Ethnicity and earnings in urban Perú

Our approach to the concept of ethnicity involves the usage of instruments in many of its several dimensions: mother tongue, parental background, religion, migration events and race. In order to approximate what can be called racial differences in a context like the Peruvian in which "racial mixture" is the main characteristic of the population, we use a score-based procedure to capture both the differences and the mixtures. By means of this procedure each individual is assigned intensities by pollsters in each of the four categories that correspond to the most easily recognized distinct racial groups in the Peruvian society: Asiatic, White, Indigenous, and Black. We find that the multidimensional race indicator is correlated with several human capital and physical capital assets, as well as with access to public services. Using Blinder-Oaxaca (B-O) decompositions we find that a substantial part of the earnings differences between racial groups cannot be explained by differences in individual characteristics. To take into account the fact that B-O doesn't consider the probability distribution of the individual characteristics, and specifically race in our case, we also use a semi-parametric technique for the estimation of differences in hourly earnings. This estimation treats the typical wage equations in a linear fashion but let estimators for the racial intensity effects to interact freely, without restricting them to a functional form. The results suggests that among wage earners after controlling for a large set of characteristics, there are racially related earnings differences in favor of predominantly White individuals. In the case of the self-employed, none of the empirical distributions of earning differences attributable to race is substantially above zero.Departamento de Economí