working papers

The Electoral Consequences of Cellphone Coverage Expansion. (with Shuning Ge, Katrina Kosec, Apoorva Lal, Benjamin Laughlin)
Revise & Resubmit at Political Science Research and Methods, 2023. drawing

Abstract Using the case of Ghana, we analyze the electoral effects of cellphone coverage expansion in a developing country setting. We construct constituency-level panel data of electoral results for Ghana’s six general elections during 1996--2016 and combine these with high-resolution geocoded annual coverage information using a difference-in-differences design. We find that cellphone coverage benefited incumbents in both presidential and parliamentary elections. This effect appears to be due to cellphone coverage expansion improving both citizens' wealth and faith in the economy as opposed to general perceptions of government performance, political knowledge, or vote-buying. The results highlight the potential for even basic information and communications technologies to make retrospective voting more common in settings where ethnic voting, clientelism, and vote-buying are common.

Do More Disaggregated Electoral Results Deter Aggregation Fraud? (with Miguel Rueda, Shuning Ge)
2023. drawing

Abstract It has been argued that the level at which electoral results are published can affect the election integrity. Publishing more granular results (e.g., at the polling station level) can allow citizens to verify the vote totals that determine election outcomes, thereby deterring voting aggregation fraud. While this logic undergirds the recommendations of international organizations monitoring elections to publish disaggregated electoral results, to date there have not been systematic assessments of how variation in aggregation is linked to electoral miscounting. We address this lacuna by constructing a new dataset of the level at which electoral results are reported in 125 low- and middle-income countries since 2000. We find a clear negative association between the granularity of published electoral results and perceptions of counting irregularities. We do not observe substitution of election malpractice: increased granularity is not linked to perceptions of other forms of manipulation like electoral violence, opposition harassment, or clientelism.

Can Community Policing Improve Police - Community Relations in an Authoritarian Regime? (with Rob Blair, Anna Wilke)
2023. drawing

Abstract Throughout the developing world, citizens distrust the police and hesitate to bring crimes to their attention — a suboptimal equilibrium that makes it difficult for the police to effectively combat crime and violence. Community policing has been touted as one solution to this problem, but evidence on its efficacy in developing country contexts is sparse. We present results from a large-scale field experiment that randomly assigned a home-grown community policing intervention to police stations throughout rural Uganda. Drawing on administrative crime data and close to 4,000 interviews with citizens, police officers, and local authorities, we show that community policing had limited effects on core outcomes such as crime, insecurity, and perceptions of the police. We attribute these findings to a combination of turnover, treatment non-compliance, and resource constraints. Our study draws attention to the limits of community policing’s potential to reduce crime and build trust in the developing world.

When Refugee Exposure Increases Incumbent Support through Development: Evidence from Uganda. (with Yang-Yang Zhou)
2022. drawing drawing

Abstract In higher-income democracies, studies have found that refugee shocks cause voters to punish incumbents and turn to far-right parties. Yet there is a dearth of studies on the electoral consequences of refugee-hosting in low-income countries, where most refugees reside. Combining information on refugee settlements with four waves of national elections data in Uganda, we find that a one standard deviation increase in refugee presence leads to a 7.4 percentage point increase in incumbent support. Original longitudinal data on healthcare, schools, and roads coupled with national survey data suggest that the mechanism is positive externalities of refugee-hosting on local public goods.