WORKING PAPERS

Cash Transfers, Clientelism, and Political Enfranchisement: Evidence from Brazil
​(Revision Requested at the Journal of Public Economics)

This paper uses Brazil’s Bolsa Família to show that redistributive policies that are shielded from the influence of political intermediaries can reduce incumbency advantage for mayors, increase both electoral competition and candidate quality, reduce support for clientelistic parties, and lead incumbents to increase redistributive spending. The paper exploits a nonparametric multivariate regression discontinuity design and employs a novel identification strategy for the variation in program coverage. The theory proposes that cash transfers, by reducing the vulnerability of poor voters, make clientelism a less attractive strategy to incumbent mayors. Consequently, incumbents reallocate effort away from the practice into public good distribution.

​Download

Do Reelection Incentives Improve Policy Implementation? Accountability vs. Political Targeting
​(Under Review)

Although reelection prospects can increase policy effort by incumbents, it can also create incentives for politically-motivated targeting of resources, which might jeopardize both distributional efficiency and electoral competition. While existing empirical tests examine the potential countervailing effects of accountability and targeting in isolation, this article analyzes their net effect in the context of the world’s largest cash transfers program, Bolsa Família (BF) in Brazil. Using administrative data on more than 11 million households from both the BF registry and party membership rolls, we estimate these effects using a regression discontinuity design based on mayoral term limits. The evidence supports political targeting over accountability: although reelection prospects only weakly improve effort in program expansion, they drive mayors to aggressively extend BF benefits to ineligible, nonpoor households that are core party supporters. Furthermore, only politically-motivated targeting is positively correlated with future reelection probability.

​Download

Strategic Allocation of Anti-Clientelism Goods and the Breaking of Political Machines 
​(Under Review)

The success of clientelism depends on the machine’s ability to deliver benefits that elicit voters’support while still keeping them impoverished. This requires targeted goods to be valuable, but revocable. Exploring the flip side of this, this article explains why and how parties target voters with valuable but irrevocable goods: delivering these anti-clientelism goods erodes the grasp of opposing dominant machines. We empirically assess this redistributive logic with previously untapped administrative data on a Brazilian program of distribution of cisterns, an archetypal irrevocable good, employing panel data and a regression discontinuity design. Results indicate that incumbents primarily target cisterns to opposition strongholds, while avoiding their own. In contrast, they favor their their own strongholds with revocable goods. This mechanism is stronger in areas where cisterns are more valuable to voters, and where machines are more dominant.

​Download

A Variance Decomposition Approach for Bounding Neighborhood Effects (with Scott Abramson)
​(Under Review)

Political scientists have pointed to neighborhood contexts as an important factor explaining individual voter behavior. This paper presents a variance-decomposition method for placing bounds on the impact of neighborhood effects on political attitudes. The method can be used to complement other empirical approaches used by the discipline in the study of neighborhood effects, as it provides better external validity than the typical experimental design, and less restrictive assumptions than most observational studies. Then, using data from the universe of voters in North Carolina, we present two applications that indicate that neighborhoods have little or no effect on political behaviors. We estimate that neighborhood effects explain at most 1.9% of the total variation in individual turnout decisions, and 4% of the observed variation in party registration, after accounting for the race of households. In both applications, household-specific traits are a much more relevant factor determining those individual outcomes.

WORK IN PROGRESS

Sleeping with the Enemy: Dynamic Electoral Competition in a Clientelistic Environment (with Sergio Montero and Gabriel López-Moctezuma)

Pricing Partisanship: Evidence From North Carolina's Housing Markets (with Scott Abramson)

Building Weak Parties with Ambitious Politicians (with Sergio Montero and Gabriel López-Moctezuma)

Do More Representatives Improve Representation? Evidence from a Natural Experiment

​Do Programmatic Policies Induce Political Engagement?