Agglomeration economies are externalities that impact on prices in the economy. The interactions that occur in agglomerations are relevant for the understanding of the benefits generated by proximity. These benefits directly affect workers' wage, real estate prices, etc. Many papers have focused on non-market interaction in aggregated labor markets, but intra-urban labor markets have received less attention. Seeking to fulfill such lack, 1 km² areas of the Metropolitan Area of São Paulo (MASP) are taken as the scope of analysis of this study. The central objectives of this dissertation are twofold. First, in chapter 3, I identified and characterized the most relevant areas in terms of job agglomeration in the MASP, named subcentral business districts (SBD). For this purpose, the geocoded matched employer-employee database of the Ministry of Labor (RAIS-TEM) was used. I developed a new empirical approach to identify the SBD, using Geographicaly Weight Regression and cutoff rules of identification. The results identified three SBD in the years of 2002 and 2008, and only two in 2014. Considering the two initial periods, the SBD are located in the municipalities of Barueri (SBD-BAR), São Paulo (SBD-SAO) and São Caetano do Sul (SBD-SCS). In the last year, the SBD are located in Barueri and São Paulo municipality only. The employment located in the central areas of São Paulo shows a relatively higher amount of employee than in the other SBD areas. SBD-BAR and SBD-SCS lost not only in terms of employment, but also in terms of area. In 2002, these last two SBD occupied areas of 5 and 7 km² respectively. In 2014, the SBD-BAR occupied 1 km², while the SBD-SCS is not ranked at all. In São Paulo, the area ranges from 79 km² to 90 km². The results stemmed from the first paper suggest a high spatial concentration of employment in the MASP. In the second paper (chapter 4), the objectives are to identify the impact of agglomeration on workers' wages and test the agglomeration spatial attenuation hypothesis (SAH). For that, I use employer-employee RAIS database and consider a specification with multiple fixed effects and spatial lags of the employment agglomeration as a strategy to SAH identification. Even in the face of a more restrictive specification, the results suggest a positive effect of agglomeration, which is attenuated as the spatial distance increases. In other words, agglomeration affects positively workers' wage in the workplace area. Estimates without solving for endogeneity between wage and agglomeration indicate a direct effect of 0.039% (in the area itself), a first-order indirect effect of-0.11% (in the contiguous areas), and a second-order effect of-0.23% (in the ring around the contiguous areas), normalized by 100,000. When considering instrumental variables, the estimated direct effect on wages becomes 1.78%, the first-order effect changes to-2.12%, and the second order effect is not statistically significant.