The Impact of Agricultural Policy Reforms on the Agricultural Sector in Brazil in the 1990s: Implications for Pro-Poor Agricultural Policies*

Steven M. Helfand University of California, Riverside

Paper prepared for discussion at the

OECD GLOBAL FORUM ON AGRICULTURE: DESIGNING AND IMPLEMENTING PRO-POOR AGRICULTURAL POLICIES

Paris, 10-11 December, 2003

* I would like to thank Brian Moore and André Hiroshi for valuable research assistance. Much of the material in section 2 of this paper was written with Gervásio Castro de Rezende in the context of a larger project on the Brazilian economy called NEMESIS, based at IPEA-RJ. I am thankful for a grant from the Brazilian government under a program called PRONEX that supported the NEMESIS project.

1. Introduction This paper seeks to contribute to the OECD Global Forum on Agriculture by examining the Brazilian experience of the past two decades in light of many of the questions that OECD has been studying as part of its two year project on the disaggregated impacts of agricultural policy reform and structural adjustment. These questions include: a) the transmission of agricultural prices from international to Brazilian markets, b) the differentiated impact of policy reforms on the Brazilian agricultural sector, and c) the impact of policy reforms on the rural poor. The paper identifies issues that require additional research and provides recommendations regarding policies that could benefit the rural poor. The paper is organized as follows. The second section of the paper analyzes the impact of policy reforms, and of changing macroeconomic conditions, on the Brazilian agricultural sector during the past two decades. The section stresses four issues: events outside of agriculture were central to the performance of the sector and to the timing and sequence of policy reform; reform involved far more than trade liberalization; the impact of reform on input markets and productivity was key for understanding the period; and policy reform had a highly differentiated impact on the sector. Although the benefits were distributed unequally across farm sizes, regions, and types of crops, agriculture became the most dynamic sector of the economy in the 1990s as a result of the reforms. The third section of the paper focuses on rural poverty in Brazil. It looks at changes in the level of rural poverty in the 1990s, and the heterogeneous geographical distribution of rural poverty in Brazil. The fourth section of the paper asks: what can be learned from the Agricultural Censuses about rural poverty in Brazil? It extends the analysis of poverty to an exploration of agricultural income, and analyzes the extent to which small farms have shared the benefits of growth in the 1990s. The final section of the paper provides conclusions and policy recommendations. 2. The Impact of Policy Reforms on the Agricultural Sector in Brazil in the 1990s1

The Brazilian economy began a process of restructuring in the 1990s as a result of dramatic changes in economic policy. The government abandoned policies associated with the import substitution industrialization (ISI) model and the country initiated a process of shaping a new path of development. The government liberalized trade, privatized state owned enterprises, deregulated domestic markets, and helped to create a South American Common Market (MERCOSUL). The agricultural sector was no exception. The country carried out a transition from an agricultural policy regime designed for a closed economy with substantial state intervention to a new regime tailored to an open economy and a curtailed role of the state.2 In this section of the paper we analyze the impact of the policy reforms, and of the changing macroeconomic conditions, on the agricultural sector in Brazil. We emphasize four

1 Section 2 of this paper draws heavily from Helfand and Rezende (forthcoming) which analyzed the period through 1998. This paper extends the analysis to include the period 1999-2001 which was heavily impacted by the floating of the currency in January 1999. 2 See Graham et al. (1987) for an analysis of Brazilian agriculture prior to the mid-1980s.

aspects of the reform period that were unexpected or did not receive sufficient attention from authors writing prior to the reforms. The first issue relates to the importance that events outside of the agricultural sector had not only for the performance of the sector but also for the timing and sequence of policy reform. In the ISI period it was clear that indirect policies such as an overvalued currency and industrial protection played a critical role in shaping the performance of the agricultural sector, and it was expected that the reform of these policies would have a positive impact on the sector.3 What was unexpected—and this is especially true for the case of Brazil—was the difficulty and length of time that would be necessary to stabilize the economy. The numerous stabilization plans that the Brazilian government adopted in the 1980s and 1990s joined the more traditional indirect policies as a key force that shaped the performance of the sector in this period. In this context, the policy reforms affecting the agricultural sector were driven, not primarily by agricultural policy objectives, but by the painful quest for price stability and the decision to abandon ISI policies. A second issue that we emphasize is that policy reform involved far more than trade liberalization. Deregulation and the reform of credit and support price policies were central as well. In fact, the most dramatic transformations in the agricultural sector took place for those products that were most heavily regulated, such as wheat, milk, sugarcane, and coffee. The products that lost import protection or gained a reduction in export taxation as a result of trade liberalization also were affected, but to a lesser degree. A third issue to which the pre-reform analyses did not give sufficient attention was the impact of policy reform on input markets and productivity.4 Liberalization altered relative input prices and increased access to high quality imported inputs. It also exposed domestic producers to greater competition. Both factors contributed to productivity gains and falling costs. The performance of some activities, such as animal production, was greatly aided by these changes. A fourth and final issue that we address is that policy reform had a highly differentiated impact on the sector. Reform was neither uniformly beneficial, nor entirely prejudicial. Thus, our analysis seeks to distinguish between different groups of products, such as importables and exportables, and different geographic regions, farm sizes, and sub-periods. Since not all reforms were introduced simultaneously, the 1990s should be treated as a decade of transition in which the old model was replaced, but not all of the features of the new model were firmly established. This section is organizes as follows. Section 2.1 provides an overview of the policy reforms, section 2.2 discusses the expected effects of the reforms, and section 3.3 analyzes the impact of the reforms on agricultural prices, output, trade, productivity, and input markets. 2.1 Overview of Policy reforms Related to Agriculture A. The Role of the Macroeconomic Environment

A first wave of policy reform began in the early 1980s in response to the debt crisis. Fiscal adjustment led to the reform of rural credit policy, reducing the volume of credit and

3 See Krueger, Schiff, and Valdés (1988), and their five volume study on this subject. 4 Quiroz and Opazo (2000) have recently addressed this issue.

2

eliminating the subsidies that had exceeded US$6 billion in 1979 (1996 dollars).5 At the same time, the government depreciated the currency and expanded the support price program. The government used a variety of additional policies in the 1980s to address the balance of payments problem and accelerating inflation. The sugar-alcohol program and wheat policy are two examples that will be discussed below.

The macroeconomic environment of the late 1980s and early 1990s also played a crucial

role in shaping the evolution of the reforms that affected the agricultural sector. The government liberalized trade and deregulated agricultural markets, in addition to making changes in the rural credit and support price policies, as part of an overall strategy to fight the threat of hyperinflation. Macroeconomic events also caused considerable instability for the agricultural sector. The numerous stabilization plans that were adopted in this period were almost always accompanied first by euphoria and then by a deep financial crisis for the sector. The instability was expressed through price cycles in agricultural asset markets—principally land and cattle—as well as in agricultural commodity markets since commodity stocks also served as real assets. Figure 1, which shows the real price of land, captures these cycles in 1986-87 (Cruzado Plan), 1989 (Summer Plan), 1990-91 (Collor Plans I and II), and 1993-95 (Real Plan). According to Goldin and Rezende (1990) and Rezende (1993), the launching of these plans increased uncertainty and reduced the attractiveness of financial assets. This, in turn, caused the prices of land, cattle and commodities to rise, which led to increased borrowing and investment in agriculture. As these plans failed, financial assets once again became more attractive. The consequence was an abrupt fall in the prices of agricultural assets and commodities and deep financial problems for the sector.

There are important similarities, and differences, between the Real Plan and the previous

failed attempts at macroeconomic stabilization. First, as with all of the stabilization plans in this period, the Real Plan generated an asset price cycle that led to increased indebtedness. When combined with the currency appreciation and high real interest rates that characterized the 1995-98 period, however, the result was one of the most severe financial crises that the sector experienced in the past two decades. Second, following the asset price cycle, land prices stabilized at a level that was about half of what they had been in the early 1980s (Figure 1). This facilitated access to land for competitive producers, lowered the cost of carrying out a state-led redistributive land reform program, and generated a healthy debate in Brazil about the causes of falling land prices.6 Finally, as we will show in detail below, the stabilization of the economy produced important gains for the agricultural sector and for consumers. B. Trade Liberalization

Trade liberalization for agriculture took place in the context of the economy wide reforms of the late 1980s and early 1990s. The sector benefited from a rapid fall in industrial protection and from the elimination of taxes and quantitative restrictions on agricultural exports. In the case

5 See Helfand (2001) for estimates of credit subsidies in the 1970s and 1980s. 6 Many authors have argued that land prices were lower in the late 1990s because land was no longer being held as a hedge against inflation. Rezende (2003), in contrast, argues that falling land prices were due to an increase in the supply of high quality land. Several of the most important contributions to this debate and other issues surrounding the Real Plan are Homem de Melo (1999), Dias and Amaral (1999), and Mendonça de Barros and Miranda (1998).

3

of coffee, for example, the export tax had been as high as 50% a decade earlier. Trade reform for primary exports advanced further in 1996 when the government removed the 13% value added tax in order to ease balance of payments pressures without a devaluation.

At the same time as the initial reforms for exportables took place, importables lost their

tariff and non-tariff protection. In cases like wheat, protection had contributed to raising producer prices to as much as double their international counterparts. For most importables, tariffs fell from the 35%-55% range prior to 1988 down to around 10% in 1991 (Helfand, 2000). Fernando Collor abruptly removed non-tariff barriers in 1990 when he became President. Shortly thereafter, Brazil signed the Treaty of Asunción that created the South American Common Market (MERCOSUL). MERCOSUL eliminated the tariffs on imports from Argentina and Uruguay, two countries with very competitive agricultural sectors.

C. The Reform of Agricultural Credit and Price Support Policies Reduction of the Role of the State

In contrast to trade liberalization, which happened quickly and broadly, the pace and coverage of reforming credit and price support policies was slow and varied. Due to its fiscal cost, the government eliminated the subsidies and substantially reduced the volume of real credit provided to the agricultural sector in the 1980s. Relative to the peak years of 1979 and 1980, annual lending was about 50% lower in 1988 and 1989. The government accelerated the credit contraction in 1990 when the flow of new credit fell by an additional 43%.7 This reduction was not solely an expression of a policy decision to reduce government involvement in agricultural finance. It was also the result of an 18 month freeze on financial assets that was one component of a macroeconomic stabilization plan aimed at combating inflation. With the exception of 1994—the first year of the Real Plan—the average annual volume of real agricultural credit in 1990-98 remained at about half of its 1988-89 level.

As the government scaled down credit policy in the 1980s, it simultaneously expanded

the price support program. The purpose of the program was not only to stabilize commodity prices. Perhaps more important, the goals of the program were to guarantee an adequate domestic supply of food, to save foreign exchange, and to contribute to controlling inflation. As an indication of the level of activity of this policy in the second half of the 1980s, in 1987 the government purchased 27% of the rice and 24% of the corn harvest. In addition, it provided storage credit to cover 85% of the cotton, 30% of the rice, and 25% of the soybeans harvested in that year (Goldin and Rezende, 1993). After several years of inactivity, the government once again acquired a large volume of agricultural stocks in the period 1992-95. This time, however, it happened at the same time as the private sector was carrying out imports. These events made it clear that policies had to change.8 From this point on, the government began to develop price support instruments that were consistent with an open economy and that involved a much lower fiscal cost. By 1999, government stocks of rice, beans, corn, wheat and soybeans had each been reduced to less than 2% of annual consumption (Villa Verde, 2001).

7 The credit data come from the Central Bank Recor database and refer to crops and animals. They are deflated with the IGP-DI inflation index of the Getulio Vargas Foundation. 8 See Rezende (2003) for a more in depth analysis of price support policy in the 1990s.

4

As a result of the changes in credit, price support, and other agricultural policies, the government dramatically reduced the level and share of fiscal expenditures related to agriculture. In the 1980s, for example, 5.65% of all federal expenditures targeted the agricultural sector. In the period 1995-99, this share was only 2.11% (Gasques, 2001). Redefinition of the Role of the State and Increased Role of the Market At the same time as the government reduced the volume of credit, it also significantly altered the sources of finance. In the 1980s there were two main sources: the Treasury and compulsory lending by private banks. Private banks had the option of lending a specified fraction (often around 25%) of their demand deposits to agricultural producers at interest rates set by the government, or depositing those funds in an account that earned zero interest at the Central Bank. In 1985, for example, 64% of agricultural credit came from the Treasury and 32% from required lending. By the late 1990s, this picture was altered dramatically. Compulsory lending became the most important source of agricultural credit after the economy stabilized under the Real Plan. In the late 1990s, it accounted for around 40% of rural credit. Treasury funds, in contrast, fell to under 2%. In their place, the government used Constitutional Funds that had been created by the 1988 Constitution to support regional development efforts, a Workers’ Support Fund which was financed with a tax on businesses, and other Funds such as for coffee and commodities. These Funds were financed by earmarked taxes and, depending on the Fund, the money could be lent at below market rates of interest. There were two important drawbacks of the new model. First, the public sector continued to provide most rural credit, or to heavily regulate it as in the case of required lending. Second, because most rural finance did not pass through the Congressional budget process, there was a substantial lack of transparency. Thus, this is one area where additional reforms were still necessary. In addition to creating new public sector sources of credit, the government and the private sector sought to develop new sources of private finance.9 We offer three examples. First, the Brazilian government adopted measures intended to increase the flow of foreign capital into the agricultural sector. In 1995 the government eliminated the Financial Operations Tax--a tax used to control short run flows of foreign capital--only for funds that were destined for agriculture. A second government innovation was the creation of the Rural Product Note (Cédula do Produto Rural, or CPR). CPRs permitted agricultural producers to acquire liquidity at the time of planting through advanced sales of their products. CPRs reduced this risk of these operations because they were legal documents that, if not honored by either party, could be enforced through the courts, and because they were guaranteed with insurance that was provided by a bank. Finally, most authors agree that there were many private sector innovations for providing agricultural producers with credit from processors, input suppliers, and traders. However, because these mechanisms did not require reporting to the Central Bank, they are much more difficult to quantify. In general, they were limited to the most modern segments of the agricultural sector and to the producers of highly traded goods.10

9 See Faveret (2002) and Gasques and Conceição (2001) on credit in the 1990s. 10 Although the relative importance of the private sector increased as a source of lending, the performance of commercial agriculture still depended on its relationship with the official credit system because of a substantial amount of debt held by farmers. Since default prevented access to new credit, in 1995 the government initiated a process of debt renegotiation. Due to the power of the lobbies that represented large producers, the renegotiation resulted in large implicit subsidies for the debtors. Thus, the government signaled that it was still willing to bail out and subsidize large indebted farmers. This signal ran counter to the spirit of the reforms.

5

Among the many changes in the price support policy, there were two key innovations that are worthy of mention. Both new instruments sought to guarantee a predetermined minimum price at a lower cost than the traditional policy and without leading to the formation of public stocks of agricultural goods. Both instruments also made use of auctions to lower costs and increase transparency. First, the government created an options contract that gave farmers the option to sell a given quantity of a product to the government at a predetermined price and date. The cost of the contract was determined in an auction. The owner of the contract could chose to exercise it when market prices fell below the price specified in the contract. When this happened, the government could choose to purchase the good or to pay the difference between the market price and the price specified in the contract. The other important innovation was the creation of a Marketing Premium (PEP). When market prices fell below support prices, rather than purchase the product directly or finance storage, the government agreed to make a payment to commodity purchasers in exchange for a commitment to purchase the targeted crop from farmers at the support price. The purpose of the auction was to determine the minimum payment necessary to induce the private sector to purchase the good. In addition to reducing costs, both the options contract and the PEP sought to increase private storage and, consequently, the incentives for investment in storage facilities.

D. The Agrarian Reform Program

The government rapidly expanded the agrarian reform program in the second half of the 1990s and targeted an increasing share of rural credit to small farmers. Under the program the government redistributed more land since 1994 than in the 1964-94 period. Between 1995 and 1999, over 370 thousand families are reported to have received land. The decline in land prices to their lowest level in several decades (see Figure 1) reduced the cost of the program. Since the land reform program occurred so late in our period, it is too soon to evaluate its success and it will not be a major focus of this paper. However, an analysis of how the existing small farms have fared in the environment of the 1990s is extremely relevant for the new small farms that have been created through the land reform program. E. The Deregulation of the Domestic Markets of Sugarcane, Coffee, and Wheat In addition to price support policy in the 1970s and 1980s, which was aimed primarily at corn, rice, beans, soybeans, cotton and cassava, the government utilized elaborate systems of regulation for sugarcane and its derivatives (sugar and alcohol), wheat, coffee, and milk. The specific policies that existed in the 1970s and 1980s are described in more detail in Helfand Rezende (forthcoming). The key point here is that the government deregulated the markets for these goods in the 1990s. In the case of wheat, for example, subsidies began to be phased out in 1987, and the market was deregulated in 1990. Sugarcane subsidies were also largely withdrawn in the 1980s, sugar exports were liberalized in the mid-1990s, and the sugarcane and alcohol markets were both deregulated in the late 1990s.11 In the case of coffee, falling prices in the second half of the 1980s led to the gradual withdrawal of the export tax. The Collor Administration abolished the Brazilian Coffee Institute (IBC) in 1990 as part of its reforms aimed at liberalizing trade and deregulating markets. 11 See Lopes and Lopes (1998) for detailed studies of sugarcane, coffee, and wheat in the 1990s.

6

2.2 The Expected Impact of Policy Reform The traditional view that agriculture was uniformly taxed as a result of ISI led some observers to expect that the sector as a whole should benefit from the move to a less interventionist and more outward oriented model (World Bank, 1986, Chpt. 4). A more nuanced analysis of the impact of ISI on the agricultural sector emerged in the 1980s, and was crystallized in the 18 country World Bank study headed by Krueger, Schiff, and Valdés. The results of this project highlighted the importance of policies outside of the agricultural sector, such as overvalued currencies and industrial protection, and demonstrated that these indirect policies implicitly taxed much of agriculture more than sector specific policies may have benefited it. Within the agricultural sector, direct policies tended to tax exportables, protect importables, and partially compensate for indirect discrimination through input subsidies. The Brazilian experience was broadly consistent with the international pattern, with the notable exception that its subsidized credit program was far larger than any of the other countries in the sample (Brandão and Carvalho, 1991). In this context, we ask: what would be the expected effect on the agricultural sector of the policy reforms that were discussed in the previous section? The expected impact of trade policy reform would be highly differentiated. All of agriculture would benefit from a reduction in industrial protection, which would raise the relative price of agricultural goods. Exportables would also benefit from reduced export taxation and an elimination of restrictions on trade. Importables, in contrast, would be harmed by the reduction in tariff and non-tariff barriers, and would be forced to compete more directly with imported goods. If these markets also had been heavily regulated, deregulation would increase competition and lead to falling product prices. Thus, in terms of output prices, we would expect the impact to be positive for exportables, negative for importables, and mixed for non-tradables. The evolution of the real exchange rate would also be an important determinant of relative prices. The real exchange rate appreciated in the late 1980s, and then appreciated again with the adoption of the Real in mid-1994. The situation finally became unsustainable in January of 1999 when the currency was allowed to float and depreciated by 50%. The failure to depreciate the real exchange rate in the 1990s should not be interpreted as an incomplete reform of the previous model. It resulted from the particular macroeconomic circumstances of the period.12 Thus, even though most of the anti-agriculture bias in policy was removed, the bias against tradables remained. The negative impact on importables should have been even more severe and the benefits to exportables more modest. There are strong grounds to believe that policy reform would lead to improved resource allocation and increased productivity and efficiency. Within the sector, we would expect a change in the product mix as area shares come to more accurately reflect each crop’s comparative advantage. Thus, importables would experience a decrease in their share of area.

12 Real exchange rate appreciation was not uncommon in Latin America in the early 1990s. Quiroz and Opazo (2000) suggest that optimistic expectations related to the policy reforms in the region could partially explain the large capital inflows of the 1990s and the resulting currency appreciations. While this seems plausible for many of the countries in the region, the 1994 real exchange rate appreciation in Brazil was much more an outcome of the adoption of a new currency and the stabilization of the economy in that year.

7

This would likely lead to a rise in average productivity as the least competitive producers were driven out of their respective sectors. Exposure to import competition would pressure the remaining producers to increase efficiency and lower costs, which could lead to positive dynamic effects on investment, growth, and productivity. Since most exportables were already highly competitive in international terms, the short run effects on productivity would not be as strong. The elimination of industrial protection should also lead to forces that contribute to increased productivity. Since the industries that produced inputs for agriculture had been protected, a fall in the prices of inputs such as tractors, irrigation equipment, and fertilizer, should accompany the policy reform. This would contribute to lower unit costs and increased productivity. As Quiroz and Opazo (2000) argue, however, these changes are unlikely to be neutral across farm sizes. A fall in the price of capital and intermediate inputs relative to the price of agricultural labor would create incentives for the adoption of labor saving techniques. This process is likely to favor large farms and could lead to increased out-migration. The impact of policy reform on the size distribution of farms is more complex, however, and it is not obvious that small farms should have been harmed the most. Some factors should have hurt the large more than the small, while others should have mitigated the impact on the small. The withdrawal of the credit, sugarcane, and wheat subsidies, for example, should have harmed large farms more, as there is ample evidence that they benefited from these policies disproportionately. Similarly, many small farms are not fully integrated into the market, or are net buyers of the goods that they produce, with the implication that falling product prices might not harm them, and could even provide a benefit. A final factor that we have not discussed thus far was the extension of social security benefits to rural areas in the 1990s. This should have had a positive impact on the welfare of the rural poor, and it is likely to have stemmed out-migration. Due to these offsetting forces, we believe that the impact of policy reform on the number and size distribution of farms is a question that can only be answered empirically. The reform of rural credit policy should have led to multiple results. First, with heavily subsidized credit in the 1970s and early 1980s, a considerable portion of the highly fungible funds were diverted to non-agricultural uses or employed in low priority investments. An improvement in resource allocation should have occurred as a result of the subsidy reduction. Second, the experience of credit amnesties in the late 1980s and debt refinancing in the 1990s contributed to producers--especially the large ones--forming the expectation that a significant portion of the costs of default would ultimately be absorbed by the government. Thus, the move toward a more private system of credit that increased the costs of default should have resulted in efficiency gains. In contrast to the observations above, the reduction in the volume of credit available to the sector in the 1990s and the difficulty of small farms in accessing private credit are likely to have created obstacles to investment and growth. The reform of support price policy should also have had important consequences. First, since this policy served to expand production on marginal lands, its phasing out should have led to improved resource allocation within the sector. Second, since the government was not a very discriminating purchaser of agricultural goods, nor a very able manager of commodity stocks, an increased role of the private sector in marketing and storage should have led to improvements in product quality and in the country’s warehousing sector. Finally, the reduction in the

8

government’s role in marketing and storage has stimulated the development of private futures and insurance markets that could help to facilitate storage and manage risk. However, these markets are still incipient, and small producers are likely to face substantial obstacles of access. 2.3. Agricultural Performance A. Agricultural and Food Prices In this section we analyze the evolution of agricultural prices at the farm-gate. The analysis permits us to quantify the impact on prices of policy changes. We emphasize five key points: a) the products that had been heavily regulated were affected most by policy reform, b) unfavorable international price movements more than offset the positive impact of policy reform in the early 1990s on the relative prices of most exportables, c) most real domestic prices fell dramatically from the mid-1980s through the late 1990s, d) the real exchange rate was the principal force causing prices to fall prior to the 1999 devaluation, e) a substantial fall in international prices in the 1999-01 period largely offset the beneficial effects of the devaluation on domestic prices.

We use a methodology based on the law of one price to quantify the impact of policy changes on domestic agricultural prices.13 In order to do this, we decompose the percentage change in a good’s real domestic price into three components: the percentage change in the real international price, the percentage change in the real exchange rate, and a residual that captures the percentage change in policy and other factors. We begin by writing the domestic price of a tradable good as:

)T1)(1(EPP ititt*itit +θ+= (1)

where and are respectively the nominal domestic price, and the nominal international price measured in foreign currency, of good i in time t. is the nominal exchange rate in period t. is a markup factor that includes the transactions costs and a competitive profit margin that are necessary to make the domestic price comparable with the international price.

is the residual proportional difference between the two prices after the markup has been considered, and can be thought of as the tariff equivalent, or nominal rate of protection (NRP). NRPs capture the effects of trade taxes, non-tariff barriers, alternative market structures, and other policies that drive a wedge between domestic and border prices. Non-competitive or state regulated markets, for example, can have larger markups than a competitive market.

itP *itP

tE

itθ

itT

If we divide both sides of (1) by a domestic inflation index ( ), multiply and divide the right hand side by an international inflation index ( ), and then take logs and first differences, we obtain:

dtINF

*tINF

)T1ln()1ln(RERlnplnpln ititt*itit +∆+θ+∆+∆+∆=∆ (2)

where dt

itit INF

Pp = is the real domestic price, *t

*it*

it INFPp = is the real international price, and

dt

*tt

t INF)INF(E

RER = is the real exchange rate. Equation (2) provides a simple method for

13 As far as we know Quiroz and Valdés (1993) are the first ones to develop the approach used here as a simple tool for analyzing changes in policy. Jaramillo (2001) also uses this approach.

9

observing the extent to which real international prices and the real exchange rate do not explain ement in , and

the other factors as a residual in the empirical w rk that follows. Large changes in the residual can usually be relat

mn 1 of able

reverse

rices then fell by another 45% in the early 1

the mov real domestic prices. Since itp RER*itp , t are easily observable, we treat

ed to large changes in policy or to a good that is not fully tradable.o

14

Table 1 decomposes the movements in the real domestic prices of five of the most important agricultural imports and four of the most important agricultural exports.15 The entries in the Table are the percentage changes in each variable between selected periods that correspond roughly to policies. Within each category, we calculate a simple average for those

roducts that exhibited similar behavior in terms of the impact of policy reform. ColupT 1 highlights the fact that the real domestic price of most agricultural products fell dramatically throughout the period. The product prices in Table 1 were all 50% to 60% lower in 1995-98 than in the period 1982-86, with the exception of coffee (-34%), cocoa (-70%), and wheat (-75%). The appreciation of the real exchange rate (column 3) was the principal factor leading to falling prices. It accumulated a 57% drop during the same period.16

The final column of Table 1 shows that the effect of policy on the real domestic price of

most importables—as measured by the residual—was quite limited. On average, there was only an 8% fall in the late 1980s in the real domestic prices of beans, corn, cotton, and rice beyond what could be explained by the percentage changes in the real international prices of these goods (-4%) and by the real exchange rate (-21%). The negative effect on prices was then fully

d in the 1990s, with positive residuals equal to 4% in the early 1990s and 8% in the late 1990s. Given that tariffs on these products fell from the 35% to 55% range prior to 1988 down to 10% in 1991 (and later 0% for imports from MERCOSUL), it is clear that these products did not experience a dramatic reduction in prices as a result of trade liberalization. Rather, by the late 1980s the prices of these goods were already close to their import parity equivalents. The most significant impact on this group came from the elimination of non-tariff barriers which increased the competition with imports. Real prices were falling, but this was attributable to the real exchange rate appreciation.

The effect of policy reform on wheat, which involved far more than trade liberalization,

was dramatic. Column 4 of Table 1 shows that the removal of the wheat subsidy in the late 1980s led to a 26% fall in the domestic price beyond what could be explained by the international price (1%) and the real exchange rate (-21%). The combined effect was to generate a 46% drop in the domestic price of wheat. Real wheat p

990s as a result of low international prices (-16%), the appreciation of the real exchange rate (-24%), and other factors (-9%). We show the consequences of such a substantial fall in domestic prices on production and trade in the following section.

14 It is important to point out that equation (2) is an accounting identity. The international price need not be exogenous, and we identify cases when this is unlikely to be true. 15 The domestic prices come from the Getulio Vargas Foundation (FGV). The international prices were taken from the commodity markets in Chicago (wheat, corn, and soybeans) and New York (cocoa, coffee, and cotton), and from the FAO (beans and rice). The Brazilian inflation index is the IGP-DI from the FGV, the foreign inflation index is the U.S. producer price index from the U.S. Bureau of Labor Statistics, and the exchange rate is the commercial sale price of a dollar in terms of domestic currency. Consumer prices are from the national CPI of IBGE. 16 Prior to this period of appreciation, there was a 30% real devaluation in 1983.

10

Consistent with our expectations, the effect of policy reform on exportables was positive.

The producer prices of cocoa, oranges, and soybeans all benefited between 10% and 20% from a combination of eliminating export taxes, quantitative restrictions, and the value added tax on exports in 1996 (Column 4, Table 1). All three products, however, suffered from international prices that were 28% lower on average in 1990-94 than in 1987-89. This negative shock more than offset the gains from reform in the early 1990s.

eement in 1989 and the abolition of the Brazilian Coffee Institute (IBC) in 1990. Without the domestic and international institutions that had con

y about 60%. The conclusion that we draw is that consumers derived significant benefits from falling food pr

s of real exchange rate appreciation or depreciation. Figure 2a, for example, shows that the real domest

As we described in Section 2, coffee was one of the most heavily regulated products in

the sector. As in the case of wheat, policy reforms and other factors had a pronounced effect on prices. Even though Brazil reduced the quantity of exports in the late 1980s, international coffee prices fell by 20% due to a significant expansion of exports from other countries. This was followed by the end of the International Coffee Agr

tributed to regulating supply, Brazil and other coffee exporters increased their exports by about 10% each in the first half of the 1990s. The result was an additional 32% fall in international prices (Table 1, Column 2). The situation began to improve in 1993 when Brazil and other exporters formed the Association of Coffee Producing Countries and undertook a coordinated effort to implement a voluntary system of export targets. This, in combination with several frosts in Brazil and the creation of the Deliberative Council for Coffee Policy in Brazil in 1996, contributed to bringing coffee prices back—at least temporarily—to roughly the same level as in the 1987-89 period.17 An important consequence of export targets, however, was that Brazil continued the long term trend of losing its share of the international coffee market. In the 1960s it had accounted for one third of world exports. In the 1990s its share fell to 19%.

Table 1 also shows the evolution of real food prices. Food prices fell by 31% between

1982-86 and 1995-98, which is slightly more than half of the average decline in agricultural prices. This was somewhat less than the other tradable components of the Consumer Price Index--household goods (-50%) and clothing (-60%)--but considerably more than the non-tradable components--housing, healthcare, and personal expenditures--which all rose b

ices in this period, and that falling agricultural prices contributed to make this possible. Figures 2-7 show the evolution of domestic and international agricultural prices with a

long run perspective. The upper graph of each Figure shows annual real domestic and international price indices between 1981 and 2001. The only difference in the lower graph is that the real international price was measured in local currency after having been multiplied by the real exchange rate. Thus, the difference between the two graphs isolates the effect of period

ic price of soybeans fell by substantially more than the real international price in the second half of the 1980s. The Figure also shows the depressing effect on domestic prices of the real exchange rate appreciation in the period 1994-98. Both of these facts were quantified in Table 1. Figure 2b, in contrast, shows that when one measures the international price in

17 Due, in part, to the difficulties of cooperation among coffee exporters, the recuperation in prices was short lived. International prices in 1999-00 were back to their 1990-94 levels. See the special issues of Agroanalysis on coffee published in November 1997 and November1998.

11

domestic currency--thus capturing the joint effect of movements in the international price and the exchange rate--the evolution of the real domestic and international prices are almost identical in the long run. This might not be so surprising in the case of soybeans, as this has probably been the exportable crop in Brazil that has always had the lowest degree of government intervention and the highest degree of transparency and integration between international and domestic markets. The small divergence between the two indices that is evident as of 1996 in Figure 2b is likely due to the elimination of the ICMS tax on exports which allowed the domestic price to rise in relation to the international price.

Figures 3, 4, and 5 show an extremely similar story for corn, rice and wheat. The upper

graph of each Figure shows that domestic prices fell by much more than international prices in the second half of the 1980s, and that the real exchange rate appreciation between 1994 and 1998 depressed domestic prices. The lower portion of the graphs for corn and rice show that although annual price changes in the two series can be somewhat large in one year or smaller in another, the long run trends have been remarkably similar. In the case of wheat, the lower graph shows the rem

fell by 29% and 33% respectively.

80s, and both the industrial and services sectors in the 1990s. Between 1980 and 1998, real GDP grew by about 40%, while real agricultural output rose by about 70%. This is a

l evidence that has documented how the share of griculture in GDP tends to decline during the process of economic development. The reversal is

ct of the policies that the Brazilian government adopted in this period.

oval of the wheat subsidy which caused domestic prices to fall by much more than international prices in the second half of the 1980s. It also shows an extremely close relationship between the two indices since 1989. Figures 6 and 7 tell largely the same story for sugar and coffee. Figure 6 shows the gradual withdrawal of the sugarcane subsidy throughout the 1980s, and the positive impact of the ICMS tax withdrawal in 1996. The lower portion of both Figures 6 and 7 show a much higher degree of integration between domestic and international markets in the 1990s than in the 1980s. This is especially true in the case of coffee.

A final observation is that the effect of the exchange rate depreciation of January 1999,

which measured 45% in real terms between 1995-98 and 1999-01, is largely imperceptible in Figures 2-7. There is no sharp increase in domestic prices in this period. The reason is that a decline in international prices largely offset the real depreciation in this period. The average international prices of the importables and exportables shown in Table 1

B. Output and Trade of Agricultural Products Aggregate Agricultural Output

In contrast to the 1950-80 period, the agricultural sector outperformed the industrial sector in the 19

remarkable fact in light of the vast internationaatestimony to the powerful impa

18 One of the most striking features of this period was the dynamism of the animal sub-

sector since the mid-1980s. Growth was most rapid for poultry production, which achieved rapid

18 On the positive performance of agriculture in Brazil during the 1980s, see Goldin and Rezende (1990). de Janvry and Sadoulet (1993) show that adjustment policies favored agriculture throughout Latin America during this period.

12

gains in productivity (Helfand and Rezende, 1999). The appreciation of the currency did not create serious obstacles for chicken exports because it simultaneously helped to lower the costs of feed (primarily corn and soybeans) and of the imported genetic material used for breeding. Althou

eeze on importables. As can be seen in Table 2, the total value of agricultural imports tripled between

approximately two billion to over six billion 1998 U.S. dollars

• The impact of policy reform on cotton production and trade was also dramatic. Imports surpass

nd price. Since the devalua

% in the 1990s, but output grew due to rising yields.

d two thirds of the third harvests. • As a share of production, rice imports grew from 6% in the late 1980s to around 15% in the late 1990s. Relative to the other regions in Brazil, two states in the South (RS and SC) produced a higher quality rice, used a different technology (irrigation), and obtained yields that

gh poultry production grew the fastest, rising by 182% between 1980 and 1996, cattle and hog slaughter also grew rapidly, rising by 98% and 70% respectively. Hog production, with about a decade lag, has followed the same path of modernization as poultry production, and there is evidence that cattle production is now entering a period of rapid intensification as well.19

Importables: Trade and Output Consistent with our expectations, policy reform led to a substantial decline in the harvested area of the domestically produced importables and to a dramatic increase in spending on imports. The appreciation of the currency between 1994 and 1998 contributed to the squ

1985-89 and 1995-98, rising fromper year. Table 3 shows that harvested area for wheat, cotton, corn, rice, and beans fell by 20% in this same period. Since the 1999 devaluation, imports dropped by 40%, or almost back to the level of the first half of the 1990s, and production grew due to rising yields. In what follows, we briefly highlight some of the most important changes that affected each crop.

• The withdrawal of the wheat subsidy and the deregulation of the wheat market led to a decline in production of more than 50% between 1985-89 and 1990-94 (Table 3). By the late 1990s imports had nearly tripled (Table 2). Since wheat was produced almost exclusively in the South of Brazil, the burden of adjustment fell most severely on this region.

ed exports in the late 1980s and grew to an average of 700 million dollars per year in 1995-98 (Table 2). Simultaneously, cotton production declined by 50% (Table 3). Area and production doubled in the Center-West, however, as new varieties contributed to the ability of this region to successfully compete with imports in terms of quality a

tion, production has nearly recovered its pre-trade liberalization level, and the value of imports fell to below the level of the early 1990s.

• In contrast to wheat and cotton, corn imports as a share of domestic production remained constant at about 3% in the 1990s. Total area harvested in corn was stable in the 1990s, with increased output coming almost exclusively from rising yields. As with cotton, the Center-West was the most dynamic region of the country. • Area harvested for beans fell by about 20

Imports remained a modest share of domestic production. Bean producers were highly differentiated, with an increasing share of production coming from irrigated second (23%) and third (7%) harvests. Although farms over 200 hectares only produced 15% of the first harvest in 1995-96, they were responsible for a third of the second an

were triple the national average. These states suffered considerable pressure from imports, but

19The data come from Pesquisa Mensal de Abate de Animais (IBGE). We restricted the period to 1980-96 because a change in the methodology of the survey made the data after 1996 less comparable. On modernization of the cattle sector, see Agroanalysis, June, 2000.

13

m d to hold their ground. Area and output fell substantially in the rest of the country, but has recovered in the Center-West, in part due to new varieties of rice which have boosted yields . Exportables: Trade and Output

Policy reform had a positive impact on exportables prior to the 1999 devaluation, bu

anage

t it

wo observations are worthy of mention. First, a reduction of the barriers imposed by developed

most important agricultural exports would benefit Brazilian exports

--actually performed the worst in the 1990s. In the case of cocoa, even though area harvested remained unaltered, output and yields fell by more than 30% as a result o

oybean output grew more than any of the other crops in the 1990s, with virtually all of

Sugarcane and oranges both expanded their area, output, and to a lesser extent, yields in

was not of the same magnitude as the impact on importables. Table 2 shows that the total value of agricultural exports rose by about four billion 1998 dollars between 1985-89 and 1995-98, or the same amount as the value of agricultural imports. For imports, however, this represented a 300% increase, while for exports it was only 34%. Since the devaluation, the total value of exports actually fell. Since the quantity of exports (in metric tons) rose for all exportables in Table 2 other than cocoa, it is clear that the fall in the value of exports was due to falling prices. Tcountries on many of Brazil’s

both in terms of export quantities and price. Second, additional research is called for on the degree to which the growth in Brazilian exports of many commodities has contributed to depressing world prices.

Policy reform also led to substitution of importables in production within the agricultural

sector. Table 3 shows that the area harvested for our sample of importables fell by six million hectares between 1985-89 and 1995-98. The harvested area for the exportables, in contrast, rose by 1.6 million hectares. A considerable amount of area was freed for more productive activities --such as animal production--or was no longer being used because it was not profitable to do so.

Two of the products which might have benefited the most from the elimination of export taxation--cocoa and coffee

f the “witches broom” fungus which spread throughout the cocoa growing region of Bahia. The area harvested in coffee, in contrast, fell by 30% between 1985-89 and 1995-98. This was a reflection of the difficulties the sector had coping with falling prices between 1987 and 1993 and finding new institutional forms of organization and representation.

Sthe expansion coming from the Center-West (Table 3). As discussed later in the paper, falling input prices and rising productivity allowed soybean producers to deal with falling product prices rather easily. The value of exports was relatively constant between 1980 and 1994, yet favorable external prices in 1996-97, in addition to the incentives created by the elimination of the ICMS sales tax on exports contributed to raising the value of exports by about 50% in the 1995-98 period (Table 2). the 1990s. These crops were grown principally in the state of São Paulo, where 75% of the oranges and 50% of the sugarcane in Brazil were produced. Orange producers benefited from falling input prices, and considerable modernization of transport services and of the ports.20 After the government freed sugar exports from quantitative restrictions in the mid-1990s, exports responded rapidly and increased to nearly two billion dollars per year (Table 2).

20 On the transformation of the citrus sector, see Agroanalysis, May 1999 and June, 2001.

14

Table 2 also provides data on meat exports. Beef exports were stagnant in the past

enty years, yet finally took off in the 1999-01 period. Chicken and pork exports, in contrast,

. Productivity Gains and Inputs

x as well as a growth accounting approach. He also devises three different dexes to proxy for capital services. Notwithstanding the problems of measurement error that

ctivity, the results point to some important conclusions. First, ith six different specifications, TFP measured with the growth accounting approach increased

he period 1975-95. The first column of Table 4 shows the results for

lly in the Center-est where farms were much larger and where expansion occurred often by incorporating the

latest t

twhave grew rapidly since the late 1980s. The expansion of soybeans and corn in the Center-West was one of the forces leading to dramatic growth in the 1990s in the production of animals in this region (Helfand and Rezende, 1999). This is another dimension of the challenge to the competitiveness of the South in the 1990s, and of the expansion of lucrative activities in the Center-West. C Total Factor Productivity Several recent studies have measured total factor productivity (TFP) at an aggregate level in the Brazilian agricultural sector over the past several decades. The results have been qualitatively similar, and in what follows we focus on Barros (1999). In order to test the sensitivity of his results to alternative methods and specifications, Barros uses a non-parametric Tornquist indeinare common to all studies of produwbetween 16% and 36% in tone of the intermediate specifications. Although TFP rose by 20% between 1975 and 1995 under this scenario, all of the net growth came after 1986, and most of it was in the 1990s. The correspondence with the period of policy reform is significant. Second, as a result of the differential evolution of inputs, land and labor productivity grew by about 30% between 1985-86 and 1994-95, while TFP rose by only 15% in this period. This is an important result because it demonstrates that gains in land productivity (yield) overstate the gains in TFP.

Barros concludes that about two thirds of the growth in output between 1975 and 1995 was attributable to growth in inputs, and about one third to growth in TFP. It is important not to attribute the growth in TFP solely to technological change. As the least productive farmers chose to exit, as farmers withdrew the least productive land from production, and as they shifted crops to more productive regions, average productivity should have risen even without any technological change. In the context of increased competition in the 1990s, it is likely that efficiency gains within the farm were also an important reason for increased productivity. Technological change and scale effects should have contributed as well, especiaW

echnologies.21 Productivity gains were especially important for certain activities, such as milk, poultry, and hog production, where rapid modernization took place.

Land Productivity The final four columns of Table 3 show changes in yields (land productivity) for the principal importables and exportables in the 1980s and 1990s. Simple averages show that importables preformed better than exportables in both the 1980s and 1990s. Between 1980-84 21 Gasques and Conceição (2001) confirm that TFP grew faster in the Center-West than in any other region.

15

and 1985-89, yields rose by 16% on average for importables, while for exportables there was no gain. Between 1985-89 and 1999-01, the average gain in yields for importables was 53%, while for exportables (with cocoa excluded) it was only 26%. This is a striking result which reverses the pattern that had prevailed in the period 1950–80 when exportables almost uniformly

stic food crops (Graham et al., 1987). For the 1980s it reflects, in part, the turns

modest for oranges, sugarcane, and coffee, the tter in spite of a significant contraction in area. Soybeans was the one major exception, with

most of

Input Prices and Quantities

According to the Getulio Vargas Foundation’s input price index for agriculture, real input e periods 1987-89 and 1990-94, and then by another 8% between

1990-9

rvesting of a record grain crop in the 1994-95 agricultural year as well as in response to the increase in econom

employed should have risen, but strong pressures to shrink the rural labor force must surely have

outperformed domere on a series of successful investments that the Brazilian government had made in agricultural research. The result for the 1990s is consistent with the view that the contraction in area for importables led to a rise in average productivity as the least competitive producers were driven out of their respective sectors. Exposure to import competition also pressured the remaining producers to increase efficiency and lower costs. Since most exportables were already highly competitive in international terms, the effects of policy reform on productivity were not as strong.

With the exception of corn, the other four importables experienced substantial contractions in harvested area. Corn and cotton in the Center-West were the two main exceptions to the pattern of contraction, as they both increased area and yields in the 1990s. Increased productivity in this region was associated with technological improvements and scale effects.

As expected, the gains in yields were quite la

the growth coming from the Center-West. Rapid expansion of area in this region (38% between 1985-89 and 1995-98) was accompanied by rapid growth in yields (22%). Without a doubt, the expansion of cotton, corn, and soybeans in the Center-West, along with the associated animal based agroindustries, made this the most dynamic agricultural region of the country.

In section 2.2 of this paper we argued that policy reform was expected to lead to a significant realignment of relative input prices within the agricultural sector. This should have led to a greater reliance on techniques that used tradable inputs intensively and that economized on non-tradable inputs such as land and labor. Thus far we have discussed the contraction in the amount of land used by agriculture. We now present selected evidence on other inputs.

prices fell by 19% between th4 and 1995-98. Average input prices fell less quickly in the second half of the 1990s

because of wages. While the real prices of pesticides, fertilizers, services, fuels, and seeds fell by an average of 21% between 1990-94 and 1995-98, agricultural wages rose by 49% (see Column 5, Table 4). Monthly data show that the increase in wages occurred almost entirely between July, 1994 and April, 1995. Agricultural wages rose due to the planting and ha

ic activity that was associated with the early phase of the Real Plan. Agricultural wages were subsequently sustained at a higher level by an increase in the minimum wage from R$ 70 to R$ 100 in May of 1995. Living standards for those rural workers that succeeded in remaining

16

been felt. Column 2 of Table 4 captures this, as it shows a contraction in employment of more than one million starting in 1996. Table 4 also provides information about the impact of policy reform on fertilizer consumption and tractor use. Fertilizer consumption rose from one to four million tons per year between 1970 and 1980, in large part due to substantial subsidies. After contracting in the early 1980s, consumption stabilized around 3.5 million tons from 1984 to 1992 (Column 4). Trade liberalization and changing relative prices in the 1990s led to more than a 60% increase in fertilizer consumption between 1992 and 1998. This is one of the key factors that contributed to increasing TFP in the 1990s and it reflects the intensity of efforts to improve competitiveness in the 1990s. As the distribution of yields for corn and beans demonstrate, however, most of the

ains in productivity were concentrated in large farms.

tractors rose to nearly 1000 per year in 995-97 (Barros, 1999). Thus, even though it is likely that credit subsidies induced excessive

investm

me poverty ne (indigence) of R$65.07, equal to the estimated cost of a minimum food basket containing

the following discussion I focus on the rural population estimated to e living in extreme poverty.

g

In contrast to fertilizer consumption, Column 3 of Table 4 shows that the estimated stock of wheel tractors, which represents 75% of agricultural machines, peaked in 1994. Although real tractor prices were falling in the 1990s, tractor sales also depend on the availability of investment credit, which was severely curtailed in the 1990s. As a consequence, sales of domestically produced tractors fell from an annual average of 35 thousand in the 1980s to under 19 thousand for the years 1990-98. Only in 1994, as a result of the Real Plan, were sales comparable to the level of the 1980s. Imports of tractors have only partially compensated for falling sales of domestically produced tractors. The number of imported1

ent in tractors in the late 1970s and early 1980s, we reach the troublesome conclusion that high real interest rates and the lack of investment credit in the 1990s contributed to an absolute decline in the stock of tractors in the second half of the 1990s. This has been partially reversed in 2001 and 2002 as new lines of credit have permitted increased purchases. 3. Rural Poverty in Brazil 3.1 The Level of Rural Poverty in Brazil This section of the paper examines the level of poverty in rural Brazil. The data are drawn from Ferreira and Lanjouw (2001) who conducted one of the most careful recent studies of rural poverty in Brazil. The authors combine detailed information on expenditures from the 1996 living standards measurement survey (PPV) with the national household survey (PNAD). They use a poverty threshold of R$131.97 household income per capita, and an extreli2,288 calories per day.22 In b Ferreira and Lanjouw focus on the Northeast and Southeast of Brazil, the two regions where the PPV study was conducted. They estimate that 49% of the rural population in the Northeast of Brazil has incomes below the extreme poverty line, whereas only 25% of the rural

22 The authors employ small-to-large survey imputation techniques to take advantage of the strengths of the PPV (detailed information on consumption expenditures, and data that permit the construction of regional price indices) and the strengths of PNAD (a much larger sample that is representative at the state level rather than simply at the macro-regional level).

17

population in the Southeast falls below this poverty line. These poverty rates are considerably higher than in the urban areas of these two regions. 36% of the urban population in the

ortheast, and 8% in the Southeast, is estimated to live in extreme poverty.23 Based on the

l rates of extreme poverty in Brazil. The Figure shows that overty rates vary by more than a factor of six. São Paulo, the three states in the South, and

%. One state in the North and five states in the

ipalities with poverty rates between 56% and 82%. Another 1400 municipalities lie etween these extreme categories. While Ferreira and Lanjouw’s study suggests that rural

to household income growth and duction in inequality, his estimates suggest that growth in income accounted for around 80% of

lity accounting for the remaining 20%. Given the fact that agriculture was the most dynamic sector of the economy in the 1990s

Ndistribution of the population, the authors conclude that 22.69 million people, or about 20% of the population, lived in poverty in these two regions of Brazil in 1996, with approximately 75% of them residing in the Northeast. 3.2 Spatial Heterogeneity of Poverty in Brazil The recently released Atlas of Human Development in Brazil (2003) provides additional information about the spatial distribution of poverty in Brazil. Based on the Demographic Census of 2000, and using a different poverty line than the one used in the study by Ferreira and Lanjouw, Figure 8 shows state levepBrasilia have rates of extreme poverty below 8Northeast, in contrast, have between 30% and 41% of their populations living in extreme poverty. It is important to note that there is considerable heterogeneity in poverty rates even within individual states. Figure 9, for example, shows municipal level rates of extreme poverty in the Northeast of Brazil. The Figure shows that at one end of the spectrum there are 166 municipalities in this region with poverty rates under about 28%. At the other extreme, there are 225 municbpoverty might not be quite as variable across space as urban poverty, the information from the Atlas of Human of Development on municipal level poverty rates suggests that spatial variation in rural poverty is likely to be high when one examines data disaggregated to the municipal level. Pro-poor agricultural policies, consequently, cannot be designed in a one-size-fits-all manner. They must take account of this geographical heterogeneity.24 3.3 The Evolution of Rural Poverty in the 1990s According to Barros (forthcoming), who uses PNAD data and an extreme poverty line of 0.5 minimum wages in constant 11/1999 currency, or R$68 of monthly per capita household income, poverty in rural areas declined from 61.1% in 1992 to 54.5% in 1999. When he decomposes the decline in poverty into components attributablerethe change, with a decline in the degree of inequa

and that there was a substantial increase in agricultural wages in 1995, as documented in section 2 of this paper, it is not surprising to see that income growth contributed to reducing rural

23 The authors also present poverty estimates based on income reported in the PNAD survey, which is a common way of measuring poverty in Brazil. They show that this approach overestimates extreme poverty by roughly twenty percentage points in the Northeast, and ten percentage points in the Southeast. 24 In contrast to agricultural policies, pro-poor polices that focus on human capital can be more uniform across space because households are mobile and they can take the acquired human capital with them.

18

poverty in the 1990s. Out migration of the poor, however, is likely to have helped reduce rural poverty rates as well. According to the demographic censuses of 1991 and 2000, the total population in rural areas of Brazil declined by 11% in this period, or by about four million people. The rate of decline nearly doubled what was observed in the 1970s and 1980s. By 2000, only 19% of the population in Brazil still resided in rural areas. The decline in the number of people employed in the agricultural sector was of the same

agnit

. What can be Learned From the Agricultural Census About Rural Poverty in Brazil?

t based on an econometric model of over 9000 presentative farms in the region.

Before examining the Census data, several points should be clarified. First, based on the

Second, according to Ferreira and Lanjouw (2001), agricultural income in 1996

regard to gricultural employment. While non-agricultural sources of income have grown in importance,

m ude as the decline in the rural population in this period. According to the national household surveys (PNAD), total employment (including the self-employed and their unpaid family members) fell by 14% between 1992-93 and 2001-02. Figure 10 shows data from the Southeast of the country where, as a result of crops like coffee and cotton undergoing substantial down sizing in the state of São Paulo, the contraction reached nearly one quarter of the workforce. In the Northeast, the refuge of rural poverty in Brazil, Figure 11 shows that employment in the sector only fell by 6%. 4 In light of the household level information on rural poverty presented in section 3, this section of the paper uses data from the 1995-96 Agricultural Census to explore a) at what farm size can the income earned from agricultural production lift rural households out of extreme poverty, b) the extent to which small producers have benefited from the growth in the production of exportables in the 1990s, or have been squeezed by the contraction in the production of importables, and c) the determinants of income per family member engaged in agricultural production in the Brazilian Center-Wesre PNAD surveys, Graziano da Silva and Del Grossi (forthcoming) show that roughly two thirds of the employed people in rural areas of Brazil worked in agriculture in 1999.25 According to the Agricultural Census of 1995-96, which is quite consistent with PNAD 1996, 76% of the people employed in agriculture were classified either as the head of an establishment or as unpaid family labor. Thus, even though non-agricultural rural activities have grown in importance in Brazil, the importance of agricultural activities proper should not be underestimated. Most rural residents were still employed in agriculture, and most of these were self-employed. represented about 70% of total income for the poor in the Northeast of Brazil, and over 60% of total income for the poor in the relatively better off Southeast region. An important difference between the two regions, however, was the relative importance of agricultural income earned from production versus from labor. Agricultural labor income was only 16% of total income for the poor in the Northeast, whereas it was 39% in the Southeast. Of the remaining income of the poor in both regions, around 15% came from non-agricultural labor income and 15% from other sources, the most important of which were government transfers and pensions. Thus, the conclusion with regard to agricultural income is similar to the conclusion witha

25 The PNAD household surveys exclude the Northern region of the country.

19

and in some locations can be a driver of economic growth or a key component of a household’s strategy to escape poverty, it is still the case that most of the income of the rural poor is earned in agriculture (either as producers or as workers). Consequently, policies to assist the rural poor to increase their income earned from agricultural production are important. So too are policies aimed at increasing earnings for workers in agricultural labor markets. In this regard, additional research is required to explore the degree to which labor legislation and high taxes that employers pay when hiring permanent and temporary agricultural workers have encouraged excessive informality and mechanization (Carneiro, forthcoming). If so, labor market reforms could reduce the advantages of mechanization, increase off-farm employment opportunities, and thus contribute to the economic viability of small farms that rely on multiple income sources. 4.1 Agricultural Production Income and Rural Poverty Based on the 1995-96 Agricultural Census, I calculated monthly agricultural profit per employed family member by farm size. Profit is defined in a short run sense as the value of agricultural production net of variable costs.26 For Brazil as a whole, Figure 12 shows that on average establishments in the 5-10ha class earned R$65 per employed family member. Thus, if they had to rely only on income earned from agricultural production, the average agricultural establishment in the 5-10ha class would have an income equal to the extreme poverty line. 13% of the establishments in Brazil were of this size, and another 37% were smaller. Thus, agricultural production income alone would leave roughly one half of the establishments at or

elow the line of extreme poverty.

e is considerable spatial variation. In the Northeast, where overty rates were the highest, 88% of the establishments were smaller than 50ha in 1996, and

ate enough production income to surpass the higher overty line of $131.97. Thus, once again, we see that poverty is a much more extreme

b With reference to the poverty line of R$131.97, we observe that even the average establishment in the 20-50ha class was unable to lift itself above the poverty line. The average establishment in this size class earned R$120 per capita per month from agricultural production. 81% of the establishments in Brazil had 50ha or less in 1996. Thus, we can conclude that only about a fifth of the establishments in Brazil were able to generate enough income through agricultural production to escape poverty. Figures 13 through 16 show monthly profit per family member for four of the five macro regions in Brazil. As expected, therpeven the average establishment in the 20-50ha class did not earn enough income from agricultural production to escape extreme poverty. In contrast to the Northeast, in the other three regions (Southeast, South, and Center-West) there were generally fewer than 20% of the establishments in farm sizes with average production income at or below the extreme poverty line. In each of these other regions, we estimate that between one third and one half of the establishments had sufficient land to generpphenomenon in the Northeast than in the other regions of the country. 4.2 To What Degree Have Small Farmers Benefited from Agricultural Growth in 1990s? We answer this question in three ways in this section. First, we examine the evolution over time of the share of production that was produced by small farms in the dynamic

26 Rented land and hired labor were considered as costs, but no attempt was made to impute the opportunity cost of owned land or family labor.

20

exportables vs. the contracting or stagnant importables. Second, as an indicator of competitiveness, we examine changes in productivity by farm size over time. Finally, with reference to revenue earned from the ten principal crops, we examine the extent to which farms of different sizes succeeded in holding average revenue per farm constant in the 1990s by increasing output fast enough to offset falling prices. The conclusion from all three exercises is consistent with the view that small farms have been losing production share, especially in the

ost dynamic crops, that small farms have not increased productivity at the same rate as large

grew y 87% between the periods 1985-89 and 1999-01 (Table 3), with exports of soybeans, meal, and

d (Table 2). Although establishments under 20ha produced

nd the value of exports down.

mfarms, and that with few exceptions crop revenue per farm has only grown on large farms. Production Shares of Small Farms In absolute terms, the exportable crops that grew most rapidly in the 1990s were largely not produced by the smallest farms. Proportionately, it is also true that small farms produced a smaller share of these crops than they did of value of production in general, and there is some evidence that the share of production of small farms fell in the production of these crops in the 1990s. Table 5, for example, shows that the 50% of agricultural establishments in Brazil with 10ha of land or less produced 12% of the value of agricultural output in 1995-96, and the 64% of establishments with 20ha or less accounted for 21% of production. Let us compare this with the production of soybeans, the most dynamic crop in the agricultural sector. Soybean output boil also growing rapidly in this perio21% of output in the entire agricultural sector, Table 6 shows that they were responsible for only 6% of soybean production in 1995-96. And, although a comparison between the 1985 and 1995-96 censuses is somewhat compromised due to a change in the reference period of the latter census,27 it appears that the share of production coming from farms under 20ha fell by 4 percentage points in this period. In the Center-West region of the country, where soybean output grew most rapidly in the 1990s, farms in the 1,000-10,000ha class gained output share, while farms under 100ha only produced 2% of output in 1995-96. The conclusions are similar for sugarcane and oranges, two other exportables that grew rapidly in the 1990s. Establishments under 20ha only produced 3% of the sugarcane and 10% of the oranges in Brazil, and in both cases the share of production coming from small farms appears to have fallen. For coffee and cocoa, in contrast, small farmers produced 19% and 17% of the value of output, respectively. These shares were much closer to their share of producing value in the entire sector. These two crops, however, were in crisis in the 1990s. Cocoa production fell by 50% and exports by 85%, and coffee production was down 15% to 20% throughout most of the 1990s. Coffee production and the quantity of exports recovered in the period 1999-01, but this was likely a contributing factor to driving world prices a

27 On problems of comparability between the two censuses, see Helfand and Brunstein (2001). Due to the change in the reference period of the 1995-96 census, and the likelihood that not all establishments that existed in that year were surveyed, the use of the censuses for comparing the levels (or shares) of variables in 1985 and 1995/96 is problematic. Comparisons across census years of ratios, such as yields or output per establishment, which are variables that are presented in the following two sections, are unlikely to be influenced by the change in the reference period of the 1995/96 census. A significant bias would only occur if the farms that were not surveyed were systematically different than the farms that were included in the census. Because the conclusions from the data based on shares (Table 6) are entirely consistent with the conclusions based on yields (Table 7) and on output per establishment (Table 8), we have more confidence in the results and have chosen not to exclude the data in Table 6.

21

As discussed in section 2 of the paper, the importables generally experienced a contraction in area, substantial growth in yields, and mixed results in terms of production. Table 6 shows that small farms were responsible for a larger share of production of importables than of exportables. However, for these importables as a group, the share produced by farms under 20ha appears to have fallen from an average of 22% in 1985 to 16% in 1995-96. In the cases of beans, wheat, and rice in the South, small farms were better able to maintain their production shares during the initial phase of trade liberalization, although for wheat this was in the context of a generalized decline in production across all farm sizes. Farms in the 1,000-10,000ha class gained production share for all importables in Table 6. The Evolution of Yields by Farm Size In the context of falling output prices, productivity gains were likely to be an important means for maintaining profitability. Productivity gains, however, often required adopting technologies for which a minimum scale of operation was necessary. When this was the case, it presented obstacles to the viability of small farms.28 Table 7 provides evidence on the evolution of yields by farm size between 1985 and 1996 that is suggestive of problems of competitiveness for small farms and is consistent with the data presented in Table 6. Because yields are only a partial measure of productivity, the evidence is not conclusive. For beans and corn, which are integral to the production portfolios of millions of small farms, yields grew much more rapidly

r large farms. Average yields grew between 80% and 165% for bean producers over 100 s under 100 hectares. For corn, yields grew by more than

was not able to compensate for falling prices. ice in the South was the only general exception to this pattern. Thus, even after accounting for

the number of agricultural establishments in the 1990s, and the least efficient producers were the most likely to have exited the sector, it is still

fohectares, but by less than 30% for farm70% for farms over 100 hectares, yet by less than 30% for farms under 20 hectares. To varying degrees, most of the crops reveal relative gains for large producers.29 Rice production in the South, where small farms increased yields faster than large farms, is the one important counter example. This is likely the explanation why small farms were more successful at maintaining their production share in rice than in other crops. The Evolution of Average Output by Farm Size Section 2 of this paper showed that domestic agricultural prices of most crops fell 50% to 60% between the early 1980s and the late 1990s. In this section we explore the extent to which the average farm of different sizes was able to at least hold revenue constant by increasing output fast enough between 1985 and 1995-96 to compensate for falling prices. Average crop output for a given farm size could rise due to an increase in yields (Table 7), an increase in the average farm size within a given size class, or an increase in specialization (the share of land dedicated to the production of a specific crop). Table 8 shows the striking result that, with few exceptions, the average farm in all size classes below 100ha Rthe fact that there was a substantial decline in

28 This was not uncommon throughout Latin America. Neo-liberal reforms created obstacles to the competitiveness of small farmers in many countries in the region. See, for example, Carter and Barham (1996). 29 In the case of soybeans, where the 0-5 hectare group appears to have performed well, this group only accounted for a small percentage of domestic production. Soybeans produced on farms between 0-5ha, for example, accounted for less than 0.5% of soybean output.

22

the case that among the remaining producers in 1995-96, the revenue earned from crop production by the average farm in nearly all farm size classes under 100ha was smaller in real terms than in 1985. The farms that increased output per farm at a much faster rate than the price decline were generally larger than 1000ha. Rice in the South, and producers in the 100-1000ha size classes of beans, corn and cotton also belonged to this group. The growth observed in yields only provides a partial explanation for their success. Increased specialization was also an important part of the story. For example, the average producers of beans, corn and cotton in the 1000-10,000ha class all increased their land allocated to these crops by between 100% and 150%. 4.3 The Determinants of Income per Family Member in Agricultural Production Based on disaggregated data from the 1995-96 Agricultural Census, Helfand (2003) used Data Envelopment Analysis to estimate the technical efficiency of agricultural establishments in

e Brazilian Center-Western, and then used an econometric model to explain differences in

in the variation in is dependent variable along the lines of Helfand (2003).

There are three important conclusions that can be drawn from the regression results

thefficiency across municipalities, farm sizes, and types of land tenure with a host of policy relevant explanatory variables. The data set included over 9000 observations on “representative farms,” where a representative farm is defined as the average of all establishments of a particular farm size (15) and type of land tenure (4) in a given municipality (426). With this same data, I calculated restricted profit for this paper, defined as the value of output net of variable costs, per employed adult family member of the establishment. I then sought to explath Table 9 shows preliminary regression results from four models, where the dependent variable is the log of profit per employed adult family member. The first three regressions were estimated with county level fixed effects to take account of spatial heterogeneity, such as differences in soil quality or rainfall, that was not captured in the regressors. The fourth equation used a county level measure of the economic distance to the city of São Paulo, which can be viewed either as the industrial center of the country where important inputs are produced, as the center of effective demand in the country, or as a proxy for the distance to ports. This variable was constructed from information on distance and the quality of highways.30 All equations use White’s heteroskedasticity-consistent standard errors. presented in Table 9. First, all four regressions show that access to assets, such as land and tractors, are an important determinant of farm income per capita. Farm size (in logs) has a positive and non-linear relationship with income per capita, suggesting that as farm size grows income per capita rises more than proportionately. The log of the ownership of tractors--measured in horse power units--also has a positive and significant impact on income per capita. The log of the stock of animals--measured in cattle equivalent--in contrast, has a much smaller impact on income per capita than land or tractors, and the results appear less reliable because they change depending on the specification of the model.

30 I thank Newton de Castro, a professor at the UFRJ in Rio de Janeiro, Brazil, and a Co-PI on the research project NEMESIS for graciously providing me with this variable.

23

The second conclusion is that access to institutions, and public goods and services, are also an important determinant of income per capita. These variables are all defined as the proportion of farms within a given representative farm that had access to a specific item. Thus, regressions 3 and 4 show that income per capita rises as the proportion of farms with access to credit, electricity, technical assistance, and irrigation grows. Belonging to a cooperative did not have a statistically significant impact on income per capita. The third conclusion is that the further the economic distance to São Paulo, the lower the agricultural income per capita. Thus, policies such as infrastructure creation that lower

ansportation costs could increase income per capita in agriculture.

occupants. In the same way that elfand (2003) found that renters were more technically efficient than owners, we find here that

part-time farmers. Finally, lthough cross-sectional spatial variation in prices is less than ideal, prices were added to

, through infrastructure creation for example, can also aise per capita income in agriculture.

s contributed to making agriculture the most dynamic

tr There are a number of other interesting results in Table 9. There is a strong negative relationship between the number of adult equivalent family members employed in establishments and income per capita, which suggests that there is a substantial amount of underemployed family labor in agriculture in the Center-West. The type of land tenure matters, although land ownership does not dominate renters, sharecroppers, and Hrenters earn more income per capita in the Brazilian Center-West. The hypothesis offered in Helfand (2003) was that renters were a more uniform, profit oriented group, whereas owners were a more heterogeneous group, including subsistence farmers and aregressions 3 and 4 as controls. The price of cattle had a large positive impact on profit per capita, and the price of land had a somewhat smaller positive impact. Although higher land prices increase costs, they are also correlated with proximity to urban areas and higher value agricultural activities such as horticulture. The results from this exercise suggest that a) access to assets--especially land--matters for raising income per capita in agriculture. Thus, land reform which transfers assets to the poor has the potential for helping the poor out of poverty. Interestingly, access to land through rental contracts appeared to be no worse, and even slightly better, than ownership in terms of raising income per capita. Thus, improving the functioning of rental markets could also be an important part of the policy package. b) Land alone is not enough. Complementary agricultural policies that permit access to credit, technical assistance, electricity, and irrigation are also extremely important mechanisms for raising income per capita in agriculture. c) Finally, increasing market integration by reducing transportation costsr 5. Conclusions and Policy Recommendations The conclusions of this paper can be separated into three parts: 1) the impact of policy reforms on Brazilian agriculture, 2) trade liberalization and international price transmission, and 3) policy recommendations for reducing rural poverty. 1) The Impact of Policy Reforms on Brazilian Agriculture One of the most striking conclusions of the paper is that, in spite of real domestic prices falling since the mid-1980s, the reform

24

sector of the economy in the 1990s. Within agriculture, there were winners and losers. The sproportionately from the reforms, as did the

enter-West region of the country. Import competing products and the South suffered the most om th

s recuperated.

ector had little to do with the type of

and to lower the price of ood for consumers. Lower food priced contributed to reducing urban poverty in the 1990s.

lt of ese forces, there was a significant reduction in agricultural employment in the 1990s,

exportable and animal sub-sectors benefited diCfr e initial shock of liberalization in the first half of the 1990s and the period of currency overvaluation between 1994-98. Since the devaluation of 1999, and after considerable structural adjustment, even the production of many importables ha Some of the most profound transitions within the sgood or the region where it was produced. They came as a result of a redefinition of the role of the state. Thus, in the cases of wheat, coffee, sugarcane, and milk, the transition resulted not only from a change in the level of protection, but also from the withdrawal of the state from its traditional role of setting prices, managing production, and regulating the activities of marketing and trade. Taken as a whole, these policy changes led to more competition within the agricultural and processing sectors, and to a larger role for the market in coordinating the relationship between them. Among the most important benefits of these reforms was to force improvements in resource allocation, productivity, and product quality, f Policy reform also had a differentiated impact across farm sizes. The available evidence suggests that small farms experienced considerable problems of competitiveness. They did not increase productivity at the same rate as large farms. They lost production share, especially in the most dynamic crops. And, in terms of revenue, they did not increase output fast enough since the mid-1980s to compensate for falling real prices. As a consequence, with the exception of irrigated rice in the South, the revenue derived from the production of ten of the most important crops in the sector fell in real terms between the mid-1980s and mid-1990s for the average farm in all farm size classes under 100ha. The farm sizes that increased average output at a much faster rate than the price decline were generally larger than 1000ha. As a resuthespecially in the South and Southeast of the country. It was in this context that the government expanded the agrarian reform program in the second half of the 1990s and targeted the provision of official credit to small farms. Additional research is required to investigate the complementary policies that are necessary, in addition to land, to contribute to the viability of the land reform beneficiaries in the new policy environment.

2) Trade Liberalization and International Price Transmission There is considerable descriptive evidence that movements in the real exchange rate and international prices are transmitted to domestic agricultural markets, and sometimes vice versa, in the medium to long run in Brazil.31 Especially since the beginning of the 1990s, it appears that Brazilian agricultural markets have been well integrated with international agricultural markets in the sense that the long run trends in prices are broadly similar. Thus, although this paper does not examine the extent to which prices can deviate in the short run from their long run trends, it does suggest that agricultural prices have not been very distorted in Brazil since the early 1990s.

31 A comprehensive econometric analysis of this question would provide a valuable contribution to this debate.

25

Since the late 1980s, Brazil pursued substantial unilateral trade liberalization and dramatically reduced the role of the state in regulating agricultural markets. Today, Brazil has a very open agricultural sector with exports and imports facing few restrictions and little or no taxation. Thus, the potential gains from additional trade liberalization on the part of Brazil are quite small. The agricultural sector in Brazil, however, could achieve substantial gains through liberalization of international agricultural markets (Brandão and Lima 2002; Cypriano and Teixeira, 2003). International liberalization would have a positive impact on the prices of Brazil’s exportables and would facilitate growth in Brazilian agricultural exports. Recent events

Cancun and Miami suggest that progress on this front is likely to be slow.

re relevant for the esign of pro-poor policies:

regions.

gion of the country, and this region is ome t

rking family members above the extreme poverty line. There are me examples of small farms being incorporated into the production of irrigated crops, but it is

i. Agricultural income, including income earned from production and in the agricultural labor

cers or as workers). onsequently, policies to assist the rural poor to increase their income earned from agricultural

and temporary agricultural workers have encouraged excessive informality and mechanization

in Because many of the gains from unilateral trade reform have already been captured, and gains from multilateral trade reform depend on forces outside of Brazil’s control, the search for pro-poor policies in rural Brazil must focus on domestic policies that can be designed and implemented by the Brazilian government. 3) Policy Recommendations for Reducing Rural Poverty Based on an analysis of trends in rural poverty, sources of income for the rural poor, and agricultural census data, the paper reaches a number of conclusions that ad i. Rural poverty is extremely heterogeneous across space. Pro-poor agricultural policies, consequently, cannot be designed in a one-size-fits-all manner. They must be designed with substantial spatial variation across and within ii. Extreme poverty rates are highest in the Northeast reh o more than half of the population below the extreme poverty line. In this region, 88% of the farms had less than 50ha in 1996, and on average farms with less than 50ha did not earn enough profit to lift the wosonot clear to what extent this can be generalized. Clearly, there is an urgent need for more research on strategies to alleviate rural poverty in the Northeast of Brazil. iimarket, still represents about 70% of total income for the poor in the Northeast of Brazil, and over 60% of total income for the poor in the relatively better off Southeast region. Of the remaining income of the poor in both regions, around 15% comes from non-agricultural labor income and 15% from other sources, the most important of which are government transfers and pensions. Thus, while non-agricultural sources of income have grown in importance and are often a key component of a household’s strategy to escape poverty, it is still the case that most of the income of the rural poor is earned in agriculture (either as produCproduction are important (see below). So too are policies aimed at increasing earnings for workers in agricultural labor markets. In this regard, additional research is required to explore the degree to which labor legislation and high taxes that employers pay when hiring permanent

(Carneiro, forthcoming). If so, labor market reforms could reduce the advantages of

26

mechanization, increase off-farm employment opportunities, and thus contribute to the economic viability of small farms that rely on multiple income sources. iv. This paper provides preliminary results from a regression model that seeks to explain agricultural income per working family member based on a large dataset from the Brazilian Center-West. The results suggest that: a) Assets, such as land and tractors, are an important determinant of per capita family income. Asset transfers, such as through the current land reform program in Brazil, thus provide the potential of lifting beneficiary families out of poverty. b) Access to land via other mechanisms, such as rental contracts, could potentially contribute to reducing poverty as well. Controlling for other factors, the renters of land had slightly higher income per capita than the owners. While this result could reflect the unique haracteristics of owners and renters in the Brazilian Center-West, it does suggest that improving

er rural development policies--such as credit, technical sistan

e just as important.

. There is abundant evidence that targeted income transfers, such as through the rural social

al security benefits to rural reas in the 1990s all contributed to reducing poverty. On the other hand, the rural population

mendation, improved education and other policies to increase the uman capital of future migrants should be a high policy priority.

cthe functioning of rental markets could also be an important part of the policy package. c) Access to agricultural and othas ce, and electricity--are also important determinants of income. Thus, access to assets alone may not be sufficient for raising the income of the poor. The institutional, policy, and broader economic context in which the assets are used can b d) A reduction in the economic distance between markets in Brazil brought about through improved infrastructure could raise income per capita in the agricultural sector. This follows from the result that distance from São Paulo, the economic center of the country in terms of industrial production and final demand, reduced income per capita. vsecurity and bolsa alimentação programs in Brazil, have been quite successful at reducing poverty and malnutrition. Thus, although this paper focused on agricultural policies, there is no doubt that income transfer policies targeted at the poor constitute an important part of the anti-rural poverty policy mix. vi. The percentage of the rural population in poverty and the total number of poor people in rural areas both fell in the 1990s. Additional research is required to investigate the relative importance of alternative causes of this decline. On the one hand, rapid growth in the sector, a large increase in agricultural wages in 1995, and a considerable expansion of sociaand employment in agriculture both fell by more than 10% in the 1990s as a result of the problems of competitiveness experienced by small farms and the adoption of labor saving technologies by large farms. Migration of poor people from rural to urban areas will undoubtedly continue in the foreseeable future. Thus, although this is a rural rather than an agricultural policy recomh

27

6. References Barros, Alexandre L. Mendonça de, “Capital, Produtividade e Crescimento da Agricultura: O

Brasil de 1970 a 1995,” Ph.D. diss., University of São Paulo, ESALQ, Piracicaba, 1999. Barros, Ricardo Paes de, “Evolucao da Pobreza e do Trabalho Rural no Brasil dos Anos 90,” in

Blanco, Mauricio F. (editor), Estrutura Agrária, Mercado de Trabalho e Pobreza Rural no Brasil, IPEA, forthcoming.

Bonelli, Regis, and Renato Fonseca, “Ganhos de Produtividade e de Eficiência: Novos Resultados para a Economia Brasileira,” TD No. 557, Rio de Janeiro: IPEA, 1998.

Brandão, Antonio Salazar P., and José L. Carvalho, "Brazil," Chapter 3 in The Political Economy of Agricultural Pricing Policy, Volume 1: Latin America, edited by Krueger, Anne O., M. Schiff, and A. Valdés, Baltimore: The Johns Hopkins University Press, 1991.

Brandão, Antônio Salazar P., and Elcyon C. R. Lima, “Impacts of the U.S. Subsidy to Soybeans on World prices, Production and Exports,” mimeo, August 2002.

Carneiro, Francisco G., “An Assessment of Rural Labor Markets in Brazil,” in Blanco, Mauricio F. (editor), Estrutura Agrária, Mercado de Trabalho e Pobreza Rural no Brasil, IPEA, forthcoming.

Carter, Michael R., and Bradford L. Barham, “Level Playing Fields and Laissez Faire: Postliberal Development Strategy in Inegalitarian Agrarian Economies,” World Development, Vol. 24, No. 7, 1996, pp. 1133-49.

Cypriano, Luiz Alberto, and Erly Cardoso Teixeira, “Impactos da ALCA e do Mercoeuro no Agronegócio do Mercosul, Revista de Economia e Sociologia Rural, forthcoming.

de Janvry, Alain, and Elisabeth Sadoulet, “Adjustment Policies, Agriculture and Rural Development in Latin America,” in Economic Crisis and Third World Agriculture, edited by Singh, Ajit, and Hamid Tabatabai, Cambridge: Cambridge University Press, 1993, pp.118-146.

Dias, G. L. S. and C. M. Amaral. “Mudanças Estruturais na Agricultura Brasileira: 1980-1998,” in Bauman, R. (org.), Brasil: Uma Década em Transição, Rio de Janeiro: Editora Campus, 1999a.

-----, “A Evolução do Preço da Terra no Brasil: 1966-1998,” mimeo, São Paulo, FIPE-USP, 1999b.

Faveret Filho, Paulo “Evolução do Crédito Rural e Tributação Sobre Alimentos na Década de 1990: Implicações Sobre a Cadeias de Aves, Suínos e Leite,” BNDES Setorial, Rio de Janeiro, No. 16, Sept. 2002, pp. 31-56.

Ferreira, Francisco H.G., and Peter Lanjouw, “Rural Nonfarm Activities and Poverty in the Brazilian Northeast, World Development Vol. 29, No.3, pp. 509-28, 2001.

Ferreira, Francisco H.G., Peter Lanjouw, and Marcelo Neri, “A New Poverty Profile for Brazil Using PPV, PNAD, and Census Data,” Texto para Discussão No. 418, Economics Department, PUC-RIO, March 2000.

Ferreira Filho, Joaquim Bento de Souza, “Os Desafios da Estabilização Econômica para a Agricultura Brasileira,” in Gomes, M.F.M. and F. A. da Costa (eds.), (Des)Equilibrio Economico and Agronegócio. Viçosa: Universidade Federal de Viçosa, 1999.

28

Gasques, J.G., “Gastos Públicos na Agricultura,” in Transformações da Agricultura e Políticas Públicas, Gasques, J.C. e J.C.P.R. Conceição (eds.), Brasília: IPEA, 2001.

,

-----, s, Gasques, J.C. and J.C.P.R. Conceição (eds.), Brasília:

Goldin rom

Graham Economic

Grazian oyment and Incomes in

-----, ditor), Estrutura Agrária, Mercado de Trabalho e Pobreza Rural no Brasil,

Helfan, edited by Helfand, Steven M.,

-----, roups

-----,

Helfand, Steven M., and Luis P. Brunstein, “The Changing Structure of the Brazilian

Helfan Castro de Rezende, “The Impact of Sector-Specific and 8,”

porary Economic Policy, 37 msc. pages, forthcoming. O

Homem Brasileira: Perspectivas,” Revista de

IBGE, e

-46. Krueger, Anne O., Maurice Schiff, and Alberto Valdés, "Agricultural Incentives in Developing

Lopes, 26-30.

Gasques, J.G. and J.C.P.R. Conceição, “Transformações Estruturais da Agricultura e Produtividade Total dos Fatores,” in Transformações da Agricultura e Políticas PúblicasGasques, J.C. and J.C.P.R. Conceição (eds.), Brasília: IPEA, 2001. “Financiamento da Agricultura—Experiêcias e Propostas,” in Transformações da Agricultura e Políticas PúblicaIPEA, 2001. , Ian, and Gervásio Castro de Rezende, Agriculture and Economic Crisis: Lessons FBrazil, Paris: OECD, 1990. , D.H., H. Gauthier, and J.R. Mendonça de Barros, “Thirty Years of Agricultural Growth in Brazil: Crop Performance, Regional Profile, and Recent Policy Review,”Development and Cultural Change, Vol. 36, No. 1, 1987, pp. 1-34. o da Silva, José, and Mauro E. Del Grossi, “Rural Nonfarm EmplBrazil: Patterns and Evolution,” World Development 29(3), March 2001, pp. 443-53. “A Evolução das Rendas e Atividades Rurais Não-Agrícolas no Brasil,” in Blanco, Mauricio F. (eIPEA, forthcoming.

d, Steven M., “Os determinantes da Eficiência Técnica no Centro-Oeste Brasileiro,” in Região e Espaço no Desenvolvimento Agrícola Brasileiroand Gervásio Castro de Rezende, Rio de Janeiro: IPEA, forthcoming. “The Distribution of Subsidized Agricultural Credit in Brazil: Do Interest GMatter?” Development and Change, 32(3), June 2001, pp. 465-490. “Interest Groups and the Pattern of Protection in the Brazilian Agricultural Sector,” Contemporary Economic Policy, Vol. 18, No. 4, October 2000, pp. 462-476.

Agricultural Sector and the Limitations of the 1995/96 Agricultural Census,” Revista deEconomia e Sociologia Rural, Vol. 39, No. 3, Jul./Sept., 2001, pp. 179-203.

d, Steven M. and GervásioEconomy-Wide Policy Reforms: The Case of Brazilian Agriculture, 1980-9Contem

-----, “Mudanças na Distribuição Espacial da Produção de Grãos, Aves e Suínos no Brasil:Papel do Centro-Oeste,” Planejamento e Políticas Publicas, No. 19, June, 1999. de Melo, Fernando, “O Plano Real e a Agricultura Economia Política, Vol. 19, No. 4 (76), October-December/1999. Censo Agropecuario, Rio de Janeiro: FIBGE, various years.

Jaramillo, Carlos Felipe, “Liberalization, Crisis, and Change: Colombian Agriculture in th1990s,” Economic Development and Cultural Change Vol. 49, No. 4, 2001, pp. 821

Countries: Measuring the Effect of Sectoral and Economywide Policies," The World Bank Economic Review, Vol. 2, No. 3, 1988, 255-72.

Mauro de Rezende, "Nova Energia", Agroanalysis, V. 20, N. 4, April 2000, pp.

29

Lopes, Mauro de Rezende, and Ignez Vidigal Lopes, coordenators, Efeito Das Políticas Públicas Sobre a Competitividade das Agroindústrias Brasileiras, Fundação Getulio Vargas,

Mendoasilia: Embrapa-SPI/

Quirozrspective,” Economic Development and Cultural Change, 49(1), October

Quiroz, Jorge and Alberto Valdés, “Agricultural Incentives and International Competitiveness in

PRI, 1993.

Schuh, rican Journal of

United iro, Atlas of Human

Villa V de Preços

World 1986.

IBRE, Centro de Estudos Agrícolas, 1998. nca de Barros, José Roberto, and Evandro Fazendeiro de Miranda (eds.), Agricultura e Estabilização no Brasil: Coletânea de Artigos 1995-1998, BrMinistério da Fazenda, 1998. , Jorge, and Luis Opazo, “The Krueger-Schiff-Valdés Study Ten Years Later: A Latin American Pe2000, pp. 181-96.

Four African Countries: Government Interventions and Exogenous Shocks,” in Valdés, A. and K. Muir-Leresche (eds.), Agricultural Policy Reforms and Regional Market Integration in Malawi, Zambia, and Zimbabwe, Washington D.C.: IF

Rezende, Gervásio Castro de, Estado, Macroeconomia e Agricultura no Brasil, Rio de Janeiro e Porto Alegre: Editora da UFRGS e IPEA, 2003.

-----, “Heterodox Stabilization Plans and Agriculture in Brazil, 1986-9l,” in Ian Goldin (ed.), Economic Reform, Trade and Agricultural Development, New York: St. Martin’s Press, 1993, pp. 189-211.

Rossi Jr., José Luiz, and Pedro Cavalcanti Ferreira, “Evolução da Produtividade Industrial Brasileira e Abertura Comercial,” TD No. 651, Rio de Janeiro: IPEA, 1999.

Edward G., “The Exchange Rate and U.S. Agriculture,” AmeAgricultural Economics, Vol. 56, No. 1, 1974, pp. 1-13. Nations Development Program, IPEA, and Fundação João PinheDevelopment in Brazil (2003). erde, Carlos Monteiro, “Modificações Recentes na Política de Garantia

Mínimos,” in Transformações da Agricultura e Políticas Públicas, Gasques, J.C. and J.C.P.R. Conceição (eds.), Brasília: IPEA, 2001. Bank, World Development Report 1986. New York: Oxford University Press,

-----, World Development Report 2000/2001. New York: Oxford University Press, 2001.

30

FIGURrop

Dec

-88

E 1and

Dec

-89

Rea

Jun-

86

Dec

-86

l Pri f C L (1 =100

0

100

200

300

400

500

600

Jun-

80

Dec

-80

Jun-

81

Dec

-81

Jun-

82

Dec

-82

Jun-

83

Dec

-83

Jun-

84

Dec

-84

Jun-

85

Jun-

88

Jun-

89

Jun-

Jun-

91

Dec

-92

Jun-

98

Note: Deflated with IGP-DI inflation index.Source: Getulio Vargas Foundation.

ce o

Jun-

87

Dec

-87

2/98

90

Dec

-90

)

Dec

-91

Jun-

92

Jun-

93

Dec

-93

Jun-

94

Dec

-94

Jun-

95

Dec

-95

Jun-

96

Dec

-96

Jun-

97

Dec

-97

Dec

-85

Dec

-98

31

Figure 2aReal Price Indices of Soybeans: Domestic and Foreign (1991=100)

0

50

100

150

200

250

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic Chicago

Figure 2b Real Price Indices of Soybeans:

Domestic and International in Local Currency (1991=100)

0

50

100

150

200

250

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic Chicago

32

Figure 3aReal Price Indices of Corn: Domestic and Foreign (1991=100)

0

50

100

150

200

250

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic Chicago

Figure 3bReal Price Indices of Corn:

Domestic and International in Local Currency (1991=100)

0

50

100

150

200

250

300

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic Chicago

33

Figure 4aReal Price Indices of Rice: Domestic and Foreign (1991=100)

0

20

40

60

80

100

120

140

160

180

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic FAO

Figure 4bReal Price Indices of Rice:

Domestic and International in Local Currency (1991=100)

0

20

40

60

80

100

120

140

160

180

200

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic FAO

34

Figure 5aReal Price Indices of Wheat: Domestic and Foreign (1991=100)

0

50

100

150

200

250

300

350

400

450

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic Chicago

Figure 5bReal Price Indices of Wheat:

Domestic and International in Local Currency (1991=100)

0

50

100

150

200

250

300

350

400

450

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic Chicago

35

Figure 6aReal Price Indices of Sugar: Domestic and Foreign (1991=100)

0

50

100

150

200

250

1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Domestic NY

Figure 6bReal Price Indices of Sugar:

Domestic and International in Local Currency (1991=100)

0

50

100

150

200

250

1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Domestic NY

36

Figure 7aReal Prices of Coffee: Domestic and Foreign (1993=100)

0

100

200

300

400

500

600

700

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic NY

Figure 7bReal Prices of Coffee:

Domestic and International in Local Currency (1993=100)

0

100

200

300

400

500

600

700

1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001

Domestic NY

37

38

Figure 8: Percent of the Population with Monthly Per Capita Income Below R$37.75 States of Brazil, 2000

39

heast of Brazil, 2000

Figure 9: Percent of the Population with Monthly Per Capita Income Below R$37.75 Municipalities in the Nort

Figure 10Total D): Southeast Employment in the Agricultural Sector (PNA

3,000,000

3,500,000

4,000,000

4,500,000

5,000,000

1992

1993

1994

1995

1996

1997

1998

1999

2002

2000

2001

40

Figure 11Total Employment in the Agricultural Sector (PNAD): Northeast

7,000,000

7,500,000

8,000,000

8,500,000

9,000,000

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

41

Figure 12Brazil: Agricultural Profit per Working Family Member per Month (1995-96)

22 2541

6597

120

169

262

538

0

100

200

300

400

500

600

0-1 1-2 2-5 5-10 10-20 20-50 50-100 100-200 200-500

49% of estab. < 10ha

81% of estab. < 50ha

42

Figure 13Northeast: Agricultural Profit per Working Family Member per Month (1995-96)

15 20 2635

4357

85

128

287

0

50

100

150

200

250

300

350

0-1 1-2 2-5 5-10 10-20 20-50 50-100 100-200 200-500

88% of estab. < 50ha

96% of estab. < 200ha

43

Figure 14Southeast: Agricultural Profit per Working Family Member per Month (1995-96)

8851 68 87

122

170

323

558

921

0

100

200

300

400

500

600

700

800

900

1000

0-1 1-2 2-5 5-10 10-20 20-50 50-100 100-200 200-500

21% of estab. < 5ha4% of estab. 51% of estab. < 20ha

44

Figure 15South: Agricultural Profit per Working Family Member per Month (1995-96)

49 5276

103142

198

326

588

1125

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

0-1 1-2 2-5 5-10 10-20 20-50 50-100 100-200 200-500

19% of estab. < 5ha

64% estab, < 20ha

45

Figure 16Center-West: Agricultural Profit per Working Family Member per Month (1995-

96)

46 42 52 64 7297

132

222

491

0

100

200

300

400

500

600

0-1 1-2 2-5 5-10 10-20 20-50 50-100 100-200 200-500

59% of estab. < 100ha

13% estab. < 10ha

46

Real Real Real PolicyProduct Period Domestic International Exchange +

Price Price Rate Residuala

(1) (2) (3) (4)

Importables Beans, corn, (1982-86)-(87-89) -32 -4 -21 -8 cotton, and rice (1987-89)-(90-94) -22 -3 -24 4

(1990-94)-(95-98) -17 4 -29 8

Wheat (1982-86)-(87-89) -46 1 -21 -26(1987-89)-(90-94) -45 -16 -24 -9(1990-94)-(95-98) -14 11 -29 7

Exportables Cocoa, oranges (1982-86)-(87-89) -21 -7 -21 6 and soybeans (1987-89)-(90-94) -42 -28 -24 3

(1990-94)-(95-98) -16 8 -29 7

Coffeeb (1982-85)-(87-89) -29 -20 -21 8(1987-89)-(90-94) -34 -32 -24 14(1990-94)-(95-98) 41 49 -29 35

Food Prices Food component (1982-86)-(87-89) -14 of CPI (1987-89)-(90-94) -13

(1990-94)-(95-98) -9

Notes:a) The residual is presented net of the interaction between the real international price and the real exchange rate in order to isolate the impact of policy.b) The 1986 coffee prices were excluded because this was an atypical year. Prices were more than double those of 1985 and 1987.Source: See text for details and footnote 14 for data sources.

Percentage change

TABLE 1

for Selected PeriodsDecomposition of Changes in Domestic Agricultural Prices

47

Product 1980-84 1985-89 1990-94 1995-98 1999-01

ImportsWheat1 1,215,680 390,897 685,278 1,098,715 890,987 Cotton2 6,965 108,420 389,019 709,076 258,871 Milk3 52,338 178,810 190,838 542,449 324,804 Rice4 103,231 149,484 290,585 411,559 222,869 Corn 147,727 107,667 138,465 137,031 112,313 Beans 20,300 29,624 52,342 95,409 44,009

Subtotal 1,546,241 964,902 1,746,527 2,994,238 1,853,853 Index (1985-89 = 100) 160 100 181 310 192 Share of Total 0.60 0.46 0.54 0.48 0.50

Total Ag. Imports 2,561,215 2,098,302 3,231,770 6,263,002 3,700,236

ExportsSoybeans5 3,404,617 3,205,105 3,139,822 4,695,374 4,203,743 Orange Juice6 942,706 1,077,317 1,117,855 1,196,668 993,440 Sugar 1,106,185 423,545 708,410 1,815,317 1,709,508 Cocoa7 793,497 703,678 305,658 146,680 96,551 Coffee8 3,177,229 2,756,582 1,657,798 2,587,764 1,808,463 Beef 598,885 593,680 530,222 493,221 826,795 Pork 7,152 21,224 63,072 144,250 222,224 Chicken 361,403 292,376 492,453 774,025 943,826

Subtotal 10,391,674 9,073,508 8,015,290 11,853,298 10,804,548 Index (1985-89 = 100) 115 100 88 131 119 Share of Total 0.83 0.82 0.78 0.80 0.80

Total Ag. Exports 12,464,345 11,029,268 10,240,186 14,788,598 13,569,234

Table 2Average Annual Trade of the Principal Agricultural Products (Thousands of 1998 U.S.$)

Notes:1) Includes wheat plus flour.2) Includes carded and combed, lint, linter, and waste.3) Milk equivalent as defined by the FAO.

8) Includes green, roast, and extracts.Source: Authors' calculations based on FAO data.

4) Includes rice, broken, and husked.5) Includes beans, soy cake, and oil.6) Includes concentrate and single strength.7) Includes cocoa butter, powder, cake, paste, and beans.

48

Products1980-84 1985-89 1990-94 1995-98 1999-01 1980-84 1985-89 1990-94 1995-98 1999-01 1980-84 1985-89 1990-94 1995-98 1999-01

ImportablesBeans 4,996 5,392 4,924 4,362 4,330 96 100 115 111 117 103 100 126 139 146Corn 11,663 12,774 12,688 12,510 12,660 86 100 114 135 147 94 100 115 138 148 Corn (CW) 1,105 1,533 1,631 1,839 2,031 57 100 114 154 178 80 100 108 130 136Cotton 1,468 1,771 1,291 828 797 75 100 76 51 95 90 100 104 109 208 Cotton (CW) 85 115 155 223 446 76 100 127 217 783 104 100 95 114 200Rice 5,766 5,506 4,316 3,567 3,576 85 100 90 88 105 81 100 115 135 161 Rice (CW) 2,248 1,859 1,057 779 1,001 100 100 64 62 99 83 100 113 149 183 Rice (RS+SC) 784 923 1,017 1,005 1,100 73 100 119 123 152 86 100 108 113 128Wheat 2,298 3,349 1,903 1,430 1,507 40 100 48 43 46 61 100 86 100 103

Total Area 26,191 28,792 25,122 22,697 22,870 Average Index 92 100 87 80 79 76 100 89 86 102 86 100 109 124 153

ExportablesCocoa 540 663 699 710 685 86 100 84 70 49 106 100 80 65 47Coffee 2,360 2,801 2,554 1,981 2,294 89 100 85 81 111 104 100 94 113 135Oranges4 599 757 954 953 944 78 100 123 141 147 98 100 97 112 118Soybeans 8,607 10,240 10,541 11,683 13,584 81 100 112 147 187 97 100 109 130 142 Soybeans (CW) 1,531 3,186 3,656 4,407 5,519 42 100 121 167 238 87 100 106 122 138Sugarcane 3,130 4,074 4,179 4,790 4,959 73 100 105 128 132 95 100 103 109 109 Sugarcane (SE) 1,518 2,012 2,198 2,811 3,034 74 100 111 137 z 96 100 105 107 107 Sugarcane (NE) 1,144 1,373 1,291 1,203 1,139 80 100 88 86 83 96 100 93 96 100

Total Area 15,236 18,535 18,928 20,116 22,466 Average Index 82 100 102 109 121 81 100 102 113 125 100 100 96 106 110 Av. Index w/out cocoa 80 100 106 124 144 98 100 101 116 126

4) Data for 1999 and 2000 only, due to change in measurement.

Area (thousands of hectares)

Area Harvested, Production, and Yield for the Principal Products in Selected PeriodsTable 3

3) The average indices are simple averages of the national totals.

Yield (1985-89=100)Production (1985-89=100)

Source: Authors' calculations based on data from Produção Agrícola Municipal (IBGE).

Notes:1) CW= Center-West region; RS=Rio Grande do Sul state, and SC=Santa Catarina state are in the Southern region; SE=Southeast region; NE=Northeast region.2) Prior to 1989 Tocantins (TO) was part of Goias. For consistency, TO has been added to the CW in the 1989-98 period.

49

Total Factor Agricultural Tractor Fertilizer Ag. Wage4 Fertilizer Price4

Productivity1 Employment2 Stock1 Consumption3

(1) (2) (3) (4) (5) 6)(1980=100) (millions) (units) (millions of MT) (1987=100) (1987=100)

1980 100 - 555,124 4.20 - -1981 102 13.20 575,220 2.75 - -1982 97 14.04 590,603 2.73 - -1983 104 13.01 601,926 2.29 - -1984 101 14.85 631,013 3.36 - -1985 105 15.10 663,487 3.20 - -1986 98 14.22 718,652 3.78 - -1987 110 13.98 749,686 3.76 100 1001988 106 14.10 770,119 3.73 61 921989 110 13.90 790,239 3.38 66 861990 103 14.04 801,914 3.16 62 761991 111 - 805,559 3.39 71 711992 109 14.04 800,949 3.54 59 681993 110 13.80 800,766 4.45 57 551994 117 - 809,941 5.02 55 501995 118 13.63 788,574 4.21 85 471996 - 12.61 753,037 5.02 93 501997 - 12.60 711,661 5.56 92 481998 - 12.12 - 5.74 93 46

Notes:1) TFP is from p. 111, and wheel tractors from p. 66, in Barros (1999). See text for details.2) Agricultural employment is from Pesquisa Nacional de Amostra de Domicílios (PNAD) , IBGE, various years. PNAD does not cover the Northern region of the country. Due to a change in methodology in 1992, for comparability over time we have assumed that 1992 employment equals 1990 employment. See Helfand and Brunstein (forthcoming) for details.3) Fertilizer consumption is from the U.N. Food and Agriculture Organization, www.fao.org.4) The agricultural wage and fertilizer price are the labor and fertilizer components of the FGV monthly index of the prices paid by agricultural producers. The index was created in June, 1986. We deflated these components with the IGP-DI inflation index, which is the

Table 4Selected Indicators of Productivity, Input Quantities, and Input Prices

50

Farm Size (ha) EstablishmentsBrazil Brazil NE SE S CW

0 - 5 0.37 0.07 0.18 0.05 0.06 0.015 - 10 0.13 0.05 0.07 0.04 0.08 0.0110 - 20 0.14 0.09 0.08 0.07 0.17 0.0220 - 50 0.17 0.15 0.13 0.14 0.22 0.0450 - 100 0.08 0.10 0.10 0.12 0.10 0.05100 - 1,000 0.10 0.32 0.29 0.40 0.26 0.341,000 - 10,000 0.01 0.17 0.12 0.16 0.10 0.4310,000 - 0.00 0.04 0.04 0.03 0.01 0.10

Total 1.00 1.00 1.00 1.00 1.00 1.00

Shares of Value of Agricultural Output by Region and Farm Size: 1995-96

Value of Output

Table 5

51

Farm Size (ha)1985 1995-96 1985 1995-96 1985 1995-96 1985 1995-96 1985 1995-96 1985 1995-96

0 - 5 .01 .00 .00 .00 0.01 0.00 0.03 0.02 .04 .04 0.03 0.045 - 10 .02 .01 .00 .00 0.01 0.01 0.03 0.02 .07 .05 0.04 0.0510 - 20 .07 .05 .00 .00 0.02 0.02 0.06 0.06 .11 .10 0.08 0.0820 - 50 .15 .10 .02 .01 0.05 0.04 0.17 0.14 .17 .19 0.20 0.2050 - 100 .11 .08 .03 .01 0.06 0.05 0.15 0.13 .14 .16 0.20 0.20100 - 1,000 .42 .40 .42 .32 0.48 0.38 0.45 0.43 .40 .41 0.42 0.421,000 - 10,000 .19 .31 .43 .56 0.33 0.40 0.12 0.17 .06 .05 0.03 0.0210,000 - .03 .04 .09 .09 0.04 0.10 0.00 0.02 .00 .00 0.00 0.00

Total 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

Farm Size (ha)1985 1995-96 1985 1995-96 1985 1995-96 1985 1995-96 1985 1995-96 1985 1995-96

0 - 5 0.15 0.16 .07 .04 0.07 0.04 .00 .00 0.03 0.01 .01 .005 - 10 0.13 0.11 .09 .05 0.10 0.06 .01 .01 0.03 0.01 .02 .0210 - 20 0.17 0.15 .15 .10 0.15 0.10 .04 .03 0.04 0.01 .09 .0720 - 50 0.22 0.19 .22 .16 0.22 0.14 .08 .08 0.08 0.05 .24 .2050 - 100 0.12 0.10 .12 .10 0.13 0.10 .09 .08 0.08 0.06 .18 .16100 - 1,000 0.18 0.20 .28 .36 0.27 0.34 .49 .49 0.39 0.34 .40 .441,000 - 10,000 0.03 0.08 .07 .18 0.06 0.15 .27 .29 0.29 0.44 .05 .1110,000 - 0.00 0.01 .00 .02 0.00 0.07 .02 .03 0.06 0.09 .01 .00

Total 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

Notes:a) The data for the South refer to the states of Rio Grande do Sul (RS) and Santa Catarina (SC). CW is the Center-West region. BR is Brazil. b) There were fewer than 50 farms in each of these cases, and they had an insignificant amount of production.Source: Authors' calculations based on data from the agricultural censuses, IBGE.

Rice (CW)

Coffee Cocoa

WheatImportables

Beans Corn Cotton

Table 6

Sugarcane

Shares of Value of Agricultural Output by Crop and Farm Size: 1995-96

Exportables

Rice (RS+SC))

OrangesSoybeans (BR) Soybeans (CW)

52

Farm Size (ha) Beans Corn Cotton Rice Rice Wheat Cocoa Coffee Oranges Sugarcane Soybeans (south)a (CW)a

0 - 5 15 14 18 72 1 -10 -36 -2 -23 -18 505 - 10 11 24 -4 68 1 -9 -35 -4 -24 -18 2410 - 20 21 29 -6 56 6 -6 -38 5 -23 -9 2920 - 50 19 38 -6 43 12 -3 -41 13 -22 -7 2550 - 100 29 55 12 21 17 -4 -44 17 -16 -6 24100 - 1,000 81 73 44 13 50 13 -45 16 -13 -2 251,000 - 10,000 164 95 49 11 56 35 -40 9 10 7 2910,000 - 107 108 31 -11 45 -b -b -b 42 -1 24

Total 34 63 24 19 45 6 -43 12 -12 2 28

Number of farms 2,946 3,461 438 228 140 143 112 526 889 403 420 in 1985 (1000s)

Notes:a) The data for the South refer to the states of Rio Grande do Sul and Santa Catarina. CW is the Center-West region.b) There were fewer than 50 farms in each of these cases, and they had an insignificant amount of production.Source: Authors' calculations based on data from the agricultural censuses, IBGE.

Percentage Change in Yields by Farm Size (1985-1996)Table 7

53

Farm Size (ha) Beans Corn Cotton Rice Rice Wheat Cocoa Coffee Oranges Sugarcane Soybeans (south)a (CW)a

0 - 5 25 11 49 132 -38 -25 -19 13 30 -13 1105 - 10 12 18 16 145 -40 -32 -14 -14 21 -18 3010 - 20 16 23 -7 149 -37 -35 -10 -6 27 -4 4420 - 50 12 35 -8 194 -32 -29 -13 14 16 1 4250 - 100 16 55 16 164 -28 -26 -9 18 21 -14 31100 - 1,000 78 149 117 116 20 -8 -6 23 45 -21 411,000 - 10,000 441 386 262 58 124 28 -20 32 151 38 9210,000 - 565 640 468 61 150 -b -b -b 859 127 86

Total 28 96 85 203 47 -16 -18 9 42 21 123

% change in price -59 -61 -56 -57 -65 -75 -75 -61 -69 -45 -58(1984-86)-(1995-97)

Notes:a) The data for the South refer to the states of Rio Grande do Sul and Santa Catarina. CW is the Center-West region.b) There were fewer than 50 farms in each of these cases, and they had an insignificant amount of production.Source: Authors' calculations based on data from the agricultural censuses, IBGE.

Percentage Change in Output per Farm (1985-1996)Table 8

54

Variable (1) (2) (3) (4)

Farm sizeSize 0.15** 0.10** 0.38** 0.33**Size2 0.05** 0.04** 0.02** 0.02**

Inputs Family Labor -0.87** -0.69** -0.79**Tractors 0.14** 0.10** 0.14**Animals 0.06** -0.03** -0.01

Land tenure (relative to owners)Renter 0.11** 0.18**Sharecropper -0.17 -0.23*Occupant -0.13** 0.00

Access to institutions/public goodsCredit 0.42** 0.43**Electricity 0.45** 0.55**Technical assistance 0.35** 0.37**Cooperatives 0.03 0.03Irrigation 0.38** 0.22*

PricesSoybean Price 0.05 -0.02Cattle Price 0.35** 0.42**Land Price 0.09** 0.16**Labor Price 0.03 -0.07**

Distance to São Paulo -0.25**

Constant FE FE FE 5.81**

Adjusted R2 0.89 0.99 0.99 0.99

Number of observations 8403 4682 1295 1295

Notes: 1) Regression 1-3 have county fixed effects. All regressions coorect for heteroscedasticity using White’s heteroscedasticity-consistent standard errors2) ** = statistically significant at the 1% level. * = statistically significant at the 5% level.

Regression Results for the Determinants of Profit Table 9

per Employed Adult Family Member

55