This thesis studies empirically the agglomeration forces posit in the economic literature. It overcomes several of the methodological limitations of previous empirical works. It privileges applications to developing countries that experienced increased integration to the world economy through trade liberalization or colonization. The empirical applications maintain a close connection with theoretical models, especially those issued from the recent literature known as New Economic Geography. Chapter 1 investigates the evolution of the world economic geography. A theoretical relationship between the proximity to demand (termed as “market access”) and the level of manufacturing wages is tested. Using trade gravity equations for the period 1980-2003 for 27 industries, measures of market access are built. Results indicate that 16 out of 27 manufacturing industries exhibit this wage response. The fact that panel data econometrics (static and dynamic) is employed reduces considerably the risks of omitted variable bias. The absence of effect of market access can be attributed to the influence of labor relocation, human capital requirements, supplier access, and labor regulations. Chapter 2 and 3 focus on the impact of a recent trade reform on Brazilian regional labor markets, and take advantage of detailed information on trade at intranational level, as well as microdata on firms and workers. In chapter 2, the same hypothesis postulated in Chapter 1 is tested, but using individual wages and a market access that captures demands from the rest of Brazil and international partners. Results confirm the impact of market access on wages, which is robust to controls for firm productivity, knowledge spillovers, and endowments. Chapter 3 continues the exploration of Brazil economic geography in a dynamic setting. In a general equilibrium framework, it is expected that wage differentials are arbitrated away through migration. The chapter explores the impact of market access on bilateral migration. Microdata are used to warrant that self-selection is properly taken into account, and to build migration flows breakdown by skill levels and industrial affiliation. Results confirm an important migration response to the evolution of market access, although migration costs and selection bias limit migration flows. In Chapter 4, the role of history in agglomeration processes is studied by following the evolution of cities founded in the American continent over the more than 3 centuries of Spanish rule. Major events affecting these cities (natural disasters, wars, etc) are employed to assess the role of the shocks on city relocation. The identification is made by looking at determinants of the duration of a city in a particular site and the effect of past outcomes (hysteresis) is measured by using the age of a city. Results indicate a great urban instability in the earlier periods of the Conquest, where political disorders together with an orientation to extractive activities seem to explain a high instability. In later periods, when the empire stabilizes politically, there are important changes towards strong urban stability. All these results suggest strong non-linearities and path-dependency in the agglomeration process.