Programmatic Parties and Local Politics                     

Do programmatic parties matter for local politics? This paper argues that anticipated electoral pressures can push local politicians from strong, programmatic parties to govern the same as those from non-programmatic parties. Specifically, I show that when programmatic politicians expect to be sanctioned for visible actions, this can lead to linkage switching, where politicians who initially campaign using programmatic appeals seek re-election using pork-based appeals. I support this claim with evidence from a regression discontinuity design, which estimates the effect of electing a mayor from Brazil's Workers' Party (PT), a programmatic party in a weak, non-programmatic party system. I find that PT mayors pursue visible policies, but have few impacts on less visible policies. The semi-automated text analysis of campaign platforms shows that PT candidates shift their rhetoric when seeking re-election, and the qualitative analysis of internal party documents illustrates that the party has historically perceived tensions between its programmatic and electoral goals.

(email for latest version)


Visibility and Local Electoral Accountability

Advocates for decentralization claim that the devolution of power to directly elected local governments increases electoral accountability, but it remains unclear if and how citizens use local elections to hold politicians accountable. The analysis presented here suggests that the optimistic assumptions used to justify decentralization reforms are likely misguided. Whereas previous work suggests that citizens have more information at the local level, I claim that they only have access to information about visible actions and outcomes. I test this argument with evidence from Brazil, including fine-grained budgetary data, electoral results from 2000-2012, and an original database of municipal election surveys. Consistent with expectations, I demonstrate that citizens do not sanction incumbents for actions that align with their stated preferences, such as spending on health care and education; instead, they sanction them for visible spending on infrastructure projects and material purchases.

(email for latest version)


Who Punishes Corrupt Politicians? Educational Attainment and Electoral Accountability           

When will voters punish corrupt politicians? Recent research has focused on contextual factors that determine whether voters sanction corrupt incumbents, but has largely neglected variation in individuals' beliefs about - and responses to - corruption. This paper provides evidence that educational attainment shapes individuals' beliefs about corruption and conditions how they respond to corruption revelations. Drawing on survey data, randomized corruption audits in Brazilian municipalities from 2003-2009, and disaggregated electoral returns, I find that: (1) low-education voters are more likely to believe all politicians are corrupt and express more willingness to tolerate corruption in exchange for competence; (2) low-education voters are less likely to punish incumbents for corruption; and (3) low-education voters are more willing to tolerate corruption in return for visible expenditures. These interactions between education and electoral accountability provide support for the main theoretical claim: educational attainment plays an important and under-appreciated role in shaping citizens' beliefs about - and responses to - corruption.  

(email for latest version)


Does Relative Performance Matter? Evidence of Asymmetric Benchmarking          

How do individuals use information about relative performance to evaluate incumbents? This paper draws on evidence from an original survey experiment to demonstrate that individuals incorporate information about relative performance when evaluating incumbents, but do so asymmetrically: they punish incumbents for poor relative performance more than they reward them for good relative performance. These results are consistent with psychological research on negativity bias and suggest that benchmarking may not be sufficient for accurate incumbent evaluations.

(email for latest version)