Rising Inequality in Mexico: Returns to Household Characteristics
By César Bouillon, Arianna Legovini, Nora Lustig (10/98, En)
Between 1984 and 1994, Mexico experienced a sharp increase in household income inequality. This trend is observed for a set of inequality measures: the Gini coefficient, the mean-log deviation, the Theil index, and the transformed coefficient of variation. The rise in inequality is robust to adjustments to income that account for underreporting and also robust to alternative measures of welfare - per capita and per adult-equivalent units (Table 1).
What accounts for this increase in income inequality? Two possible factors come immediately to mind. One is the widening gap in the returns to skill in the labor market between the less educated and the highly educated. Another factor could be uneven regional development with some parts of the country lagging behind.
In order to test these hypotheses, we estimate a reduced-form household income regression model and apply a simulation methodology on the overall distribution of income. The methodology, first proposed by Almeida dos Reis and Paes de Barros (1991) and Juhn, Murphy, and Pierce (1993) in the context of earnings equations, follows the generalization to the household income model proposed by Bourguignon, Fournier, and Gurgand (1998). The simulation allows the decomposition of the observed changes in income distribution by source. That is we can identify the contribution of changes in household "endowments" separately from the changes in the "returns" to those endowments (and from the changes in the residual term). Furthermore, the simulation allows isolating the effects of specific returns individually, such as returns to education and regional effects. In order to test the robustness of the results, the simulation is carried out for four inequality measures for both income per capita and per adult equivalent unit.
Because the method is applied to a reduced form household income model, strictly speaking the estimated coefficients are not "returns". While in the earning functions à la Juhn et al., the estimated coefficients represent the market returns to individual characteristics, the estimated coefficients presented here capture a whole range of endogenous decisions such as labor force participation and occupational choice, in addition to the market returns. For simplicity we will still refer to them as "returns."
Our results indicate that changes in the returns to education account for the lion share of the observed changes in inequality. In addition, the deterioration in the relative position of the southern part of the country is the second most important factor accounting for the increase in inequality between 1984 and 1994.
Last updated: 04/26/07
