Essays on the Distributive Politics of Bureaucracy

Bureaucrats are hired to produce public goods. Yet, despite the distributive implications of this canonical rationale, bureaucrats are generally absent from theories of "who gets what." The three papers in this dissertation advance a role for bureaucrats in the distribution of public goods and services premised on their work in policy implementation. I provide new theory and evidence to answer three questions. The first paper asks the question: under what conditions do bureaucrats’ actions generate inequalities in the provision of public services? The second paper inquires: how does the design of bureaucratic oversight institutions influence a state's capacity to implement policy and citizen access to services? The third essay asks: how does the co-production of public goods by politicians and bureaucrats influence voters' ability to hold politicians to account? In the first paper, I study the conditions under which bureaucratic bias (discrimination) in the allocation of services generates inequality in access. I argue that citizens' principal mechanism of control over bureaucrats is to complain to a politician. When politicians respond to complaints by tightening oversight of bureaucrats, differences in citizens' access to complain induce bureaucrats to devote more effort to groups with the loudest voices. I test this theory using a national-scale factorial audit experiment of Colombia's two largest national social welfare programs to measure bureaucratic effort behaviorally. I find that bureaucrats provide less information about social welfare programs to poor citizens and internal migrants. Consistent with the theory, such bias manifests most strongly in places with greater inequalities in citizens' ability to access the state and on tasks where oversight from politicians is most likely. These results are unlikely to reflect taste-based discrimination or screening. This paper shows that inequality in access to public goods and services can emerge even when politicians' budget allocations to public goods are equitable. In the second paper, I examine the distributional consequences of the use of citizen complaints in bureaucratic oversight. I study the adoption and consequences of bureaucratic oversight institutions in the context of service provision. Specifically, I consider a politician's choice to use (or ignore) information generated by complaints when monitoring a bureaucrat. Complaints generate information that direct a politicians' remediation of bureaucratic decisions and may increase bureaucratic effort. However, when costs of complaint vary across the population, the use of this information generates inequality in the distribution of service outputs, improving the access of citizens that can complain while reducing the access of citizens that cannot. Further, relying on citizen information can build or erode a state's capacity to accurately implement public policies, depending on the distribution of these costs across the population. This paper introduces citizen complaint systems as an institution that shapes both policy implementation capacity and distributional outcomes in comparative perspective. In the final paper, I start from the observation that in many theories of electoral accountability, voters learn about an incumbent’s quality through the observation of public goods outcomes. However, politicians rely on bureaucracies to produce public goods. Across contexts, politicians work with bureaucracies of markedly different qualities. In this paper, I argue that accountability relations between voters and politicians yield different empirical implications at different levels of bureaucratic quality. I introduce a model of electoral accountability with a voter, a politician, and a bureaucrat. The model identifies observational equivalencies between (i.) the implications of pooling equilibria that emerge at high and low levels of bureaucratic quality (with informed, rational voters) and (ii.) the findings of existing studies that are interpreted to indicate a lack of accountability due to uninformed or irrational voters. I demonstrate the plausibility of the model by introducing and validating an original measure of bureaucratic quality in Brazilian municipalities. I use this measure to extend four studies on corruption and accountability. I conclude with implications for the comparative study of accountability across the world's democracies