How Transparency Affects Distributional Politics (with Brigitte Seim). Manuscript

Supplementary Information
Pre-Analysis Plan

How does transparency affect distributional politics? We theorize that it conditions how officials choose recipient communities, compelling them to allocate to needy communities rather than to core supporters. We present the results of a field experiment in which 333 elected incumbent councillors in Malawi made real and meaningful decisions about the allocation of NGO-provided development goods to schools in their constituency. Prior to allocating goods, half of the incumbents were informed that letters about their decisions would be sent to local development oversight committees. We find that this transparency treatment caused incumbents to allocate goods to recipient school communities with greater economic need. They were also less likely to allocate to schools with strong political support. To our knowledge, this is the first experimental evaluation of theoretical claims about the role of transparency in distributional politics using in-office elected leaders as participants and observing real distributional decisions.


How Violence and Fraud Neutralize Information Effects on Turnout: A Field Experiment in Uganda (with Mark Butaine, Daniel Nielson and Paula Pickering). Manuscript.

Email for latest draft
Pre-Analysis Plan
Information about corruption should motivate voters to show up at the polls to “throw the bums out.” However, this effect presupposes that ballots remains secret and that voters face a low risk of violence at polling places. Yet in electoral autocracies the accountability relationship can break down, since showing up to vote in support of opposition candidates may place a voter at risk of government repercussions, especially if votes are not expected to be private. Where violence and privacy violations are anticipated, exposure to information about corruption may have little effect on turnout. We evaluate this theory using a large randomized­ controlled trial(n=16,083) conducted in Uganda for the 2016 district elections. We treated eligible voters with factual, non­partisan information about irregularities in the management of local government budgets using SMS messaging. We find that this information caused the expected turnout effects, but only in the absence of nearby electoral violence and for voters who had relatively high expectations that their votes would remain secret. These results imply that electoral violence and privacy violations are key impediments to voters’ seeking accountability from their elected officials at the polls and suggest that repressive tactics may be particularly useful for incumbents when corruption is rampant.


How Foreign Aid Affects Election Outcomes. Manuscript.

Email for latest draft
Foreign aid often appears to contribute to leader survival, corruption, patronage, and poor governance; yet we know little about when and why these effects occur. I propose a formal model to explain when governments are successful at using aid for political ends. The model illustrates the strategic interaction between governments interested in winning an upcoming election, and a donor interested in maximizing the effectiveness of an aid project. I show that the ability of donors to prevent the political capture of aid depends upon the policy objectives of the donor and the level of political competition in the recipient state. I confirm these propositions by evaluating the effect of aid spending on the outcome of all developing country elections from 1960 to 2011. To assess causation, I instrument for aid spending using regional aid shocks. The results confirm that aid has a positive effect on the probability an incumbent is re-elected, especially when donors have political interests in the recipient state.


How Information Affects Distributional Politics: An RCT with Elected Officials. In preparation.

Email for latest draft
Most models of distributional politics assume that governments have the capacity to make efficient decisions about the allocation of development; however in many cases politicians lack crucial information about the needs, politics and existing resources of their constituencies, particularly when newly elected. This information gap can potentially create biases in the distribution of public resources, and may reinforce clientelism and pork barrel politics. Additionally, information about the distributional effectiveness is crucial for the proper functioning of voter accountability mechanisms. In this article we describe the results of a unique field experiment among 550 MPs and councillors in Malawi in which we test how constituency information changes distributional decisions. In partnership with a NGO, we provide elected councillors and members of parliament with the opportunity to decide which schools in their ward or constituency should receive development materials. We then randomly provided information relevant to effective decision making in order to examine the causal effects of information on distributional decisions. We include three different types of information, which we hypothesize play important roles in allocation decisions: information about the official’s electoral support in the vicinity of the schools; information about the development and economic needs of the schools; and information about pre-existing resources provided by international donors. We also evaluate the treatment effects among subgroups of competitive and non-competitive wards and constituencies, among female and male officials, and among officials with more or less informed priors about constituency characteristics. Contrary to usual assumptions about distributional politics, we document considerable gaps and errors in politicians’ pre-existing information about the needs, politics and resources within their constituency. We show how information provision can play an important role in improving politicians’ decision-making capacity, as well as addressing inequities and incentives for clientelism and corruption. Our study is unique in that we evaluate allocation decisions within the context of actual distributional decision making, and that we involve elected officials who are regularly entrusted with making such decisions. In addition to offering new insights into development and inequality, the results of this field experiment will help inform policy debates about how and when local officials can best be entrusted to make efficient decisions about development aid.


How Foreign Aid Affects Election Outcomes. Manuscript.

Email for latest draft
Foreign aid often appears to contribute to leader survival, corruption, patronage, and poor governance; yet we know little about when and why these effects occur. I propose a formal model to explain when governments are successful at using aid for political ends. The model illustrates the strategic interaction between governments interested in winning an upcoming election, and a donor interested in maximizing the effectiveness of an aid project. I show that the ability of donors to prevent the political capture of aid depends upon the policy objectives of the donor and the level of political competition in the recipient state. I confirm these propositions by evaluating the effect of aid spending on the outcome of all developing country elections from 1960 to 2011. To assess causation, I instrument for aid spending using regional aid shocks. The results confirm that aid has a positive effect on the probability an incumbent is re-elected, especially when donors have political interests in the recipient state.


Understanding the Downstream Effects of Aid on Political Perceptions and Behavior. Experiment in progress.

Pre-Analysis Plan
This experiment is an evaluation of the political downstream effects of the allocation of foreign development project aid. In 2015, the we executed a field experiment examining the allocation of NGO-provisioned school aid by local councillors in Malawi (Jablonski and Zimmerman 2017). This earlier project resulted in the constrained and randomized allocation of school aid and provides us with a unique opportunity to evaluate the causal effects of aid on perceptions of government, voting, and beliefs about government and NGO performance. This study will also allow us to speak to one of the more hotly contested questions in the politics of aid: Do voters respond to donor-branded development goods by supporting the incumbent government? Or does the provision of such goods promote ambivalence or criticism about government capacity? This question remains contested in part, because of the challenges of studying aid and aid branding in a non-experimental setting, and the associated difficulties of attributing causation.