The Doha Development Agenda and Brazil: Distributional Impacts^†

Brazil has one of the worst patterns of income distribution in the world. The persistence of this problem has interested researchers worldwide, and has stimulated a lively debate in Brazil. The role played by past trade liberalization and the possible effects of future changes such as the Doha Round, are central to this discussion. Global economic integration has complex effects; it is not obvious whether Brazil's poor will benefit. For example, although Brazil's poorest regions and workers depend on agriculture, some have argued that increased agricultural exports will benefit mainly landowners rather than poorer people.

This paper addresses the potential impacts of a Doha agreement on income distribution and poverty in Brazil, and seeks to extend previous results. It focuses on impacts within agriculture, and on effects linked to ownership of land and other production factors.

Background and Motivation

The share of agriculture-based products in Brazilian exports began to rise again by 1994, after a steady decline since the 1970s. Similarly, the rate of growth of agricultural gross domestic product (GDP) has been rising, mainly from the late 1990s, at higher rates than total GDP. For the Brazilian economy, the agricultural sector is still important. With strong forwards and backwards linkages with other sectors in the economy, agricultural GDP was about 10.3% of total GDP in 2003, and accounted for about 19% of total population (in 2002).

Studying the effects of Doha development agenda (DDA) scenarios on income distribution and poverty in Brazil, Ferreira-Filho and Horridge concluded that trade liberalization would benefit the poor. Around a quarter of a million persons would leave poverty due to the DDA scenario simulated. Their analysis, however, did not differentiate between the impacts on either different farm sizes or types of workers in Brazilian agriculture. These impacts, however, can be of considerable interest, due to regional specialization and different technological patterns across regions. This paper addresses these questions more thoroughly.

Poverty in Brazil: An Overview

Although Brazil is a large country with many poor people, it is not among the world's poorest. Drawing on the 1999 Report on Human Development, Barros, Henriques, and Mendonça show that around 64% of countries have smaller per capita income than Brazil, as does 77% of world population. They also show that, while 30% of Brazil's total population is poor, on average only 10% are poor in other countries with similar per capita income.

Focusing on income insufficiency, the same authors show that in 1999, about 14% of the Brazilian population lived in households with income below the line of extreme poverty (indigence line, about 22 million people). About 34% or 53 million people lived in households with income below the poverty line. Green, Dickerson, and Arbache arrived at the same general results, concluding that the egalitarian consequences of trade liberalization were not important in Brazil for the period under analysis.

Brazilian poverty also has an important regional dimension. According to calculations by Rocha in a study for 1981–95, the richer Southeast region of the country, while counting for 44% of total population in 1995, had only 33% of the poor. For the poorer regions, on the contrary, the share of population is lower than the share of poor: 4.6% (9.3% of poor) for the North region, and 29.4% (44.3% of poor) for the Northeast region, the poorest region in the country. Table 1 shows more information about poverty and income inequality in Brazil in 2001.

Calculations by Ferreira-Filho and Horridge show that agriculture is an important employment sector for the poorest in Brazil. Using a ten-wage class classification and data from the Brazilian Household Survey for 2001, they showed that the lowest wage class accounts for 40% of the total agriculture labor bill, and agriculture accounts for about 41% of wages of the less skilled (lower-waged) workers.

Table 2 shows the importance of each broad income source for workers in agriculture, classified by occupation: permanent workers, temporary workers, self- employed producers, and employers¹. For comparison, the first column reports income sources for nonagricultural workers. As table 2 shows, wages are the main income source for every category. The reported values for employers, however, should be regarded as including the returns from land and capital stocks. The same applies to self-employed workers.

Transfers are an important income source both for self-employed workers and employers. These are mainly retirement pensions, which account for 23% of total income of self employed workers in 2001. The income profile for both permanent and temporary workers in agriculture closely follows that of nonagricultural workers-wages are 92–95% of total income.

Table 3 shows the composition of income inside agriculture, according to the type of income source, by region. In the southeast region, the states of São Paulo, Bahia, and Minas Gerais concentrate the bulk of the temporary and permanent employed workers in Brazil. São Paulo and Minas Gerais have also high shares of employers. Rio Grande do Sul and Santa Catarina (states in southern Brazil), on the other hand, account for high shares of self-employed workers in agriculture. Santa Catarina, another southern state, is also important for the self-employed, especially if one takes into account the state's small size.

The center-west states (Mato Grosso do Sul, Mato Grosso and Goias) are also interesting cases. Mato Grosso is the most important soybean-producing state in Brazil, followed closely by Rio Grande do Sul. Mato Grosso, however, shows small shares of all worker types, including employers. This follows from its particular land ownership structure, dominated by large and more capital-intensive farms. Note that while Rio Grande do Sul state accounts for 17% of the self-employed in Brazilian agriculture, Mato Grosso accounts for just 3%, but the employers' shares are about the same for both states (6% for RS and 7% for MT).

Potential Impacts of DDA Scenario: The Methodology

To assess the impacts of a DDA scenario on poverty and income distribution in Brazil, a computable general equilibrium model (CGE) of Brazil was used, linked to a microsimulation (MS) model. Ferreira-Filho and Horridge proposed this method, which guarantees consistency between both models².

The CGE model used is a static inter-regional model of Brazil based on the well-known ORANI-G model of Australia (Horridge) and was calibrated with Brazilian 1996 data. The model's structure is quite standard. Consumption is modeled through the Linear Expenditure System over composite commodities (domestic and imported). Exporters of each commodity face constant-elasticity³ foreign demand schedules. Production for exports or domestic markets are regulated by CET⁴ functions for each firm. Production is a nested LEONTIEF/CES structure for primary factors and composite inputs and labor is a CES function of ten different types of labor. This nonlinear model is solved with the GEMPACK software, and distinguishes between forty-two sectors and fifty-two commodities⁵; ten labor occupational categories.

All quantity variables in the model are disaggregated according to twenty-seven regions within Brazil, using an elaboration of the top-down regional modeling method described in Dixon et al. This methodology recognizes local multiplier effects: many service goods are little traded between regions, so that local service output must follow local demand for services.

On the income generation side of the model, workers are divided into ten different categories (occupations), according to their wages. These wage classes are then assigned to each regional industry in the model. Together with the revenues from other endowments (capital and land rents), these wages will be used to generate household incomes. The CGE model covers 270 different expenditure patterns, composed of ten different income classes in twenty-seven regions. In this way, all the expenditure side detail of our MS data set is incorporated within the main CGE model.

The Pesquisa Nacional por Amostragem de Domicílios (PNAD)(National Household Survey, Instituto Brasileiro de Geografia e Estatística), and the Pesquisa de Orçamentos Familiares (POF) (Household Expenditure Survey, Instituto Brasileiro de Geografia e Estatística) were the main sources of information for the household MS model. The model database contains 263,938 adults, grouped into 112,055 households, with records about wage by industry and region, and personal characteristics such as the ownership of land, type of work, years of schooling, sex, age, position in the family, and other socioeconomic characteristics.

Model Closure

Model closure was chosen to mimic the global trade analysis project (GTAP) model that generated the foreign price scenario. On the supply side, total national employment is fixed by occupation, with labor moving freely across sectors. The model allows substitution between occupations, driven by relative wages. Similarly capital is fixed nationally but is mobile between sectors and regions. Land stocks (used just in agriculture) are fixed.

On the demand side, a fixed trade balance enforces the national budget balance, which is accommodated by changes in real consumption, with investment and government spending fixed. The consumer price index (CPI) is the model's numéraire.

Finally, tax revenue losses due to tariff cuts are replaced: real aggregate revenue from all indirect taxes is kept fixed, via a uniform endogenous change in the power of indirect taxes on sales to households. This mechanism is equivalent to a lump sum tax, of value proportional to each household's spending.

The DDA Scenario

The simulated DDA scenario comprises cuts in agricultural tariffs according to a nonlinear (two-tier) formula with maximum cuts of 75%, cuts in domestic support for Organisation for Economic Co-operation and Development (OECD) agriculture, elimination of export subsidies, and 50% proportional cuts in nonagricultural tariffs. In the case of special and differential treatments, developing countries were given a two-third reduction of the developed countries, while no reduction was imposed for the least developed.

Simulation Results

As a result of the shocks, agriculture and agriculture-related industries (the food industry in general) expand. Model results show a general fall in activity in the Brazilian manufacturing sectors after trade liberalization, since these are the protected sectors in Brazil in the base year. This suggests that regions specializing in manufacturing would fare worse. Indeed, employment falls in São Paulo and Rio de Janeiro in the southeast (the most populous and industrialized states) which host the bulk of Brazil's manufacturing. Employment also declines in Amazonas, which is a free exporting zone. Reduced mining activity cuts jobs in Amapá.

The trade liberalization simulation redistributes economic activity toward poorer regions because manufacturing sectors (concentrated in the richer regions) shrink, and agriculture (concentrated in relatively poorer regions) grows.

Model results show an increase of 253,066 jobs in the agricultural activities. Most (197,187) would be new jobs creation, or workers coming to agriculture from unemployment, and the other part (55,882 workers) would be a net attraction of jobs from contracting industries. The job creation benefits the poor disproportionately: 57% of the new agricultural workers belong to the first three lowest income classes (78% if only the previously unemployed are considered).

Table 4 shows income variation (percent) by type of agricultural worker and income class of the household to which the worker belongs. New jobs in agriculture help the poor (top left of table 4), while rising land rents benefit agricultural employers (last column), who are mostly landowners.⁶ Richer employers benefit proportionately more (bottom right) since more of their income arises from rents. The aggregated result is an increase in income for every agricultural worker type, and a decrease in income for nonagricultural workers.

Since the PNAD includes data for land holdings (hectares [ha] owned), the following question can be addressed: do owners of large (area) farms benefit more than small farm owners? Table 5 shows that owners of larger farms do benefit more: the larger farmers group in the richest household would get an 8.2% income increase. But other farmers also gain, regardless of their land holding. Even the poorest landless group enjoys a 6.17% income increase.

However, area of land owned is not closely correlated with income class. It is apparent from table 4 that there are employers (mostly landowners) in every household income class. Indeed, the PNAD shows that 13.3% of farmers owning less than 25 ha fall in the lower five income groups while 55.2% of these farmers belong to the two highest income groups—even though 25 ha is a small farm area by Brazilian standards. Behind this lies a regional distinction: poorer farmers with big farms are concentrated in northern Brazil, and richer farmers with small farms in the south and southeast.

The simulated DDA scenario, then, generates an income rise to landowners, and a fall in income to nonagricultural workers. However, many previously unemployed get new jobs in agriculture. Considering that initially poverty is concentrated in agricultural households and among the unemployed, the jobs/income shift helps raise income of the poorest households (POF[1] and POF[2]) and reduces poverty overall.

Conclusions

The simulated DDA scenario, which was found to be poverty-reducing in previous work by the authors, is shown to also reduce poverty inside Brazilian agriculture.

Model results, then, contradict the notion that only landlords would gain from trade liberalization in the DDA agenda, an idea that became somewhat popular recently. The strong agricultural employment effect and the distribution of land ownership must be taken into account for this discussion.