The dissertation addresses the troubled relationship between authoritarianism and public goods. Political scientists have mixed expectations about the incentives of dictators to provide public goods that foster economic development. I argue that authoritarianism is not the main obstacle in the pursuit of prosperity, but neither are authoritarian institutions a reliable advantage. The cohesiveness and infrastructural capacity of the state apparatus are more important than the political regime. In turn, to understand ‘state capacity,’ we must look at the historical origins of state institutions and their persistence over time.The dissertation contributes in this direction by studying a path of institutional development: the aftermath of a social revolution. Specifically, I study the state-building dynamics after the Mexican Revolution (1920-1940) and their impact on the public education system in the short and long terms. I show that the rivalry between the national leaders and their subnational allies was the driver of the massive expansion of primary education access. In turn, the varying political strength of the subnational allies explains the geographic patterns of schooling supply and their persistence in the long run. The dissertation develops two theoretical pieces: one on the incentives of autocratic leaders to update the state’s institutions; another one on the obstacles that state leaders face in this endeavor. On the leaders’ incentives, the dissertation offers qualitative evidence on the effect of elite conflict inside the ruling coalition on the pace of expansion of public education. On the obstacles, I present four case studies to illustrate different scenarios of subnational resistance to the supply of national schools based on the political resources of subnational elites. These cases are Yucatán, Nuevo León, Tlaxcala, and Durango. I further support my argument on incentives and obstacles with schooling data on rural and urban education, and analyze alternative explanations through a series of multivariable regressions. Finally, I study the case of Chiapas as an example of why public education can be a source of political power. Using original microdata, I study how the role of teachers in mobilizing communities to request land redistribution made the national leaders stronger than their subnational allies.