This paper aims to discuss how agglomerations economies are present in the equilibrium outcomes of the Brazilian formal labor market. There has been a wide discussion on how to correctly identify agglomeration economies given all the different types of endogeneity found in the labor market relationships, as well as taking into account all the relevant aspects that may affect the results. We make use of an individual-firm panel database from the Ministry of Labor (RAIS - Annual Report on Social Information) with information for six years (2003, 2004, 2005, 2008, 2009 and 2012). With the panel data setting, it is not only possible to account for individual unobserved characteristics constant in time, but also for sector and area effects. Moreover, by identifying skills according to the occupational position of the individuals in each firm, it is possible to control for the proximity to different skill levels (in the sector and municipality) to account for different levels of production knowledge externalities. Individual fixed effects control the potential endogeneity of the labor quality. In the case of labor quantity endogeneity, even if there is no consensus of how to best control for it, instruments based on long time lags are considered. The results show that there is a positive and significant effect of density over wages (Urban Economics literature), even when controlling for other relevant characteristics. Moreover, a measure of market potential, related to the New Economic Geography literature, does not capture this positive relationship with wages in the same way, changing sign in a specific setting. Finally, considering a quantile regression approach, there is an indication that agglomeration economies reinforce wage inequality, with a higher effect for the upper part of the wage distribution.
