The mission of development economics is, fundamentally, to explain why some communities are wealthier than others. This dissertation is a collection of two essays that make progress towards fitting together key pieces of the puzzle. My hope is that the findings will inform, not only researchers, but policymakers working to accelerate economic development.

To that end, the essays make three thematic contributions. First, they leverage data and experiments from the Middle East and North Africa (MENA). While MENA’s relevance to global markets is self-evident, to date economists have paid it less attention than other low- and middle-income regions. Second, the results advance the economic geography of development by exploring trade and forced migration. Third, the results advance the political economy of development by exploring corruption and the victims of conflict. These geographic and political factors are among the most important drivers of structural transformation, productivity growth, and poverty reduction.

In Chapter 1, joint with Nate Grubman and Jawaher Mbarek, I investigate corruption dynamics in international trade. Every year emerging markets import goods valued at more than $7 trillion, and in many countries shipments have to pass through corrupt customs administrations. Given these high stakes, policymakers require a deep understanding of both the causes and the effects of customs fraud. In addition, trade corruption can serve as a laboratory to study corruption writ large. One previously unexplored complexity is that bribe payers and bribe receivers often have repeated interactions. Given corruption’s characteristic counterparty risks and information asymmetries (not to mention the impossibility of contract enforcement), these long-running relationships likely matter for a wide variety of outcomes across a wide variety of contexts.

To pursue these general learning objectives, we overcome the data and identification challenges inherent to investigating bribery: we build an original dataset on Tunisian customs transactions using an audit study to directly observe bribes, and we leverage a natural experiment in which a computer algorithm randomly assigns customs officials to import shipments. There are three sets of results. First, we show that bribery and tax evasion are widespread, that bribery is collusive (not coercive), and that age (but not gender) predicts officials’ corruptibility. Second, in line with a straightforward Nash Bargaining model, we show that the length of official/trader relationships increases tax evasion but decreases bribe amounts. Third, we zoom out to consider the larger macroeconomic implications and show that, in terms of lost tax revenue, bribery costs the Tunisian government 0.7% of GDP or $80 per citizen.

In Chapter 2, joint with Edward Miguel, Sandra V. Rozo, Emma Smith, and Sarah Stillman, I investigate affordable housing for forced migrants. There are 82 million forced migrants in the world today, and the vast majority reside in “host communities:” they live, work, and attend school in the same neighborhoods as the citizens of their host country. As both a cause and an effect of this arrangement, many governments, IGOs, and NGOs support these refugees’ shelter through either direct provision or rental assistance. However, the evidence on the welfare implications of housing subsidies is inconclusive for refugees in particular and for high-poverty populations in general. On the one hand, studies have shown that shelter assistance has the potential to provide, not just a roof overhead, but a number of downstream benefits. On the other hand, some studies have found null or even negative results, and it is often unclear whether the positive results are due to direct housing-quality effects or indirect neighborhood effects.

To fill this knowledge gap, we ran a randomized controlled trial (RCT) testing a program that provides housing support to Jordan’s Syrian refugees. The subsidies are large (approximately $190 per month) and are not vouchers; households can only use them for their current housing, allowing us to be among the first to identify housing-quality effects absent confounding neighborhood effects. There are three sets of results. First, we estimate the positive impacts of the program. Recipients experience improvements in their living situations, as well as better education and credit-market outcomes. Second, we observe that, while the program does not cause movement of households, it does cause movement between households. Recipients welcome new individuals into their homes, likely so that family, friends, and neighbors can take advantage of the upgraded housing. Third, we estimate the negative impacts of the program and argue that these unintended consequences likely follow from the endogenous household formation. For example, recipients are less food secure, consistent with having more mouths to feed.

Together, these studies have generalizable, actionable insights. Knowing that dynamics matter for corruption can help reform-minded governments decide which levers to pull as they attempt to enhance transparency. And the large economic cost of trade corruption implies a high return on investment for the governments that succeed. Our RCT reveals both the upside potential and the downside risk of large-scale affordable-housing interventions. By refining their design, implementation, and cost-effectiveness, policymakers could improve these programs and the lives of the vulnerable communities they serve.

When trying to understand political change in democracies it is natural to think of voters as the driving force. But much political change—including the elections of Donald Trump and Boris Johnson—is as much a product of dynamics happening within existing parties—

enabling such candidates to come close to power—as it is a product of voters voting for those parties. A complete understanding of political change thus requires understanding the behavior and role not only of voters, but also of parties.

This thesis studies political change and the roles of voters and parties in driving such change. In the first chapter I study the political consequences across and within parties of refugee migration. I combine Danish administrative data on local politicians from 1993-2013 with quasi-random refugee placements to show that refugee migration does not increase voters’ support for the far right. Instead, it increases the propensity for candidates from lower social backgrounds to be elected. This effect is driven both by parties and voters: First, parties place candidates from lower social backgrounds higher on lists and switch to a party system, where individual candidate popularity, rather than party list position, determines election. Second, voters cast individual votes for these candidates from lower social backgrounds. In a nationally representative survey experiment I show that voters’ preference for candidates from lower social backgrounds can in part be explained by refugee migration increasing their preferences for redistribution towards natives, preferences that are mirrored by candidates from lower social backgrounds.

In the second chapter of my thesis, I further explore the role of parties in responding to and driving political change. Specifically, I derive a theoretical framework to study how changes in the ideological bias favoring a party affect the propensity to run an open list, and how this in turn affects their support from voters. I find that parties run closed lists if the ideological bias is small. However, when voters become sufficiently biased towards one party, that party opens up its list. This switch to open lists in turn increases their support from voters beyond the effect of the ideological bias. The reason for this is that candidates in popular parties exert more effort in open than closed lists. These results imply that large parties are more powerful in proportional representation systems where parties can choose their list structure compared to closed list systems.

Overall, this dissertation shows that when trying to understand political change, it is important to consider the role not only of voters, but also of parties in driving such change.

Corruption and political favoritism are considered major impediments to economic development. Although there is a growing consensus about the adverse efficiency consequences of corruption we still have a limited understanding of how corruption is shaped by political and economic institutions and how it affects our democracies. An increasing literature documents political favoritism and its welfare consequences relative to a no misallocation benchmark. In my dissertation, I complement this line of research by quantifying the effects of favoritism relative to relevant policy counterfactuals. My work highlights the importance of transparency and limiting regulatory discretion in improving the efficiency of public spending.

In the first chapter, I investigate the determinants and consequences of increasing a buyer's discretion in public procurement. I study the role of discretion in the context of a Hungarian policy reform which removed the obligation of using an open auction for contracts under a certain anticipated value. Below this threshold, buyers can use an alternative "high-discretion" procedure to purchase goods and services. At the threshold, I document large discontinuities in procurement outcomes, but I also find a discontinuity in the density of anticipated contract value, indicating that public agencies set contract values strategically to avoid auctions. To distinguish the causal effects of increased discretion from the self-selection of agencies into high-discretion procedures, I exploit the time variation of the policy reform. I find that discretion increases the price of contracts and decreases the productivity of contractors. To dig deeper into the motivations of public agencies, I use a structural model to identify discretion's impact on rents from corruption. I also use the same structural approach to simulate the effect of alternative value thresholds. I find that the actual threshold redistributes about 2 percent of the total contract value from taxpayers to firms and decreases the average productivity of contractors by approximately 1.6 percent. My simulations suggest that the optimal threshold would be about a third of the actual.

Moreover, case studies suggest that in addition to rent extraction corruption provides opportunities to buy political support in weakly institutionalized democracies (e.g. McMillan and Zoido (2004)). Consequently, detrimental effects of political favoritism may not be limited to misallocation of public resources but also constrain governmental accountability. In the second chapter, my coauthor Adam Szeidl and I confirm this conclusion by investigating political favoritism in the Hungarian media market. We scrutinize three different markets, printed media, billboards, and online newspapers. We establish three main results about favoritism in the Hungarian media. First, we document distortive two-way favors between politicians and the media, in the form of government advertising and media coverage. For both directions of favors, our empirical strategy is to compare the allocations of actors with changing versus unchanging connection status. More specifically we compare advertising behavior of state-owned and private companies and media content of outlets with more and less political connections. Since friendly news coverage systematically moves together with advertising favors we interpret our findings as media capture. Second, we document an organizational change in favoritism: a first phase when favored media was controlled by a single connected investor; a second phase when this relationship broke down and two-way favors were terminated; and a third phase when control of newly favored media was divided between multiple connected investors. Our preferred interpretation is that governments with more de-jure power shift the organization of favors towards a divide-and-rule style arrangement. Third, we develop and implement a portable structural approach to measure the economic cost of misallocative favoritism.

This paper examines the extent to which access to information enhances political accountability. Based upon the results of Brazil’s recent anti-corruption program that randomly audits municipal expenditures of federally-transferred funds, it estimates the effect of the disclosure of local government corruption practices upon the re-election success of incumbent mayors in municipal elections. Comparing municipalities which were randomly audited before the elections with those audited after, the analysis shows that the disclosure of audit results had a significant impact on the re-election rates of mayors found to be corrupt. For a one standard deviation increase in reported corruption, the audit policy reduced the incumbent’s likelihood of re-election by 25 percent. This effect is more pronounced in municipalities where radio stations are present and higher levels of corruption are identified. These findings highlight the value of information and the role of the media in reducing informational asymmetries in the political process, thus enabling voters to not only hold corrupt politicians accountable but also to reward non-corrupt politicians.

This dissertation studies the role of political institutions in curbing rent-seeking and corruption. In each of the three chapters, I describe and analyze how economic incentives and institutions affect the behavior of elected officials and, ultimately, have an impact on policy choices and welfare. In the first chapter, Interest Groups, Campaign Finance and Policy Influence: Evidence from the U.S. Congress, I study how the financing of political campaigns by special interest groups distorts legislative voting in the United States. In the second chapter, Do Government Audits Reduce Corruption? Evidence from Exposing Corrupt Politicians, we study how randomized audits of government resources can reduce corruption by enhancing political and judicial accountability. In the third chapter, titled Money and Politics: The Effects of Campaign Spending Limits on Political Entry and Competition, we study the effects of the imposition of limits to campaign spending on political competition.

In Interest Groups, Campaign Finance and Policy Influence: Evidence from the U.S. Congress, I study the effects of special interest group contributions on legislative voting. Empirical research on this question has led to mixed results (Ansolabehere et al., 2003). The main empirical challenge to establishing this relationship is that it is difficult to disentangle the effects of ideology and special interest groups in determining the behavior of legislators. In this paper, I address these concerns by exploiting a new comprehensive dataset of interest group positions on bills and by improving upon the identification strategies of previous studies. To identify the effects of interest group contributions, I create instruments which are based on the interaction of national industry shocks with historical connections between interest groups and politicians. I find a consistent, robust reduced-form relationship between interest group contributions and legislative behavior. I then develop and estimate a structural model which allows me to further unpack how ideology and interest groups interact in determining the votes of legislators. I find that more ideological (Republican) legislators tend to be less influenceable by contributions, highlighting the possibility that ideology can act as a commitment to voters against special interests. I also use my estimates to study the effects of a counterfactual policy change which would ban contributions from special interest groups. I find that several bills would have failed to pass under the counterfactual policy regime, including significant bills allocating several hundred billions of dollars.

Corruption and rent-seeking are major impediments to economic development worldwide (Rose-Ackerman, 1999). In the second chapter, I study how institutions play a central role in curbing corruption in the context of a developing country. In Do Government Audits Reduce Corruption? Evidence from Exposing Corrupt Politicians, written with Claudio Ferraz and Fred Finan, we study the extent to which government audits of public resources can reduce corruption by improving the selection of politicians and by enhancing accountability. We exploit a Brazilian anti-corruption program, which randomly audits municipalities for their use of federal funds. To differentiate whether the effects of the audits operate through a political selection or discipline channel, we develop and estimate a career concerns model of political accountability (Holmstrom, 1999; Persson and Tabellini, 2002). In the model, audits affect the beliefs of mayors about the costs of engaging in corruption and provide information about mayors to voters before elections. We exploit the randomized nature of audits together with the different incentives facing first-term versus term-limited (second-term) mayors to separately identify the effects of political selection, electoral discipline and legal discipline. We find that most of the reduction in corruption is due to a legal disciplining effect, as audits primarily increase the perceived non-electoral costs of engaging in corruption.

Beyond the use of randomized audits of public resources, policymakers have also focused efforts on reforming campaign finance laws to curb rent-seeking and improve economic outcomes. In Money and Politics: The Effects of Campaign Spending Limits on Political Entry and Competition, written jointly with Claudio Ferraz, Fred Finan and Carlos Varjao, we study the effects of campaign spending limits on political entry and competition. Although campaign spending limits are one of the most common used policy tools worldwide in curbing the role of money in politics, there is little causal evidence on their effects (with Milligan and Rekkas (2008) and Fouirnaies (2018) being the only exceptions). In this chapter, we exploit a reform in Brazil which imposed spending limits on mayoral elections. These limits were implemented with a sharp, unexpected discontinuity which was the result of the use of different formulas to calculate inflation rates. Using a regression-discontinuity design, we find that stricter spending limits increase political competition. Stricter limits create a larger pool of candidates running for office, which is on average less wealthy. Moreover, stricter limits reduce the likelihood that incumbent mayors are reelected and leads to the election of less wealthy mayors that rely less on self-financing for their campaigns. However, in municipalities with stricter spending limits, we find that mayors are less successful in obtaining federal block grants. These findings were found to be consistent with the predictions of a contest model with endogenous entry of candidates.

This dissertation studies the impacts of quid pro quo relationships between politiciansand their supporters on public policy and the labor market of public servants. Chapter 2 investigates the effects of a temporary shock in the likelihood of detecting officeholders’ misbehavior on the composition of public expenditures and reelection rates of incumbent politicians, taking into consideration the existence of an ongoing relationship between lawmakers and campaign contributors. This shock derives from an experiment developed by Zamboni and Litschig (2018) that temporarily and randomly increased the likelihood of selection of Brazilian municipalities into a preexisting audit program against corruption. I document a reduction in the probability that treated municipalities engaged in construction projects during the experiment, with a partial reversal in the reelection year, as the audit risk returned to its baseline mark. I report that the threat of audit mostly impacted the actions of politicians in locations with previous experience in the anti-coruption program and where mayors had close relationships with historically corrupt campaign donors, such as construction companies. Moreover, I estimate a large contraction in the likelihood of contribution from these donors to the reelection campaigns of treated mayors and a sizeable reduction in their reelection rates. Chapter 3, co-authored with Thiago Scot, analyzes the effects of a change in power at the local level on the employment of health care workers in Brazil. It compares municipalities where the incumbent party barely won a bid for reelection with those where the incumbent party barely lost, in order to estimate the effects of political turnover on the labor market for public servants. It reports a larger rate of termination of employment contracts in municipalities affected by these shocks, in comparison to the control group. It shows that employees are more likely to leave the health care labor force permanently after political turnover and that public employees on unstable career tracks drive most of the disruption in the employment contracts.

This dissertation contains three chapters that examine development and urbanization in China since 1980. The first chapter investigates the impact of Special Economic Zones established since 1992 on local economic outcomes and urbanization rates. The second chapter studies local bureaucrats’ incentives while balancing the goals of economic growth and resource or environmental protection. The last chapter examines patterns of urbanization in China as related to the administrative hierarchy between urban and rural regions.

In the first chapter, I examine the local and aggregate productivity impacts of Chinas Special Economic Zones (SEZ) program from the 1990s to 2006. Using an event study de- sign where I exploit variation in the timing of the first zone establishment across counties, I find that an SEZ increased GDP by 1% to 2% per year over 5 years. In order to assess the aggregate impact of the program, I develop a spatial equilibrium model with two sectors – agriculture and manufacturing. In the model, the economy-wide output impact of an SEZ depends on the elasticities of inter-regional and inter-sectoral labor supply, as well as the relative productivity levels of the SEZ-hosting regions and other regions. I do not find any evidence that the SEZ program induced inter-regional labor reallocation. However, the program did accelerate the shift of employment from the agricultural sector.

In the second chapter, I explore whether Chinese bureaucrats’ policy decisions depend on connections with their superiors through the appointment of bureaucrats to office. In particular, I consider appointment as a kind act on the part of the promoter, and ask whether the promoted bureaucrat enacts policies that are in the interest of his benefactor. This framework is applied to the policy outcome of rural land conversion and air pollution. I examine whether rural land conversion is more prevalent, and whether air pollution is more abundant, when the local bureaucrat works under the Party Secretary who promoted him. My analysis exploits variation in the length of politicians terms within individual and pairs of politicians. Preliminary findings suggest that promoted land bureaucrats may exhibit reciprocal behavior, while environmental protection bureaucrats may not.

In the last chapter, I study the impact on urbanization of the Cities Leading Counties (CLC) policy implemented in China beginning in 1983. This nationwide program granted large metropolitan governments administrative control over counties in the surrounding

hinterland. I use annual county-level nighttime lights data to examine changes in urbanization in the hinterland after integration with a city. I find that while the policy increased urbanization in counties subsumed into cities, the cities themselves did not experience faster urbanization.

This dissertation consists of three contributions to the literature at the intersection of development economics and public economics, set in the context of Uganda but with broader applications. In each of the three chapters, I use rigorous identification and a rich variety of administrative data to answer interesting and challenging questions about a variety of ways in which the public sector influences and is influenced by economic development.

The first chapter studies the subnational fiscal response to international aid, and examines how other donors and the Ugandan government respond to exogenous increases in the amount of World Bank funds allocated to local projects. While similar analysis has been done before at the country level, my analysis is the first that is focused on the subnational level. I find evidence of a ``crowding in'' of donors, whereby some other donors follow the lead of the World Bank in determining where to allocate funding across Uganda's smaller administrative units. The crowding in of aid, and the subsequent potential for overspending and diminishing marginal returns, may help explain why international aid sees large returns viewed locally, but smaller returns on national levels.

The second chapter focuses on a randomized control trial which I conducted with the Uganda Revenue Authority (URA), studying tax compliance in the context of low state capacity. Although the ability of the state to collect tax revenue is crucial for development, relatively little is known about what drives tax compliance and how it can be improved in poor, low-capacity countries. I show that low-cost, easily implementable tax compliance interventions can succeed, testing a highly effective tax encouragement scheme in conjunction with URA. I find a 6x rate of return to sending a simple text message to prospective taxpayers in the days before taxes are due, which increases to 13x when considering an enforcement-focused message; this intervention is estimated to have raised over \$12,500 USD. To further understand the levers of tax compliance, I build an extensive, granular, nationwide dataset of public good provision, and utilize machine learning tools to estimate state capacity using this dataset; I then examine the heterogeneity of the treatment effect across different levels of state capacity. I find that the treatment was most effective in areas where state capacity is low, and particularly in areas where state capacity was low and there have been recent investments in public goods. These results suggest that while deterrence-based methods may be most effective across the board, there is a crucial role for fiscal exchange in encouraging tax compliance in low-capacity environments.

The third chapter focuses on the effects of administrative unit proliferation, or the creation of new subnational administrative units, on the provision of governmental services and the processes of economic development. I utilize a variety of rich data sources on budgets, educational personnel and infrastructure, household well-being, economic activity, political attitudes and voting behavior in a differences-in-differences framework to examine the effects of the 2009-2010 wave of district creation in Uganda. I compare parent districts, which inherit the administrative capacity of their predecessors; newly formed districts, which must build administrative capacity from the ground up; and non-splitting districts, which are unchanged, while controlling for a variety of underlying static and temporally varying characteristics. I find that parent districts benefit from the split in terms of service provision and medium-run economic development, suggesting that smaller districts may have been better able to respond adequately to the needs of the population. However, new districts benefit from the split only in terms of service provision and not in terms of economic development, suggesting that a lack of adminstrative capacity may have hampered their internal growth. These findings suggest that even under favorable conditions for the creation of new districts, a lack of adequate staffing capability may hinder the benefits of decentralization.

Jointly, these three papers address three topics crucial to the fields of both development and public economics: fiscal response, taxation, and decisions on governmental size. In other words, this dissertation addresses how countries raise funding, both internally and externally, and where and how capably they spend it. Although I study these questions in the context of Uganda, all three papers have broader implications, whether on the dynamics of international aid, the importance of state capacity to taxation (and vice versa), or the dynamics of economic development under insufficient administrative capacity.

This dissertation combines three projects, each analyzing how markets or individuals respond to different policy or environmental changes in low and middle income countries. Each study theoretically motivates the hypotheses of interest and combines novel data from a wide range of sources to test them.

The first chapter examines the role of prices of foods, rich in different nutrients in the obesity and diet-related disease epidemic in Mexico. In response to the growing epidemic of obesity and diet-related chronic diseases, a number of governments are proposing taxes designed to reduce the consumption of unhealthy foods and thereby improve health outcomes. In this paper, I provide the first estimates of the effects of price changes of foods rich in sugar on the prevalence of obesity and diet-related chronic diseases, such as diabetes and hypertension. The analysis is made possible by rich longitudinal and nationally representative micro data on food prices and objective measures of health outcomes in Mexico for the period 1996-2010. I employ a unique bar-coded level price dataset with product-specific nutritional information combined with two datasets on health outcomes: (1) a state-level administrative dataset and (2) an individual panel dataset. Exploiting plausibly exogenous within-state variation in prices over time, I show that a decrease in the price of sugar-rich foods significantly increases the prevalence of abdominal obesity, type 2 diabetes, and hypertension. In addition, the least healthy and most impatient individuals seem to be more responsive to price changes, suggesting that time preferences are an important mechanism driving the results. Overall, the effect of sugar prices on the incidence of chronic diseases is large. Since the signing of NAFTA, I estimate that the reduction in prices of sugar-rich foods explains 20 percent of the increase in diabetes.

The second chapter, joint work with Paul J. Gertler, identifies how junk food and soda taxes are passed through to consumer prices. I use a unique bar-code level price dataset with product-specific nutritional information and the introduction of junk food and soda taxes in Mexico in January 2014 to assess whether and how the excess tax burden is split between consumers and producers. Preliminary results suggests that pass-through is the strongest for sodas, followed by snacks, candy and cakes. For these products a full pass-through materializes within six months after the introduction of the tax. On average, the tax is not fully passed through to consumer prices in the case of cereal and cookies. However, the pass through is increasing over time for the latter, reaching a full pass-through by May 2014. We also observe large variation in pass-through across cities - from over to under shifting; hence observing decreases in consumer prices as well. Using price elasticities of health from Gracner (2015), a one time introduction of junk food and soda taxes in Mexico would imply approximately one and a half percentage point decrease in abdominal obesity and between a quarter to one half of a percentage point decrease in type II diabetes prevalence, under the assumption of a symmetric response to a price change.

The third chapter, based on joint work with Paul J. Gertler, Marco Gonzalez-Navarro, and Alex D. Rothenberg, provides evidence of the effects of road quality on local economic activity using temporal variation generated by maintenance investments in roads. A long panel of firms and households allows us to shed light on the effects of road quality for pre-existing households and firms. Methodologically, we propose a new road quality instrument using a nationwide panel dataset of road surface roughness to predict road quality from temporal variation in budgets exogenously allocated to different road maintenance authorities. We first show that higher road network quality improves household consumption and income. We then show that this is partly due to job creation in the manufacturing sector. Third, we show evidence of an occupational shift from agriculture into manufacturing and higher profits for those who stay in agriculture. The gap in average income between agriculture and manufacturing employment is reduced with road quality but not eliminated. Because wages in the manufacturing sector do not change with road quality the results are consistent with dual labor markets in Indonesia.