How do elections influence foreign aid spending? This is the primary question I seek to answer in this dissertation. There is considerable concern about the perverse effects of aid; however there are few tests of how electoral incentives influence these outcomes. Also, despite the fact that most aid recipient countries now hold elections for high office, we lack basic theory or evidence to explain how these elections influence aid spending. This dissertation fills these gaps. I argue that re-election pressures and information asymmetries between donors and recipients incentivize governments to manipulate aid spending in favor of key voters and patronage networks. These re-election pressures also compel donors to underinvest in the prevention of political capture. This manipulation and underinvestment undermines electoral competition and aid effectiveness, and helps to explain a number of puzzling effects of aid on democratization, political survival and corruption. This dissertation is divided into three stand-alone chapters. In Chapter One, I model the dilemma that donors face in trying to influence economic development and policy in developing democracies. To evaluate this model, I propose an original identification strategy that uses regional aid shocks to instrument for aid disbursements. The results confirm that aid has a positive effect on the probability that an incumbent is re-elected, particularly in cases where donors have strong policy interests. In Chapter Two I evaluate whether governments successfully influence the distribution of foreign aid in favor of strategically important voters. I create a subnational dataset containing the location of multilateral aid projects in Kenya from 1992-2011. I use these data to establish that incumbents have consistently manipulated aid spending in favor of co-partisans and co-ethnics. In Chapter Three, I discuss the implications of these distributional incentives for aid effectiveness. I argue that when aid projects are located within a government's core areas of support, governments have incentives to allow aid funds to be diverted for private gain rather than public good. Using donor evaluations of success and corruption, I confirm that aid projects distributed in co- ethnic and co-partisan areas are less effective and more corrupt

Abstract: Mobile communication technologies can provide citizens access to information that is tailored to their specific circumstances. Such technologies may therefore increase citizens’ ability to vote in line with their interests and hold politicians accountable. In a large-scale randomized controlled trial in Uganda (n = 16,083), we investigated whether citizens who receive private, timely, and individualized text messages by mobile phone about public services in their community punished or rewarded incumbents in local elections in line with the information. Respondents claimed to find the messages valuable and there is evidence that they briefly updated their beliefs based on the messages; however, the treatment did not cause increased votes for incumbents where public services were better than expected nor decreased votes where public services were worse than anticipated. The considerable knowledge gaps among citizens identified in this study indicate potential for communication technologies to effectively share civic information. Yet the findings imply that when the attribution of public service outcomes is difficult, even individualized information is unlikely to affect voting behavior.