Gains and losses from trade liberalization are often unevenly distributed inside a country. For example, if budget shares vary according to household income, changes in commodity prices will redistribute an overall welfare change between household types. Household incomes will also be differentially affected. Sectoral differences in factor-intensity mean that changes in industrial structure cause redistribution of income between primary factors. Particular primary factors (such as capital, or less skilled labour) may contribute disproportionately to the incomes of certain household types. The fortunes of such households indirectly depend on the prospects of particular sectors. We emphasize these distributive issues, especially those arising from the income side. At the same time we distinguish households by regions (within the country). The regional distinction sharpens the contrast between groups of households. Particular regions have their own patterns of economic activity and so are differently affected by changes in the industrial protection structure. Since regional household incomes depend closely on value-added from local industries, economic change will tend to redistribute income between regional households. If the regional concentration of poverty is more than we could predict by regional primary factor endowments and industry structure, the addition of a regional dimension will add power to our analysis of income distribution beyond the mere addition of interesting regional detail. The paper deals with these issues more fully. We extend previous regional modeling of Brazil to include the intra-household dimension, addressing poverty and income distribution issues that may be caused by trade integration. An applied general equilibrium (AGE) inter-regional model of Brazil underlies our analysis, with a detailed specification of households. The model is static and solved with GEMPACK. The Representative Household (RH) hypothesis is abandoned; instead a micro-simulation (MS) model is used to track changes in household income and expenditure patterns. This micro-simulation model is built upon two Brazilian household studies: (1) the Household Budget Survey (POF, IBGE, 1999) covers detailed expenditure patterns for 16,013 households and 11 regions in Brazil in 1996; (2) the National Household Sample Survey (PNAD, IBGE, 1997) is a yearly survey that includes detailed information about household employment and income sources, with 331,263 observations. We integrate the two data sources to produce a detailed mapping of expenditure and income sources for 112,055 Brazilian households and 263,938 adults, distinguishing 42 activities, 52 commodities, and 27 regions. We link the AGE and MS models together, solving them iteratively to get consistency between results. After a shock the AGE model communicates changes in wages and employment by industry and labour type to the MS model that individually simulates the changes in employment, income and expenditure patterns for each household. The new expenditure pattern is then communicated to the AGE model, and the process is repeated until the two models converge. The final results from the MS model enable us to estimate changes in poverty and income distribution measures, both nationally and for regions within Brazil. We use the model to analyze poverty and income distribution impacts of the Free Trade Area of Americas formation upon the Brazilian economy. In the particular simulation we examine, freer trade leads to increased employment, especially for lower-paid workers. Poor households, which contain more unemployed adults, benefit most. This leads to a reduction in poverty in all 27 Brazilian states.
