Income distribution in Brazil: A regional approach1

Abstract

This article is an attempt to identify the determinants of regional differences in the distribution of income in Brazil using data from the personal income tax returns from 1970 to 1974. We formulated and tested linear regression models using pooled cross‐section and time series data. The most complete model considered explanatory variables such as: (1)the real mean income of the returns; (2) productive structure of the regions; (3) the proportion of labour income; (4) the proportion of single income tax filings; and (5) a set of dummy variables to capture shifts in the cross‐section relationships over time. This model specification provided a good overall fit, coefficients being, in general, statistically significant and having the appropriate sign. The estimated regression enabled us to make the important prediction that as material progress takes place, inequality will tend to stabilize, but at a much higher level than that presently experienced by developed countries of the western hemisphere.