Abstract
Intergovernmental transfers finance most local government spending in developing and developed nations. The redistribution of government revenue across subnational governments, however, might be subject to distortions for political motives, such as political connections. It is well documented by the literature that central governments favor the politically connected with increased resources, but evidence on the extent to which such distortion affects individuals in local government jurisdictions is scarce. Throughout this dissertation, I seek to answer whether co-partisanship between the heads of the central and subnational governments affect government behavior to such an extant that it also affects outcomes of local government constituencies. I examine connections between the federal and local governments in Brazil and how these connections affect local outcomes.
In chapter 1, I study how political connections between the federal and local governments in Brazil affect mortality rates of Brazilian municipalities. Whether municipalities become connected to the federal government might be correlated with unobservable shocks that affect mortality. Besides, residents can vote for the president's party in local elections if they know they can benefit from being connected to the federal government. To address these identification challenges, I exploit quasi-experimental variation in municipalities political connectedness, generated by close elections, within a regression discontinuity design (RDD). I compare municipalities where co-partisans of the president barely won and barely lost the local race. I find that municipalities that become connected to the federal government experience a 16% reduction in mortality, equivalent to 0.72 fewer deaths per one-thousand residents. The result suggests that some action of the federal government toward municipalities, motivated by thee political identity of the local politician, drives mortality rates down in aligned jurisdictions.
In chapter 2, I investigate whether distortions in transfers allocation could explain the reduction in mortality found in chapter 1. While intergovernmental transfers financed approximately 90% of all local spending in Brazil in 2017, discretionary transfers could match only a small fraction of that figure, at about 2% of total local spending. Discretionary transfers conditional on capital expenditure, however, financed one-fourth of all investments made by local governments in the same year, and large distortions in this type of transfers could be consequential for mortality. I find that political connections increase the amount of discretionary transfers for capital expenditure received by aligned jurisdictions by 53% and investment spending by 20%. The difference is substantial and equivalent to 17% of all local investments. I estimate the effect of political alignment on long-term mortality, and I find that mortality is low for several years after municipalities lose their initial connection, which supports the claim that political alignment affects mortality through capital transfers, as local investments allow for a permanent expansion of public service provision. Estimates of the effect of political alignment on cause-specific mortality also suggest that the differences in all-cause mortality found earlier are likely due to increased spending in several functions of government. This is supported by the finding that connected municipalities have lower health operating expenditures and admit fewer patients to hospital beds, which suggests that fewer individuals seek health care in connected jurisdictions due to higher spending in other areas. Finally, I test for alternative mechanisms that could explain the results in chapter 1, namely channels that could affect mortality through economic growth and the election of left-leaning politicians, and I find no evidence that these could affect health outcomes.
In 3 I study whether political connections played a role in the allocation of health resources during the COVID-19 pandemic. I leverage detailed data on health inputs available through the Brazilian publicly funded health system (SUS) to show whether resources to address the pressure created by the COVID-19 pandemic on health systems have been distributed to local governments on political grounds. These were heavily branded federal and state government resources, which allow me to conjecture about an alternative mechanism that could explain the difference in all-cause mortality discussed in chapter 1, i.e., the allocation of resources that, according to voter's perception, were entirely funded by upper-tier governments, such as national program spending and large infrastructure projects. I find no evidence that medical resources were disproportionately distributed to the politically connected during the COVID-19 pandemic, which leaves little to no room for other channels to explain the results presented in chapter 1.
