Racial Classification and the Race-Wage Gap in Brazil: Evidence from Administrative and Survey Data

Abstract

Because race in Brazil is highly subjective, the racial classification of workers in any sort of survey requires careful consideration by survey administrators and those who would use the data for analysis. This paper provides an examination of how race is reported in administrative and survey settings, how the differences between the two affect measures of racial wage gaps and what conventional earnings decomposition can tell us about reporting differences and the nature of inequality in Brazil. I show that estimates derived from administrative data smaller wage gaps, driven by racial bias; estimates from survey data yield much larger wage gaps, driven not just by racial bias, but also by differences in observable worker characteristics. In addition to providing a warning about the importance of data source selection, these results provide a range of possible values for the race-wage gap, and its components, in Brazil that vary with assumptions on the nature of Brazilian racial identity.