Electoral Competition, Income Inequality and Public Goods : A Sub-national Assessment

Does democratization increase redistribution? Does inequality affect the quality of democracy? This dissertation explores these questions by focusing on the effects of electoral risk and income inequality on the provision of basic infrastructure at the household level in Mexico. The first part of the dissertation accounts for the state of the literature, and I propose a way to start thinking theoretically about inequality as a condition of social heterogeneity that signals the presence of elites prone to capture politics. A main challenge for testing these hypotheses is the lack of data, which I address by carefully constructing inequality measures at the municipal level in the second part of the dissertation using a 10% sample of the Mexican census. I also generate cleaner measures of provision than the ones provided in aggregate by the Mexican government. I additionally use two different measures of electoral competition to measure political incentives for parties. Building on work by others scholars, I introduce the concept of bundles of public goods, and show that public goods are better understood this way and not as individual goods. The empirical analysis, using three-level hierarchical logit models, shows a robust negative relationship between income inequality and the probability for Mexican households to have full coverage of running water, sewerage, electricity and literacy (schools). A change from very low to very high inequality would represent a probability drop of 20% for a household in the first quartile of income. I also find that state electoral incentives have no association with these probabilities, but at the municipal level the structure of the party system matters while the margin of victory does not. I find systematically no significant interaction effects between inequality and electoral risk, suggesting these are both sufficient, and not necessary, conditions to shift the provision of public goods. Overall, the results show heterogeneous effects for different public goods and different levels of government, and should be taken to suggest a more careful look at the level of analysis we use to answer questions on democracy and ex-post redistributio