political support” (Weyland, 1996: 5). Thus, even under the current democratic conditions, the rural poor are being co-opted and manipulated so as to vote in accord with clientelistic loyalties rather than in line with their real interests. The ”selling” of their votes in exchange for small gifts expresses nothing else than the abdication of their citizenship (and democratic rights) for satisfying immediate basic material needs. The huge landholdings that still provide the landed elite with enormous political power are, therefore, hardly compatible with the consolidation of Brazil's fragile democracy. Thus agrarian reform is likely to be a precondition for extending ”citizenship” to the rural poor by breaking the patterns of clientelism, understood as a pyramidal system of personalist relations based on unequal though reciprocal exchange, discouraging or suppressing horizontal linkages (Weyland 1996). Despite capitalist penetration, it is clear that the objective of ”social modernisation” (Johnston and Kilby, 1982) remains valid. Institutionalist theorists, however, raise an important caveat: latifúndios, perceived as ”endogenous” institutions, may in the most backward regions still fulfil political economy functions, e.g. by providing security or informal magistracy (Lipton, 1993: 643). For agrarian reform to be ”incentive-compatible” and ”power-compatible” (ibid.), it needs to replace or dispense with such functions. This is likely to require both new horizontal social organisation and more effective state institutions in the countryside.

The exclusionary patterns associated with the agrarian structure gained a new dimension during the process of conservative modernisation. Both capital-intensive modernisation itself, the fear of agrarian reform and, ironically, protective tenancy legislation provoked the eviction of tenants and sharecroppers on a large scale. Those who tried to resist expulsion often became victims of all sorts of violence, including hundreds of assassinations. Moreover, tenants were not only evicted from land already owned by latifúndios, but also from those (frequently public) lands that were being illegally appropriated by large farms (land-grabbing) during the accelerated concentration of land in the 1970s and 1980s. A further disturbing aspect of this ”structural violence” in the countryside is the re-appearance of slavery. In 1994 the Brazilian Ministry of Foreign Affairs[5] admitted 16,442 confirmed cases (victims) of forced labour in 1992. Souza Martins[6] estimates that in the last 25 years at least 90,000 people were temporarily enslaved, mainly on new estates in the frontier areas of the Amazon. Sutton (1994) perceives this phenomenon to be an element of conservative modernisation itself, since it is not limited to pre-capitalist remnants but can be found with modern rural enterprises as well.

In the 1990s, violence in the countryside continues, but its nature has largely changed. Its primary source is no longer resistance against eviction. The initiative has shifted to the rural poor who try to regain access to land by means of occupations of idle or under-utilised land of large estates. This strategy pursued chiefly by the Landless Rural Workers' Movement (MST) has caused serious conflicts and even massacres against land occupants. According to CPT[7] in 1996 54 assassinations, 71 assassination attempts, 12 cases of torture and 220 injuries occurred in 750 land conflicts. There are basically two reasons why the rural poor appear to adopt a more aggressive stance: Firstly, even the rural poor tend to have better access to information (due to modern telecommunication, mass media etc.) and slightly more access to education which increases their capability of organisation. Simultaneously, re-democratisation has allowed demands for agrarian reform to be channelled by social movements, and ”bottom-up” pressure to become the driving force of land distribution. Secondly and more importantly, the two ”exit options” which served in the past as ”security valves” for the pressure of an impoverished and marginalised rural population, are no longer functioning. The ambitious colonisation programmes designed under the military regime to give ”the land without people to the people without land” by opening up new land on the frontier (especially the Amazon basin), have with few exceptions ended in complete failure and therefore largely been abandoned. The second exit, migration seeking employment in the urban-industrial sector, has been largely closed as a result of the crises of the 1980s associated with the exhaustion of ISI policies, the subsequent economic adjustment, and the implementation of neo-liberal reforms.

Recent studies by Paes de Barros et al. (1995), Gonzaga (1996), and Rocha (1996) have confirmed a significant decline in the level of industrial employment in Brazil during the last eleven years. Industry responded to economic crises and increased external competition after trade liberalisation by reducing its labour force. This trend has probably been reinforced by a still overvalued exchange rate (used as an anchor against inflation), that cheapens imports of labour-saving technologies while penalising the export sector. The workers laid off by the formal economy have largely been absorbed by an urban informal sector that had grown by some 70 percent (creating estimated 17 million ”jobs”) during the 1980s. The informalisation of the labour market, due to job losses in the most modern and dynamic sectors and the simultaneous increase of informal activities, has caused a serious decline in the quality of jobs. This results in very low incomes, bad working conditions, and the absence of social security systems. In this context it is the threat of further massive rural-urban migration that links the agrarian question to social welfare in urban centres. This is likely to explain the massive popular support for agrarian reform registered by Ibope earlier this year: 88 percent would support a confiscation (sic) and redistribution of idle land. Industrial growth is unlikely to be able to solve the problem of urban poverty, at least in the short term. Bulmer-Thomas argues that under the new economic model a reduction of poverty can only be expected under the condition of high rates of economic growth led by non-traditional, labour-intensive exports. A reduction of income inequality, he argues rightly, would require rising real wages through a tightening labour market as well as state subsidies and transfers to the poorest (1995: 300f).

The reduced capacity of the urban-industrial sector to absorb rural migrant labour contrasts sharply with the long-term need for rapid employment creation. Although population growth has been slowing down during recent decades, the current rate is still around 1.8 percent a year. This means a rapidly increasing pressure on the labour market in the foreseeable future and, in the absence of ”exit options” for the rural poor, increased demand for agrarian reform. Therefore, it is not surprising that employment creation has become central in the recent Brazilian debate on agrarian reform. Abramovay and Carvalho Filho see as ”the great historical mission of agrarian reform” not so much to increase food production, but to allow the transformation of the rural poor into agricultural producers (1993: 41). Romeiro believes that ”agriculture is the only sector in which job creation presents a certain flexibility, being able to absorb relatively more workforce without a fall in productivity, through the intensification of land utilisation”. For him, the actual meaning of agrarian reform in Brazil is to reduce the supply-pressure on the urban-industrial labour market (1994: 132). Romeiro's point is not to totally block rural-urban migration, but to make it compatible with the jobs available in the urban-industrial sector. In his view, agriculture must function as a regulating buffer for the supply of unskilled labour (1994a: 80). The supply pressure of rural migrants would otherwise rule out a tightening of the labour market that is required for a dynamic redistributive process aiming at a reduction in poverty. Thus, agrarian reform is perceived to have a broader systemic role in an urgently needed process to make Brazil's development model more inclusive, allowing the poor to share the fruits of growth while being themselves a source of that growth. This focus has several implications. Firstly it means that the traditional social base of agrarian reform has partly changed, though it has not been destroyed as argued by Sorj (1997). Apart from giving landless peasants access to land and consolidating the still existing family farm sector, agrarian policy is required to go further, including the rural un- or underemployed as well as those urban unemployed ”still linked or willing to return to rural life” (Sorj, 1997: 7). This implies an inverted re-absorption process, the direct opposite of what has been envisioned by theorists from Marx onwards. Secondly, it means a partial shift in the policy tasks of agrarian reform and policy: creating rural employment at the opportunity cost required to reduce rural-urban migration or even to stimulate the opposite move. In this context, it is probably secondary for non-peasant groups whether they are provided with land ownership or with reasonably well paid rural jobs. This means that besides the redistribution of land, agrarian policy is required to encourage labour-intensive production in agriculture as a whole. Moreover, it must try to create living conditions in the countryside comparable to those in the cities (electric power, education, health, water, transport etc.).

Although the social and political case for agrarian reform in Brazil has been strengthened since the 1980s, land redistribution aiming at a substantial alteration of the existing agrarian structure must be supported by an economic rationale if it is to be successful and sustainable: the new agrarian structure must be at least as efficient as the old one. This economic justification has been provided by the often observed phenomenon that productivity per unit land is higher the smaller the farm.

One of the most comprehensive studies on agrarian structure and productivity was undertaken by Berry and Cline (1979) focusing on the different patterns of factor utilisation on small and large farms. They argue that the productivity of large farms tends to be lower than that of small farms because land utilisation declines as farm size rises. Due to lower output per area, heavy use of land and capital as well as sparing use of labour, large farms are also likely to be less efficient in terms of total factor productivity (social efficiency). The conclusion is that redistribution of land can ”combine the underused land with the underused labour and raise production” (1979: 7ff). A recent article by Dyer (1996/7) reviews critically the theoretical debate on the IR and tries to replace conventional theories by a class-theoretic framework. These two contributions will serve as a starting point for the following discussion of the most important arguments being put forward to explain the emergence of the IR.

Berry and Cline (1979) (and earlier Sen) have pointed out that in developing countries there exists a ”labour market dualism” of a modern sector using wage labour to the point where its marginal product equals the marginal cost (wage) and a traditional sector using family labour to a point where the marginal product is much lower than the market wage, possibly even zero. In the absence of alternative employment, the effective price of labour is lower on small farms which, therefore, can cultivate more land and achieve a higher output per unit land. An alternative though not conflicting “institutionalist” approach focuses on incentives and transaction costs. Wage labour and family labour face fundamentally different incentive structures which is reflected in the superior quality and intensity of labour input per hectare on family farms. Family members are the residual claimants to the profit they produce. Hence, they are rewarded in close correlation to their productivity and this clearly stimulates labour effort. Moreover, they share a part of the risks involved. Family farming reduces transaction costs associated with hired labour such as recruitment, monitoring, supervision, shirking and ”free-riding”. Discipline and responsibility are imposed on small farmers ”by the pressing incentive to secure the family livelihood” (Nolan, 1988: 43). The latter argument comes close to Dyer's class-based approach (1996/7: 123), which singles out as the only admissible explanation, that the IR arises in the context of a backward, semi-feudal agriculture in which poor peasants are driven to maximise output (rather than profit) because their survival depends upon it. Thus the IR is argued to be a sign of distress rather than of relative efficiency. Yet, this argument leaves unexplained why the IR can also be observed across farms well above the levels of size and ”distress” of those struggling for their ”naked survival”. Dyer's expected ”relax of labour effort” as a result of rising income (1996/7: 107) does either not happen or is not strong enough to invalidate the IR. In this sense Dyer's argument is no stronger than that criticised by him: the postulate of a superiority of family labour that cannot explain the IR running across farms above family size. With regard to the latter, one can argue, however, that the supervision of a restricted number of wage workers on medium-sized family farms is likely to be easier and cheaper and that transactions costs are lower than those on large estates.

Land and capital market imperfections lower the effective prices of land and capital for large farms, while increasing them for small ones. According to Berry and Cline, this leads to a higher labour/land ratio and higher output per hectare on small farms, while encouraging the substitution of capital intensive techniques for labour on large farms. This may ”offset somewhat the latters' tendency to low output per unit land” (1979: 10). Dyer rejects factor price differentials as an explanation for the IR, since in the context of higher capital intensity and technological innovation on large farms the IR is argued to break down (1996/7: 119). However, Dyer does not discuss Berry and Cline's qualification that cheap capital available to large farms tends to encourage high output per hectare only if (a) it indeed leads to higher capital/land ratios, (b) capital can substitute for land and (c) capital and labour are complementary. Otherwise cheaper capital may well replace labour by machinery without ameliorating the problem of under-utilisation of land on larger estates (1979: 11). Berry and Cline distinguish three forms of under-utilisation of land: declining yield per area cropped; a falling share of cultivated or double-cropped area; or a product-mix shifting from high to low value items (e.g. natural pasture). Hence, higher yields per hectare cultivated on large farms may be compatible with declining land utilisation as farm size rises (ibid, 13).

Landholding for asset or prestige purposes has provided a further explanation for the low productivity of large estates in Brazil. Commonly land has been held as a hedge against inflation, as store value, or as a source of political power. Land quality and market access (location) may also influence land utilisation and productivity. Dyer (1996/7: 109f) discusses the argument that better soil quality, water availability etc. are likely to attract higher population density leading in the longer run to smaller average farm size. The implication is that the IR arises because larger farms are commonly characterised by lower quality soils than those of small farms. This ignores however the historical, socio-economic and political context of the patterns of land appropriation. In many Brazilian regions the powerful latifundists have tended to appropriate and monopolise the best lands. Hence, land quality differences can hardly serve as an alternative ”exogenous” explanation of the IR in most of Brazil.

The existence of economies of scale is a crucial point since in their presence the IR is destroyed and the economic rationale for agrarian reform largely invalidated. Economies of scale may derive from improved technologies leading to more efficient production processes which require scaling up. Whether or not modern technologies demand large scale depends on their divisibility. Many new ”green revolution” technologies such as seeds, fertiliser, pesticides etc. are as effective on small as on large farms. Nolan (1988) stresses that for most elements of working capital, for livestock and land there is a high degree of divisibility of factors of production. Dyer contradicts stating that while these inputs may be scale neutral, their adoption steps up pressures for mechanisation associated with scale economies. The labour-related advantages of small farms may be more than matched by new technology-related advantages of large farms. According to his class-based approach, institutions, services (credit, extension, information, marketing), and political power are biased in favour of large farmers who are able to reap the benefits of new technology. Rather than small farmers being able to catch up, the initial advantages captured by large farmers are further strengthened over time due to intrinsic scale advantages (1996: 134).

Dyer's arguments are subject to several criticisms. Firstly, the increased per-hectare productivity associated with the green revolution seems to derive more from new divisible inputs (fertiliser, seeds etc.) than from indivisible machinery/mechanisation (see Cline, 1973: 150) whose utilisation in a labour abundant economy is likely to reflect distortions of factor prices. Secondly, the class-based theory recognises that the factors causing a possible break-down of the IR are policy induced rather than farm-size inherent. However, these policy factors are taken as ”given” leading to the conclusion that the IR inevitably breaks down destroying the case for agrarian reform. A much more plausible (and socially acceptable) conclusion would be to change the policy. Thirdly, by ruling out the possibility for small farmers to catch up in the process of technological innovation, Dyer ignores contradicting evidence, for instance from India (Rao, 1977) and Mexico.[8] Berry and Cline understand this process as an S-curve along which larger farms are likely to be the first to adopt innovation, being followed by small farms, thereby re-establishing the earlier IR on a higher level of output on all farms (1979: 28), a view that is shared also by the World Bank (1990: 60). Fourthly, the problem of a minimum farm size required for indivisible machinery can partly be resolved by either renting in land or renting out machinery (Binswanger, 1990: 346).

A further possible source of scale economies is modern management. It is useful, however, to distinguish between the agricultural production itself and ”support operations”. With farming itself, it has been argued, managerial diseconomies of scale derive from inherent inefficiencies of increased complexity of organisation as farm size rises. According to Nolan (1988), these derive mainly from the sequential and dispersed nature of farm work which makes labour supervision difficult. Moreover, he argues, small farmers commonly have better knowledge of what is required on the site than the manager of a large estate. The former are able to respond quickly and in a flexible way to changes in the natural environment. As stressed by Dyer, this advantage may be outweighed, however, by the superior knowledge of managers concerning the use of modern technology and by increased scope for division of labour and specialisation on large farms (1996: 114). With regard to ”sup-port activities” like credit provision, processing, transport and marketing of inputs and outputs, there can be little doubt that central management is associated with economies of scale. Certain crops, for instance, demand tight co-ordination of growing, processing and marketing activities. In addition, foreign export markets commonly impose the need for high and uniform quality standards that can best be achieved by centralised supervision and control. Lipton (1993: 645f) emphasises therefore that ”the IR does not apply to all farm outputs, at all factor price ratios, across all activities from ploughing through weeding to marketing, or selecting any technique”. He suggests correctly that small farms do reduce unit labour-related transaction costs, whereas larger farms may reduce unit capital-related transaction costs, especially for credit, marketing and exploiting economies of scale where these apply (ibid., 648).

In terms of policy conclusions, Berry and Cline advocate a strategy of improving small farm access to credit and new technology coupled with comprehensive land redistribution. The optimal post-reform farm size, in the absence of technical economies of scale, is argued to be the total agricultural area divided by the total number of families working in agriculture (1979: 18). This conclusion, however, restricts itself to farm operations and ignores possible advantages of larger scale associated with ”support operations” and management (as discussed above). Therefore, their findings have to be qualified in Lipton's sense, who suggests forms of ”activity splitting”, allowing to combine the advantages of small farms with those of large scale management and services (where appropriate). Breakdown or preservation of the IR are likely to depend more on policy framework and institutional arrangements than on factors inherent in modern technology. Their more dynamic adoption of new technology in the process of agrarian change has given rise to the impression of a superior productivity of large farms. In this situation it is a real danger that ”the large farmers' accurate perception of their own self-interest and the policy-makers' inaccurate perception of the national interest will combine to assist the large farmers to expand at the expense of the smaller ones” by pursuing a large-farm, capital-intensive strategy (Berry and Cline, 1979: 139).

The standard arguments against redistributive policies such as agrarian reform have been that the country could not afford them, they would jeopardise economic recovery, hamper growth and put at risk macroeconomic stability through irresponsible (”populist”) expansionary policies. Moreover, they would scare off foreign capital, pose disincentives to entrepreneurs and imply losses in economic efficiency. These arguments are entirely in line with the conventional wisdom that inequality is not only a by-product of growth, but also conducive to it. The latter assumption has been justified on the grounds that the rich would save more of their income than the poor who spend most of it on consumption. Higher saving rates would increase investment which in turn accelerates growth. Cline argues that long-term growth may be adversely affected by land redistribution if large farmers have a higher marginal propensity to save than small farmers, or if the former are more receptive to technical change (1970: 175).

The ”East Asian Miracle” has challenged these widely accepted propositions. The success stories of countries like Taiwan and Korea was initiated by redistributive policies such as land reform and massive investment in human capital. Their rapid growth has been accompanied by increasing social equality. While Brazil's Gini-coefficient for income distribution lies above 0.6, East Asian figures are around 0.25. At the same time they have much higher rates of private domestic savings than Brazil and growth rates have tended to be higher in these countries than even those of pre-1980 Brazil. Although these empirical findings are not sufficient to establish a causality running from low inequality to growth, they do permit the conclusion that inequality is neither a necessary precondition nor an inevitable by-product of economic growth. Currently empirical and theoretical research on the links between inequality, growth and savings is being undertaken. Some theorists have come to establish a negative correlation between inequality and growth. Even the World Bank (1993) has recognised that low inequality or ”shared growth” has been an essential element of the East Asian success.

Birdsall et al. (1996) find that the credit-constrained poor cannot borrow even to finance high yielding investments and their poverty precludes a substantial reduction of consumption to finance investments. However, if the returns to their labour are sufficiently high, rational poor households will intensify their work effort to generate additional income for productive investment (e.g. in the education of their children or new agricultural technologies). In order to trigger an investment boom by poor households, returns on both their investments and labour will have to increase. However, there is no consensus about the virtues of redistribution with respect to savings. Schmidt-Hebbel and Servén (1996) found no evidence that income inequality affects aggregate saving across countries negatively. They argue that, on theoretical grounds, higher inequality is, ceteris paribus, likely to be associated with higher savings. Perhaps more emphasis should be given to the question of what happens to the savings of the rich when compared to those of the poorer deciles. It seems that the latter are less likely to end up as flight capital, as investments in unproductive luxury goods or as speculative capital than the former. Moreover, it is not only the quantity of private domestic savings that matters for accelerating growth, but also the quantity and quality of productive investment that is financed out of them.

Berry and Cline report that simulations for Brazil indicate only a limited impact of rural income redistribution on rural savings rates (1979: 28) whereas Birdsall and Sabot (1994) emphasise that high inequality is likely to constrain the country's growth in the long run. The latter point out that ”the ratio of the income shares of the top 20 percent to the bottom 20 percent is 26 in Brazil and 8 in Korea. Simulations suggest that if, in 1960, Brazil had had Korea’s lower level of inequality Brazil's predicted growth rate over the following 25 years would have been 0.66 percentage points higher each year. This implies that after 25 years GDP per capita in Brazil would have been 17.2 percent higher”. The policy recommendations advocated on these grounds are: improving the investment opportunities for the poor and shifting to a more labour-demanding growth path. Both are crucial objectives of Brazil's agrarian reform.

Significant scale economies in Brazilian agriculture have as often been denied as the IR confirmed. Cline (1970) showed that there are constant returns to scale for inputs actually used. Therefore, he concludes, agrarian reform could create small family farms without loss of efficiency. He verified that land utilisation declined as farm size rose. Factor ratios were found to vary systematically away from labour and towards land and capital as farm size increased. This study was based on a 1962-63 sample survey. Relying on another joint survey by the World Bank and Sudene for North-eastern Brazil in 1973, Berry and Cline (1979: 130) stress that no significant changes of the patterns of productivity had been found in the decade from 1963-1973, suggesting that the 1973 data confirm the socially inefficient land use by large farms. Thiesenhusen and Melmed-Sanjak (1990) verify, on the basis of 1970 and 1980 census data, the existence of an IR and ”close-to-constant” returns to scale both in 1970 and 1980. Guanziroli (1990: 112f), relying on data provided by Albuquerque (1987), confirms that there are in general no significant economies of scale, except for some individual crops. He agrees, however, with Albuquerque, that ”all groups of farmers are allocatively efficient” since their partial productivity rates correspond with the relative availability of factors. Thus, as farm size rises, land and capital productivities decline and labour productivity increases. Therefore, he concludes, agrarian reform cannot be advocated nor dismissed on the grounds of the superiority of either small or large farms. Guanziroli ignores, however, that it is land and capital not labour that must be treated as scarce. The allegedly ”allocatively efficient” factor utilisation on large farms is likely to reflect both the logic of the IR and heavy policy-induced distortions of factor prices shifting away from the relative scarcity values of the Brazilian economy. It is therefore inefficient and does provide an economic justification for agrarian reform.

Looking in more depth at efficiency and/or inefficiency of the small farm sector requires first a definition of what is understood by small or family farms. The three studies being compared here use different delimitations which have to be kept in mind. Thiesenhusen et al. (1990) use a single criterion to define small farms: a farm size of up to 50 hectares. FAO/Incra (1996) delimitate ”family farms” by multiple criteria: the farm work is managed by the producer himself; there are no sub-contracted services; there is (apart from family labour) an average hired workforce of either one permanent wage labourer and up to three temporary workers, or no permanent and up to four temporary workers; a farm size of up to 500 hectares in the South and Southeast and 1000 hectares in other regions. This relatively large size is justified by the need to capture the US-style medium farms that operate large areas almost exclusively by family labour. The number of these farmers is, however, likely to be very low since the average size of the so defined family farm sector is only 19.1 hectares. Whereas the two aforementioned studies do not separate ”traditional” and ”modernised” farms, mixing up different systems of production, FAO et al. (1992) analyse exclusively the performance of resettlement areas implemented under the PNRA between 1985 and 1991. Thus they refer to resettled farmers with on average 31 hectares, using fairly uniform (traditional) techniques.

Thus the 1980 census data analysed by Thiesenhusen et al. (1990) show that the small farm sector amounts to 82 percent of total rural properties, owning together just 12 percent of total farmland. The sector accounts, however, for 39 percent of total cropland, 28 percent of capital and 71 percent of agricultural labour. The small farms earned 40 percent of total farm receipts and 50 percent of total net agricultural income. A share of 78 percent of the small farms' production was marketed as against 86 percent for the entire sector, suggesting that own-consumption does not weigh too much on small farms. In 1980 the smallest farms (under 1 hectare) held 90 percent of their land under crop (mostly basic food crops) as against only two percent on the largest farms (over 10,000). Moreover, small farms used more capital per hectare than medium and large farms. In conclusion, the data confirm that small farms are more productive than large farms since they use land and capital more intensively.

FAO/Incra (1996) (based on 1985 census data) came to slightly different results analysing the ”family farm sector” as defined above. Brazil's 4,339,053 family units make up 75 percent of total ”production units”, operating 22.2 percent of total farmland, employing 59.5 percent of agricultural labour (with strong regional differences), and accounting for 28 percent of the total value of production. The authors stress that the latter figure may be understated since in the absence of accounting on many farms it is difficult to obtain values of production; farmers depending on intermediary traders tend to get lower prices; own-consumption tends to be underestimated; and some family farms may have been excluded by the above criteria. Therefore, the sector may well contribute some 40 percent to the total production value. While 44.2 percent of family farms had access to credit, their share in value terms was only 10.7 percent and when the exceptional level of the South is disregarded, the figure falls to 5.4 for the rest of the country. Because the family farm sector is quite a heterogeneous grouping, the study disaggregates it in two ways: firstly by gross monetary income dividing it into three categories: a consolidated group A encompassing 26.5 percent of the sector with an average farm size of 32.1 hectares and an annual gross monetary income of 57.1 minimum wages; a transitory group B with 23.5 percent of farms, operating on average 16.1 hectares, earning 12 minimum wages; and a peripheral group C with 50 percent of the farms, disposing of 13.7 hectares, achieves an annual gross monetary income of only half a minimum wage. The second criterion focuses on the legal condition: 52.7 percent are owners, 6.0 tenants, 3.4 sharecroppers and 37.8 occupants without any legal title. Surprisingly, the share of owners in group C is 48.8 percent. Therefore, it becomes clear that neither a certain farm size nor ownership are sufficient conditions for their advancement to income groups B or A. Further determinants are likely to be the quality of natural resources, access to credit and technical advice etc. However, the absence of ownership titles both weakens the incentives to invest and hinders the access to credit markets. Thus it is likely to result in lower productivity.

FAO et al. (1992) examined (through a sample survey) the socio-economic indicators of the 525 resettlement projects of the PNRA, redistributing land to some 94,026 families between 1985 and 1991. The most important finding was that of an average monthly income of 3.7 minimum wages per family, consisting of receipts from cropping (22.03 percent), livestock breeding (15.12), own-consumption (37.01), off-farm employment (11.17) and occasional non-agricultural receipts from wood, coal, small commerce etc. (14.67). The farms spend on average 15 percent of gross income on production costs. In terms of income generation the overall evaluation is positive since the families reached an income level that places them between the groups A and B of the above study (FAO/Incra 1996), approximately at the national average family income in 1985 (3.82 minimum wages). However, the average figures mask not only huge regional differences but also high income differentials within the settlements. The study identifies a sub-sector of emerging commercial family farms (some 45 percent), a poor peripheral segment tending to abandon the projects (15 percent) and the remainder on the way to integration. In terms of productivity the study found physical yields per hectare cultivated to be lower than those of the regional context. This is attributed to technological backwardness and the short time passed since the implementation of the projects (on average three years). In this context it is worth mentioning that it is the degree of market-integration of the smallholders that was found to be determinant for the adoption of technological innovation, not vice versa. As Abramovay (1994: 138) suggests, there are three possible criteria for evaluating the performance of resettlement projects: (1) a focus on the essentially social nature of agrarian reform, disregarding economic feasibility, (2) viability defined by the opportunity cost of the settlers' labour (followed by FAO), and (3) viability defined by productivity and competitiveness relative to the commercial capitalist sector. The study suggests that by the second criterion they perform well (achieving higher incomes than other rural or urban workers), while they tend to lose out in terms of productivity and competitiveness. In the latter sense, performance may still be disappointing, but it should be kept in mind that this ”inferior productivity” refers to physical yields per cultivated hectare, not to total farmland productivity. Moreover, it must be compared with the productivity levels of the same areas under the previous latifúndio ownership. This would probably draw a more positive picture of the settlements' performance. In addition, the FAO/Incra study highlights several resettlement projects that were able to surpass also regional productivity levels per hectare cultivated. A crucial factor for their success has been technology adoption which in turn was determined by access to credit and market integration.

The data of the 1970 and 1980 censuses suggest that the large farm sector (all above 50 hectares) tended to be less productive (per unit land, not per unit cropland) but more dynamic than small farms. This dynamism is reflected by a significant increase in its contribution to overall agricultural production, rapid adoption of new technology and high rates of productivity change. Thus, Brazil's bimodal agriculture is not characterised by stagnation, although by international standards its productivity levels are still low. It must be kept in mind that this productivity growth of large farms started from very low levels and that in the 1970s the expansion of agricultural land rather than higher yields was the primary cause of increased production (Thiesenhusen, 1990: 408). In the 1980s, however, this has probably been inverted, with productivity growth being more important than expansion of area. It is clear that the process of conservative modernisation has led to the emergence of a modern agroindustrial segment within the large farm sector that is highly integrated into both domestic and foreign markets, producing primarily higher value crops and achieving better yields per hectare cultivated. Thus, the large farm sector is not a homogeneous bloc. Modern commercial farms coexist with traditional latifúndios, but unfortunately, the census data do not allow to separate them. Nevertheless, it is useful to disaggregate the sector to at least two size categories: a farm size from 50 to 2,000 hectares and from 2,000 to over 10,000 hectares. It is worth noting that it is the former ”medium-size” sector that accounts for both the bulk of Brazil's export crops and a significant share of commercial food crop production.[9] Moreover, the census data show an abrupt decline in net income and farm receipts for the largest farms relative to the sector from 50 to 2000 hectares as well as a decrease in capital productivity by nearly 47 percent.[10] Its increased economic weight suggests that this ”medium-size” sector has been the engine of modernisation. The farms with more than 2,000 hectares are likely to exceed what could be justified by possible scale economies deriving from indivisible machinery or modern management. Therefore modernised large farms may hold huge land reserves but are unlikely to operate them intensively beyond the margin where they exceed the ”technically required size” even in the case of technical or managerial economies of scale. This excess land is commonly used for extensive cattle raising or kept idle.

Rent-seeking, capital-intensive modernisation, and the marginalisation of rural labour

An analysis of the sources of conservative modernisation and the dynamism of the large farm sector suggests that neither Brazil's latifúndio-dominated agrarian structure nor its capital-intensive modernisation are outcomes mainly of market processes or the requirements of technological progress. Instead, they can largely be attributed to the prevailing distribution of political power and the resulting pervasive state intervention in favour of those who dominate the polity and have vested interests in maintaining the status quo in the country's agrarian structure. As a result, the state engaged in two types of policies: (1) strategies aiming at substituting for agrarian reform without changing the agrarian structure, such as colonisation projects trying to deviate the pressure for land redistribution to frontier regions, disregarding both economic viability and ecological sustainability; (2) the massive use of the instruments of agricultural policy to induce the modernisation of latifúndios, pursuing a capital-intensive large-farm strategy. With the implementation of the ”National Rural Credit System” (SNCR) in 1965, subsidised credits became the most important instrument of this policy. Guedes Pinto points out that during nearly two decades real interest rates were negative and by the late 1970s the repaid value was on average less than 50 percent of the borrowed value. In the mere 20 percent of farms that had access to this kind of credit, the resources were concentrated in the hands of a small number of large landowners. In some years one percent of the largest borrowers accounted for 38 percent of total credits. This abundant and cheap money was often used to acquire more land, facilitating even further their privileged access to credit (1995:79). According to Denslow and Tyler[11] credit subsidies amounted to 21 percent of the total value of agricultural output in 1980. In addition, generous fiscal incentives and income tax breaks stimulated activities like large-scale extensive cattle raising or the production of sugarcane for the alcohol programme (substituting gasoline) and soybean monoculture. Tax policy allowed profits from other sectors being invested in agriculture to be largely exempt from income tax. Moreover, the ineffectiveness of the land tax and pervasive tax evasion by large farmers contributed further to a policy-induced land concentration and a way of modernising agriculture that was essentially based on rent-seeking. A further important factor that fuelled this process was the high inflation rate that made land an extremely attractive asset for speculation on the basis of its ”store value” rather than its productive capacity.

Since the conservative modernisation of Brazil's agriculture was accompanied by accelerated land concentration, capital/land ratios on large estates are unlikely to have increased in a way as to reduce significantly the under-utilisation of land. This is confirmed by the census data that show an inverse relationship between capital use per hectare and farm size, although this relation becomes converse for capital use per hectare cropland. Interestingly, the growth rates of machinery value per hectare cropland between 1970 and 1980 are u-shaped with the highest growth rate being registered in the category of 10-50 hectares.[12] This suggests, on the one hand, that small farms do catch up in the adoption of new technology. On the other, it is likely to reflect the distortions of factor prices (against labour and in favour of capital) associated with a conservative modernisation whose character has necessarily been capital-intensive and labour-displacing. More labour-intensive production processes would hardly have allowed the conservation of the existing agrarian structure. In summation, conservative modernisation of agriculture managed to increase both yields per hectare cultivated and overall agricultural production but failed to ameliorate the under-utilisation of land and exacerbated the problem of unemployment while making wasteful use of the scarce factor capital. Thus, agricultural modernisation and decades of dynamic growth have done little to reduce rural poverty and inequality. Instead they have played a major role in an overall development model that has been essentially exclusive, transforming large parts of the population into a marginalised superfluous mass that is neither needed as workforce nor as consumers.

De Janvry analyses the conservative modernisation of agriculture under the threat of agrarian reform in Latin America by using a game-theoretical approach. He understands this process as a non-co-operative game between landlords and the state in which the latter tried to force the former into modernisation levels that would increase total factor productivity at social prices above those of small farms. This game was lost by a state that was unable to predict the landlords' successful collective action allowing them to manipulate the state, to engage in rent-seeking activities and to extract the promise of non-expropriation-if-modernisation. The result was a political deadlock and the loss of opportunities to create net social gains via redistributive land reform (1989: 1398). Although in the Brazilian case its nature was collusive rather than non-co-operative, it is quite clear that this deviated developmentalist game came to an end during the 1980s and early 1990s as a result of the crises of the ”lost decade”, economic adjustment, and the gradual adoption of market-oriented reforms. Most of the perverse fiscal incentives and tax breaks were abolished. It was, however, primarily the dramatic reduction of credit subsidies that affected the large farm sector severely. According to Romeiro, the subsidy element of rural credits fell from 38 percent in 1980 to 2.3 percent in 1985 (1994: 129). In this context it is instructive to look at the evolution of annual agricultural growth rates between 1980-85 vis-à-vis those between 1975-80: the growth rate of the physical product increased from 1.82 to 3.01 percent, productivity growth per unit land rose from -0.58 to 2.36 percent, the growth rate of agricultural area per rural labourer decreased from 1.6 to -1.28 percent and the growth rate of the rural workforce increased from 0.79 to 1.92 percent. The growth of producer prices fell from 3.29 to -0.52 percent as a result of increased supply and falling international prices (ibid., p. 128). This suggests that the reduction of pro-capital and pro-latifúndio distortions of factor prices contributed to increasing production, productivity, and labour use in agriculture. These data, Romeiro argues, reveal the enormous potential of employment creation in Brazilian agriculture, whereby he considers reduced land speculation and open access to land to be sufficient conditions (1994: 130). However, Guanziroli points out that this liberalisation approach applied during the 1980s had only slight consequences for land prices and the agrarian structure (1990: 107). This was, however, largely due to the fact that the second engine of land speculation, inflation, not only remained in place during the 1980s but was aggravated into hyperinflation. With the massive drop in inflation after the 1994 stabilisation plan ”Plano Real”, this picture has changed sharply. With high real interest rates and low inflation, the opportunity cost of capital tied up in idle or underused land has risen in a way that makes landholding for speculative purposes unattractive. More and more landlords offer land voluntarily to Incra for expropriation, while land prices are falling. What seems to be happening is exactly the scenario envisaged by Guanziroli (1990: 135) as a desirable precondition for agrarian reform: without credit subsidies and inflation many largeholders are ”removing themselves quietly to urban real estate speculation or to liberal professions”. Thus, market reforms and stabilisation have created the most favourable conditions for decades for a deconcentration of land and substantial changes in Brazil's agrarian structure.

Thus, in the aftermath of the crises of the 1980s and the subsequent change of its development strategy, Brazil is presented with new opportunities to resolve its agrarian question. In this situation the country has the choice of two options. The first is to maintain the bimodal agrarian structure, continuing to rely on capital-intensive large farms for agricultural growth while using the new opportunities for reducing social tensions and poverty in the countryside. This indeed restricts the meaning of agrarian reform to compensatory social policy (with temporary effects). The second option would be a comprehensive reorientation of reunified or at least complementary agrarian and agricultural policies. These would seek the gradual convergence of the diverse agricultural sub-sectors towards overcoming bimodalism or, at least, narrowing its extreme polarity in terms of farm size and scale of production (technology). Thereby it would shift to an inclusive poverty-reducing and hence labour-demanding growth path. With regard to land transfers, this option would combine an induced and partly enforced reduction of farm sizes in the large farm sector with the consolidation of existing family farms, titling programmes and resettlements of landless peasants.

In the recent Brazilian academic debate there is, with few exceptions, consensus about both the continued (or even increased) social need for agrarian reform and the notion that this reform must not threaten the dynamic modern sector. Guanziroli (1990) advocates that family farms should make use of non-traditional ”niches” that are not or cannot be penetrated by agro industry. This means that the reform has to remain partial (restricted to idle and under-utilised land) and within the prevailing bimodalism. It is often emphasised that, due to Brazil's immense territorial expansion, the social objectives of agrarian reform need not touch the modern sector in terms of land. Meanwhile, the problem of this approach lies rather in the (in)compatibility of conflicting policies for different sub-sectors. The maintenance of bimodalism probably implies separate (and conflicting) agricultural and agrarian policies which are likely to create (or better maintain) adverse conditions for resettled small farmers, defeating in the longer term even the limited objectives of an agrarian reform understood as social compensation. Agricultural institutions determining the access to credit, technical advice and modern inputs, as well as price policy, marketing structures and agricultural research/extension are still biased towards large capital-intensive farms, and need to be redirected. Thus, while the reform may be partial in terms of land transfers, it cannot be so in policy terms, i.e. it cannot be built upon a two-tier agricultural strategy without being short-lived.

Therefore, the second option seems to be more coherent. If agrarian reform is to have a significant social impact, its primary objectives must be to encourage the use of labour rather than capital as well as that of divisible modern inputs rather than excessive labour-displacing mechanisation. This would allow small farms to bring to bear their genuine advantages: their superior labour-input per unit land in terms of quantity and quality. As a result, higher labour-related transaction costs on large farms would encourage the renting out of land and/or tend to create smaller production units, driving agriculture towards a small/medium farm structure. This would not only create the foundations for the strengthening of family agriculture which has recently come to prominence in Brazilian policy and debate. It would probably give significant impulses to the rural non-farm sector. Yet, such a policy must not overlook either possible scale advantages of larger farms associated with capital-related transaction costs and management as discussed in section 2. It is with this background that many authors (e.g. Binswanger, Nolan, Lipton) have suggested a combination of the advantages of small and larger farms through institutional arrangements allowing different activities and services in the farming process to be provided on appropriately different scales. Technical advice, finance, processing and marketing services, transport, and even tractor-ploughing and fertiliser application can be provided for small farms either co-operatively or privately by larger firms. Experiences from tea production in Kenya (Swainson 1986) and banana plantations in the Philippines (Hayami et al., 1990), for instance, show that contract farming arrangements can be successful even in the context of highly competitive export markets through smallholder-multinational co-operation. Lipton alerts rightly, however, that such arrangements should not be imposed exogenously by reform packages (1993: 650). The crucial policy task is to create favourable conditions for them. Moreover, monopoly/monopsony relationships between smallholders and the collective, private or public providers of lumpy inputs must be avoided. Small farmers may otherwise find themselves exploited (even by their own co-operatives) just as much as they were by traditional latifúndios. White's account from Indonesia is an instructive example in this respect (1997: 131).

While market reforms and the reduction of pro-latifúndio distortions have created favourable opportunities for structural change, they do not by themselves transfer land to the poor client groups of agrarian reform. Some form of state intervention is necessary. Brazil's government currently outlines its new agrarian concept,[13] ”reinventing agrarian reform” (Minister Raúl Jungmann) along neo-liberal principles, using market-analogous processes and incentives rather than expropriation as instruments of land transfer. Thus it is undertaking steps towards Lipton's (1993) ”new wave” land reform. The ”Cédula da Terra” programme will offer landless families (partly World Bank financed) credits to purchase land in direct negotiations with landowners. Previously registered landless families, organised in communities and supported by unions or other rural organisations, will be provided with a sort of ”credit card” issued by Incra, enabling them to choose their preferred land, negotiate the transfer and effectuate the deal with the state banks. This consensual modality is argued to be a decentralised, quicker, cheaper and more efficient way of transferring land to the poor. This is probably true, given the commonly long-drawn-out and conflictive bureaucratic and judiciary expropriation procedures that have often been subject to sabotage, patronage and corruption. If a previous pilot project is successful, this new model, it is hoped, could provide a million families with land in little more than six years. A greater role will be ascribed to municipalities in order to decentralise and accelerate the implementation of reform projects. A decentralised service of technical advice for reform projects is part of the new concept. Moreover, the government plans the titling of an estimated one million tenants (posseiros) or occupants to be localised and identified via satellite (sic). Unfortunately, the programme apparently does not include existing minifúndios willing to achieve viable farm sizes.[14]

A massive programme of credit-financed land purchases by the landless would raise two fundamental problems. Firstly, the programme could be self-defeating by driving land prices up again. A progressive land tax increasing the tax rate with rising farm size and decreasing degree of land utilisation could outweigh to some degree the price-effects of increased demand for land. Although Brazil has recently reformed its land tax legislation, some of the major causes of its notorious ineffectiveness have not been tackled. According to Hackbart (1997) the problems of tax evasion and ineffective collection remain largely unsolved due to the inexistence of an updated national land register and to a tax calculation based on land values and utilisation levels declared by the owner himself. Moreover, the previous time-progressiveness of the taxation of under-utilised land has been abolished. Although the effectiveness of the tax can possibly be improved by a new reform, the ”sticks” of conventional expropriation in conjunction with an additional land ceiling are likely to be required (Lipton, 1993: 650), if the objective is a significant change of the agrarian structure. The above discussion suggests a relatively generous limit of 2,000 hectares which would leave the bulk of the dynamic export-orientated agroindustrial sector untouched. According to the 1980 census[15] this would affect merely 20,696 properties, owning together, however, more than 127.5 million hectares or about three and a half times the size of Germany. This ceiling alone would, if effectively enforced, increase the land supply by some 86.1 million hectares (albeit not all of these lands may be suitable for land reform purposes). What will be decisive, however, is not so much the use of classical or ”new-wave” instruments of agrarian reform (or a combination of both), but the political will behind them.

The second problem associated with credit-based land purchase programmes is, of course, the question of repayment. Obviously, this point is closely related to the evolution of land prices. Binswanger and Elgin argue correctly that the land price must not contain any premium reflecting non-agricultural income, such as real future appreciation, inflation premium, income tax exemptions and credit cost advantages of large farms, otherwise the assetless poor would not be able to finance land purchases out of agricultural profits (1990: 349f). Agricultural policy will have to take this into account, if the new agrarian concept is to create viable family farms rather than offering a temporary palliative. In this context it is worth remembering that poor land purchasers will have to finance not only the acquisition of land but also initial investments, from housing, through basic infrastructure to working capital. For these investments the special credit line for agrarian reform (Procera) offers subsidised credits for resettled farmers. Given this double burden to be financed by loans, it is worth considering a subsidy element for land acquisition as well, even though the World Bank (1990: 66) finds that the poor are willing to pay the market interest rate. The question is rather whether under the current policy environment of high interest rates and falling agricultural prices they are able to do so. Moreover, the Bank's argument that subsidised credits do not reach the poor is a gross generalisation. Procera may suffer from all sorts of problems, but it clearly has reached a significant part of the resettled farmers. Although FAO/Incra (1995) propose a reform of Procera they do not question its being subsidised.

Family agriculture has become an important issue of policy and debate in Brazil. A series of studies have analysed its characteristics, delimitation, sub-groupings, economic weight, productivity, and differentiated policy requirements vis-à-vis its diverse sub-sectors (e.g. ”peripheral”, ”transitory” and ”consolidated”). FAO/Incra (1995) have worked out comprehensive ”guidelines for agrarian policy and sustainable development”. The proposals for strengthening family farms start commonly with stressing the need for a favourable macro environment, eliminating the remaining pro-latifúndio distortions, such as explicit or implicit subsidies to machines, price policy etc., as well as abolishing counterproductive tenancy and labour legislation. Eliminating distortions, although an indispensable condition, is by no means sufficient to create a ”level playing field” for family farms. Centuries of latifundism and large-farm bias have heavily disadvantaged the rural poor and smallholders in terms of their ”initial endowments” not only of wealth but also of human and social capital. Therefore, those studies rightly list up a range of crucial policy requirements belonging to the ”standard fare of small farmer development programmes” (Binswanger et. al., 1990:352): the redirection of public investment in infrastructure (water supply, irrigation, roads, electrification, storage facilities etc.) towards small farmers; improved credit supply (via official financial institutions or alternative credit schemes such as credit associations, group lending and revolving funds); broadly-based access to decent health services, basic education, professional training and technical advice; reorienting agricultural research and extension towards adapting and disseminating new technologies to small farmers; appropriate policies to encourage smallholder processing, marketing, transport services etc. Most of these objectives are on the agenda of the recently set up ”National Programme for Strengthening Family Farming” (Pronaf). Recognising the absurdity to resettle landless farmers, on the one hand, while on the other small farmers, ignored by public policies, are on the verge of losing their land, the programme's goal is to support and consolidate these family farms. It aims at benefiting 3,000 municipalities, creating 4.4 million jobs, and investing some 12 billion dollars until 1998. The programme is likely to suffer, however, from tensions arising from an unchanged bimodal strategy with contradicting policies vis-à-vis the capitalist and the family farm sectors. One of Pronaf's stated action fields is ”the negotiation of public policies with sectoral organisations” to promote the adjustment of these policies to the needs of family farmers. Thus, the adaptation of the policy framework for family farming is left to the dynamics of Brazilian corporativism which is likely to favour latifúndio interests that traditionally have dominated the Ministry of Agriculture (ex-Maara, now MAA).

Apart from the contradictions of a two-tier strategy, the question is not so much what needs to be done to promote forms of rural development that help to consolidate family farms and can, as Lipton stresses, also sometimes re-establish the IR where it has collapsed due to constraints imposed by the ”Junker path” (1993: 653). The problem is rather how to ensure that the target groups indeed benefit from such programmes. Rural Brazil has a long tradition of diverting pro-poor programmes to the advantage of the rich, which reflects its pro-rich biased institutional matrix of formal and informal ”rules of the game”. Public service delivery in the countryside is commonly plagued by poor performance, political clientelism, corruption and collusion between influential landowners, state officials and local politicians. In the terms of Paul (1991), public provision is largely characterised by its ”capture” through the local elites. In addition to the monopoly of public services and the widespread incapacity of the poor to ”demand” their delivery, this is likely to constrain public sector performance and, as a result, the effectiveness of programmes such as Pronaf. Therefore, rural development demands taking into account not only its technical requirements and political implications on the macro level, but also its public management dimension within a given social structure on the micro level.

Tendler (1991) has tried to identify the reasons for the relatively better public performance in 23 World Bank financed rural development projects considered successful in North-eastern Brazil. She found the following determinants: (1) the ease and difficulty of the task, (2) the existence of outside pressure, (3) built-in incentives to perform, (4) the involvement of interested actors and organisations at the local level. Some of Tendler's general conclusions and recommendations are worth being considered in programmes such as Pronaf: a single agency should be given sole responsibility for a project, including the power to carry out tasks itself or to contract them out; research and extension should concentrate on highly focused interventions with a clear beginning and end; results must be measured in terms of output rather than input; agencies should be subjected to ”demand shocks” by channelling funds partly through ”users”[16]. Moreover, project periods should coincide with the four-years electoral mandates etc. (1991: xxiv). The general lesson from Tendler's evaluation is that if RD programmes supporting family farming are to be effective, they have to design institutional arrangements that elicit good performance through appropriate incentives and sanctions (e.g. by linking payment to work effort measured by results). Agencies must be submitted to mechanisms of accountability by the means of competition (exit) and democratic checks and balances (voice). Paul emphasises that, in the final analysis, elite ”capture” can be countered only by the ”voice” of the rest of the public (1991: 46).

Agrarian reform itself is likely to play an important role in increasing the potential of ”voice”, by reducing income barriers and, as we have argued in section 1, institutional barriers to its use. The experience in many resettlements has been that they tend to weaken the traditional vertical/hierarchical social structure, shifting gradually to horizontal forms of organisation. Therefore, democratic accountability, agrarian reform and rural development are likely to be mutually supportive and interdependent. Together they could indeed create the foundations for a viable small/medium farm strategy in which the traditional landed elite would either become capitalist medium farmers or make use of alternative opportunities as input suppliers and in processing and marketing (under competitive conditions). An equitable and efficient agrarian structure requires both technological and social modernisation. This demands, in turn, a determined policy response in the sense of an ”inclusive modernisation” strategy as outlined above.

The debate on agrarian reform has always had a double focus: both economic efficiency and social equity have provided important arguments and justifications for structural change by the means of state intervention. Technological modernisation and capitalist transformation in the countryside have not changed the agrarian structure in a way as to remove its traditional character of being a source of poverty and inequality, clientelistic domination, social tensions, violence and political instability. Thus the traditional social and political case for land reform has remained valid. In addition, the crisis of the urban-industrial sector has sharply increased the pressure and the case for an agrarian reform whose primary role is no longer seen in paving the way for capitalism nor in its functionality vis-à-vis industrialisation but in a sense in ”picking up the pieces” (in social terms and to some degree) of a failed development strategy that has resulted in the marginalisation and social exclusion of large parts of the population both in rural and urban areas. The new ”vocation” of agrarian reform is perceived to be the reduction of migration, re-absorption of labour, creation of employment, income, and citizenship for the rural poor, thereby enhancing social equity and promoting democratisation in the countryside.

With regard to economic efficiency we can conclude that the IR still provides a strong argument for agrarian reform. Even in the dynamic context of technological change it has not disappeared in Brazil. Although the “conservative” modernisation of agriculture has created a dynamic agroindustrial sector that repeatedly has registered ”bumper crops”, overall productivity in agriculture is still low and huge amounts of land remain idle or under-utilised. What surprises in the context of a massively subsidised modernisation process is not so much the dynamism of modern commercial farms but the remarkable resilience of small farms and their still important contribution to agricultural production. Yet, it is the dynamism of commercial large farms that has led authors like Graziano da Silva to conclude that agrarian reform is no longer an economic necessity but a mere possibility. Since the largely urbanised Brazil must avoid serious disruptions in the supply of wage goods, it is certainly desirable to preserve this dynamism, as long as it is based on efficiency rooted in genuine scale advantages rather than on rent-seeking. Brazil is privileged in the sense that it need not sacrifice its dynamic modern sector for agrarian reform. It can rely on the idle and under-utilised land of the traditional latifúndio sector as well as on the huge land reserves held (equally under-utilised) by modern commercial farms.

It would be a fatal mistake, however, to equate dynamic growth with large farms, continuing to pursue a capital-intensive large farm strategy to agricultural growth. Serious social and political consequences would probably constrain this growth in the longer term. Rather than ”betting on the strong”, agricultural policy should aim at dynamising the small and medium farm sector. Given their productivity advantage arising from the IR, this would certainly not require two decades of negative interest rates and perverse state subsidies. It would demand, however, improved access to land, credit, modern technology, and technical advice as well as basic infrastructure and a redirection of research, marketing and price policies towards medium and small farms. Giving the latter the chance of catching up in the process of technological innovation could indeed deepen the IR and strengthen the case for an agrarian reform aimed at improving total factor productivity, i.e. the social efficiency of factor utilisation. Where appropriate, arrangements of activity splitting can combine the advantages of small and large farms. Finally, we can conclude that agrarian reform must be understood in terms of ”inclusive modernisation” rather than ”social compensation” if it is to make a significant contribution to overcoming the exclusive patterns of Brazil's abysmal inequality and inequity frequently labelled ”social apartheid”.

Abramovay, R & J. J. de Carvalho Filho (1993) ”Reforma Agrária: o sentido econômico de uma política distributivista” in Reforma Agrária, Abra, vol. 23, may/august 1993, pp. 31-44.

Abramovay, R. ”Agricultura Familiar e Capitalismo no Campo”, in Stédile, J.P. (ed.) (1994), A Questão Agrária Hoje, Editora da Universidade /UFRGS, Porto Alegre.

Abramovay, R. (1994), ”Comercialização e Clientelismo” in Romeiro et.al. (eds.) Reforma Agrária: Produção, Emprego e Renda. O Relatório da FAO em Debate, Vozes/IBASE/FAO, Rio de Janeiro, 1994, pp. 136-142.

Barraclough, S. ed. (1973) ”Agrarian Structure in Latin America: A Resume of the CIDA Land Tenure Studies of Argentina, Brazil, Chile, Colombia, Ecuador, Guatemala, Peru”. Lexington, MA: Lexington Books.

Berry, R. A. & Cline, W. R. (1979) ”Agrarian Structure and productivity in developing countries: a study prepared for the ILO within the framework of the World Employment Programme”, The Johns Hopkins University Press, Baltimore and London.

Binswanger, H. P. & Elgin, M. ”Reflections on Land Reform and Farm Size”, in Eicher, C. & Staatz, J. (eds.) (1990), Agricultural Development in the Third World, The Johns Hopkins University Press, London, pp. 342-354.

Birdsall, N. and R. Sabot (1994) ”Inequality as a Constraint on Growth in Latin America” Development Policy (Sept. 1994) Washington, DC: Inter-American Development Bank (IDB).

Brasil. Ministério da Agricultura e do Abastecimento (1996), ”Programa Nacional de Fortalecimento da Agricultura Familiar”, Brasília, November 1996.

Brasil. Ministério das Relações Exteriores, (1994) ”Relatório Inicial Brasileiro Relativo ao Pacto Internacional dos Direitos Civis e Políticos de 1966”, FUNAG/IPRI; Brasília.

Bulmer-Thomas, V. (ed.) (1996) ”The New Economic Model in Latin America and its Impact on Income Distribution and Poverty”, Institute of Latin American Studies, Macmillan, London.

Cline, W. (1973) ”Interrelationships Between Agricultural Strategy and Rural Income Distribution”, Food Research Institute Studies 12(2).

Cline, W. R. (1970) ”Economic Consequences of a Land Reform in Brazil”, North-Holland Publishing Company, Amsterdam and London.

de Janvry, A. ”The Role of Land Reform in Economic Development: Policies and Politics”, in Eicher, C. & Staatz, J. (eds.) (1984), Agricultural Development in the Third World, The Johns Hopkins University Press, London, pp. 263-274.

de Janvry, A. & E. Sadoulet (1989) ”A Study in Resistance to Institutional Change: The Lost Game of Latin American Land Reform”, in World Development, vol 17, no. 9, pp. 1397-1408.

de Janvry, A. (1981) ”The Agrarian Question and Reformism in Latin America”, The Johns Hopkins University Press, Baltimore and London.

Dyer, G. (1997) ”Output Per Acre and Size of Holding: The Logic of Peasant Agriculture Under Semi-Feudalism” in The Journal of Peasant Studies, Vol. 24, no. 1/2, pp 103-131.

FAO, UNDP & Maara, (1992) ”Principais Indicadores Sócio-Econômicos dos Assentamentos de Reforma Agrária”, Brasília .

FAO/Incra, (1995) ”Diretrizes de Política Agrária e Desenvolvimento Sustentável”, Brasília.

FAO/Incra, (1996) ”Perfil da agricultura familiar no Brasil: dossiê estatístico”, Brasília

Ghose, A. K. (1983) ”Agrarian Reform in Contemporary Developing Countries”, Croom Helm, London and Sydney.

Gonzaga, G. M. (1996) ”The Effects of Openness on Industrial Employment in Brazil”, Ipea, Rio de Janeiro.

Graziano da Silva, J. (1994) ”O desenvolvimento do capitalismo no campo brasileiro e a reforma agrária” in J. P. Stédile (ed.), A Questão Agrária Hoje, Editora da Universidade /UFRGS, Porto Alegre.

Graziano Neto, F. (1991) ”A Tragédia da Terra: O Fracasso da Reforma Agrária no Brasil”, IGLU Editora Ltda., São Paulo.

Guanziroli, C. (1990) ”Agrarian Reform in the Context of Modernised Agriculture: the Case of Brazil”, University College London.

Guedes Pinto, L. C. (1995) ”Reflexões sobre a Política Agrária Brasileira no Período de 1964-1994”, in Reforma Agrária, ABRA, vol. 25, jan.-april 1995, pp. 65-92.

Hall, A. L. ”Land Tenure and Land Reform in Brazil” in Roy L. Prosterman et al. (eds) (1990), Agrarian Reform and Grassroots Development: Ten Case Studies, Boulder & London, pp. 205-232.

Hayami, Y., Quisumbing, M. A. & Adriano, L. (1990) ”Toward an Alternative Land Reform Paradigm. A Philippine Perspective”, Ateneo de Manila University Press.

IPEA (1994) ”O Brasil no Fim do Século: Desafios e Propostas para a Ação Governamental”, Rio de Janeiro.

Johnston, B. F. & P. Kilby, (1982) ”Unimodal and bimodal stratgies of agrarian change”, in J. Harriss, Rural Development, Routledge, London.

Kay, C. (1980) ”The Landlord Road and the Subordinate Peasant Road to Capitalism in Latin America”, in Etudes Rurales, vol. 77, jan.-mar. 1980, pp. 5-20.

Kutcher, G. & P. Scandizzo, (1981) ”The Agricultural Economy of Northeast Brazil”, published for the World Bank, The Johns Hopkins University Press, Baltimore and London.

Ladejinsky, W. (1977) ”Agrarian Reform as Unfinished Business: The Selected Papers of Wolf Ladejinsky”. Ed. Louis J. Walinsky. Published for the International Bank for Reconstruction and Development/ The World Bank, New York and London: Oxford University Press.

Lehmann, D. (1978) ”The Death of Land Reform: A Polemic”, in World Development, Vol. 6, No. 3, pp. 339-345.

Lipton, M. (1974) ”Towards a Theory of Land Reform” in D. Lehmann (ed.), Agrarian Reform and Agrarian Reformism, Cambridge: Cambridge University Press.

Lipton, M. (1993) ”Land Reform as Commenced Business: The Evidence Against Stopping” in World Development, Vol. No. 4, pp. 641-657.

Maddison, A. and associates (1992), ”The Political Economy of Poverty, Equity and Growth: Brazil and Mexico”, published for the World Bank, Oxford University Press

Medeiros, L. et. al. (eds.) (1994) ”Assentamentos Rurais: Uma Visão Multidisciplinar”, São Paulo, Editora Unesp.

Nolan, P. (1988) ”The Political Economy of Collective Farms: an Analysis of China's Post-Mao Rural Reforms”, Polity Press, Cambridge.

Paes de Barros, R., Mendonça, R. & S. Rocha, ”Brazil: Welfare, Inequality, Poverty, Social Indicators, and Social Programs in the 1980s” in N. Lustig (ed.) (1995) Coping with Austerity: Poverty and Inequality in Latin America, The Brookings Institution, Washington, D. C.

Paul, S. (1991) ”Accountability in Public Services: Exit, Voice, and Capture”, The World Bank, Working Papers WPS 614, Washington, D. C.

Ramos, J. (1996) ”Poverty and Inequality in Latin America: A Neostructural Perspective” in Journal of Interamerican Studies and World Affairs, Vol. 38, No. 2/3 Special Double Issue: Poverty and Inequality in Latin America, pp. 141-157.

Rao, H. (1977) ”Technological Change and Distribution of Gains in Indian Agriculture”, New Dehli: Macmillan.

Rocha, S. (1996) ”Renda e Pobreza: Os Impactos do Plano Real”, Ipea/Dipes, Rio de Janeiro.

Romeiro, A. (1994) ”Reforma Agrária e Distribuição de Renda” in J.P. Stédile et.al. (eds) A Questão Agrária Hoje, Editora da Universidade/UFRGS, Porto Alegre.

Romeiro, A. et.al. (eds.) (1994a) ”Reforma Agrária: Produção, Emprego e Renda. O Relatório da FAO em Debate, Vozes/Ibase/FAO, Rio de Janeiro.

Schmidt-Hebbel, K. & L. Servén (1996) ”Income Inequality and Aggregate Saving: The Cross-Country Evidence”, Policy Research Working Paper 1561, The World Bank, Washington, D. C.

Sorj, B. (1997) ”Agrarian Reform in Times of Globalisation”, paper presented at the Conference ”Power Structure, Interest Intermediation and Policy Making: Prospects for Reforming the State in Brazil”, Institute of Latin American Studies, London 13-14 Feb. 1997.

Stédile, J. P. (1993) ”De que Programa Agrário Precisamos?” in Reforma Agrária, Abra, vol 23, may/august 1993, pp. 17-25.

Sutton, A. (1994) ”Slavery in Brazil. A link in the chain of modernisation: the case of Amazonia”, Anti-Slavery International, London.

Swainson, N., (1986) ”Public Policy in the Development of Export Crops: Pineapples and Tea in Kenya” in IDS Bulletin, vol 17, no. 1, Institute of Development Studies, Sussex.

Tendler, J. (1993) ”New lessons from old projects: The dynamics of rural development in North-east Brazil”, Operations Evaluation Department mimeo, The World Bank, Washington, D. C.

Thiesenhusen, W. (ed.) (1989), ”Searching for Agrarian Reform in Latin America”, Unwin Hyman, Boston.

Thiesenhusen, W. C. & Melmed-Sanjak (1990) ”Brazil's Agrarian Structure: Changes from 1970 through 1980”, in World Development, vol. 18, No. 3, pp. 393-415.

Weyland, K. (1996), ”Democracy without Equity: Failures of Reform in Brazil”, University of Pittsburgh Press.

White, B. (1997) ”Agroindustry and Contract Farmers in Upland West Java” in Journal of Peasant Studies, Vol. 24 (3), Apr. 1997

World Bank (1993), ”East Asian Miracle: Economic Growth and Public Policy”, Oxford: Oxford University Press.

World Bank, (1990) ”World Development Report 1990”, Oxford University Press.

[1] Abbreviations: Abra - Associação Brasileira de Reforma Agrária; Cida - Comité Interamericano de Desarrollo Agrícola; CPT - Comissão Pastoral da Terra ; Eclac - Economic Commission for Latin America and the Caribbean (UNO); FAO - United Nations Food and Agriculture Organisation; FHC - Fernando Henrique Cardoso, President of Brazil; Funag - Fundação Alexandre de Gusmão; Ibase - Instituto Brasileiro de Análises Sociais e Econômicas; IBGE - Instituto Brasileiro de Geografia e Estatística; Ibope - Instituto Brasileiro de Opinião Pública e Estatistica; ILO - International Labour Organisation; Incra - Instituto Nacional de Colonização e Reforma Agrária; Ipea/Dipes - Instituto de Pesquisa Econômica Aplicada/Diretoria de Pesquisa; IR - Inverse Relationship between farm size and productivity; ISI - Import-Substituting Industrialisation; LDCs - Less Developed Countries; Maara - Ministério da Agricultura, do Abastecimento e da Reforma Agrária; MST - Movimento dos trabalhadores rurais Sem Terra; PNRA - Plano Nacional de Reforma Agrária (1985); Procera - Programa de Crédito Especial para a Reforma Agrária; Pronaf - Programa Nacional de Fortalecimento da Agricultura Familiar; PT - Partido dos Trabalhadores; RD - Rural Development; SNCR - Sistema Nacional de Crédito Rural; Sudene - Superintendência para o Desenvolvimento do Nordeste; UFRGS - Universidade Federal do Rio Grande do Sul; UNDP - United Nations Development Programme; Unesp - Universidade Estadual Paulista; UNO - United Nations Organisation.

[2] The amount of idle land is controversial. This figure is based on the IBGE agricultural census which includes only ”rural production units” (estabelecimentos rurais) implying some sort of economic activity. Thus IBGE data refer to the some 360 million hectares of farmland (1980) out of the 850 million hectares of total Brazilian territory. The Incra land register encompasses ”rural properties” irrespective of their utilisation. Incra data show that in 1992 60.7 percent of the land appropriate for agriculture, or 197.7 million hectares were held totally idle. As pointed out by Graziano Neto (1991), the Incra land register suffers from serious problems too and its data must be treated with caution.

[3] As quoted in Bulmer-Thomas (1996: 317).

[4] World Bank (1995), quoted by Ramos (1996: 143).

[5] Brasil. Ministério das Relações Exteriores (1994: 50).

[6] In Sutton, 1994: 11.

[7] Comissão Pastoral da Terra, (1997), Conflitos no Campo. Brasil 96, Goiânia-GO.

[8] Quoted by Berry and Cline, 1979: 227 (note 38).

[9] Thiesenhusen and Melmed-Sanjak., 1990: 405, table 14

[10] Thiesenhusen and Melmed-Sanjak, 1990: 403, Table 10

[11] As quoted by Thiesenhusen and Melmed-Sanjak, 1990: 394

[12] Thiesenhusen and Melmed-Sanjak, 1990: 405, Table 13

[13] The information about this new concept is taken from the unpublished paper ”A Reforma Agrária de Mercado do Governo FHC” from May 1997 by Rolf Hackbart, agricultural advisor of the PT Congress group. Official publications on this concept do not exist at present or are unknown to the author.

[14] Interesting and innovating proposals how to provide land for the upgrading of minifundios have been made in FAO/Incra (1995: 16ff) who suggest to provide them with preferencial purchasing options for farmland being sold.

[15] Thiesenhusen and Melmed-Sanjak, 1990: 397, Table 1.

[16] In this spirit, FAO/Incra (1995) propose, for instance, that groups of small farmers be funded directly to hire their own extension workers.