This article analyzes the importance of firm characteristics to explain earnings in urban Bolivia. Initially I propose a new simple theoretical model of segmented labor market where, in equilibrium, individual and firm variables jointly determine earnings at the worker level. The key for achieving this equilibrium is that workers have both specific preferences and heterogonous skills provided by years of schooling, which are in turn associated to certain firms. Given the household surveys information, I estimate two alternative earnings functions from this model, one for unsalaried workers, for which there is detailed firm data and one for salaried workers, in which sector, size and formality are used as firm proxies. I find not only that firm characteristics are fundamental determinants of earnings but that regressions that include only individual characteristics present highly overestimated coefficients.