Using data from Medellín, second largest city in Colombia, we asses in this paper how a set of neighborhood characteristics determines wages and labor supply for workers in the city. We use GIS data to construct measures of the quality of environments where workers live. This paper focuses in the impact in labor supply and wages of the following set of characteristics: availability of public transportation, crime levels and density of economic activity. The empirical methodology consists of the estimation of linear equations for wages and worked hours, and we control for the selection of individuals into the neighborhoods they are observed. In order to do this, we estimate in a first stage a probabilistic model of neighborhood selection from which selection correction terms are obtained; these correction terms and included in the linear equations for wage and worked hours in a second stage. In addition, we control for sample selection as well. We find that the endogeneity of the location decision tends to overestimate the magnitude of the effect of neighborhood characteristics on labor market outcomes. Nevertheless, the effect of some characteristics is still significant and important after we control for the possibility of selection into neighborhoods.
