Research

This paper investigates the incentive effects of no-confidence votes (NCVs) on governance and development. While NCVs are meant to promote accountability and better service delivery, in developing democracies they are often used strategically by national elites to undermine local rivals, incentivizing local leaders to prioritize elite interests over voter welfare. Under what conditions do NCVs enhance governance? Using data from Benin’s 2016–2017 wave of NCVs against mayors, I implement a difference-in-differences design combining electoral records, audit reports, and household surveys. The findings show that NCV motions—regardless of whether they succeed—lead to significant improvements in municipal spending (41%), provisions of goods (10%), and mobilization of internal revenue (31%). Voter welfare also moderately improves, with increases in employment and incomes in affected areas. The findings suggest that accountability mechanisms that enhance conflict of interests among politicians are crucial in aligning politicians' actions with voters' preferences, even when distorted.

No-confidence votes (NCVs) provide a sanctioning mechanism to hold democratically elected politicians accountable. However, national politicians can weaponize NCVs to target potential local rivals to deter future electoral challenge. This perverse form of accountability can undermine the quality of democratic competition, thus obviating the core goal of NCVs. Empirically, I evaluate this argument in the context of a developing democracy. During the 2016 presidential campaign in Benin, mayors who did not endorse the same presidential candidate as the MP from their district were two-thirds more likely to face an NCV. MPs who initiated NCVs against non-compliant mayors were more likely to win re-election. Finally, mayors who faced NCVs (even if not removed) were less likely to secure nomination. This evidence leverages electoral records, local audit reports, and original interviews with mayors in Benin to shed new light on an understudied dimension of democratic accountability.

Designing accountability systems for bureaucracies remains a key challenge for state capacity development. We carried out a randomized control trial at scale with local bureaucrats in Benin to investigate whether performance improves with deliberation on external evaluation reports.  In the treated municipalities, bureaucrats organized internal meetings to deliberate on issues highlighted in audit reports and surveys of bureaucrats and citizens. We find a positive treatment effect on the internal performance of bureaucrats (6.8 pp or 8.7% of the control group mean), while there is no immediate effect on service delivery. We investigated the mechanisms and found that the meetings helped bureaucrats update their beliefs about local government corruption, resulting in a fall in trust and cooperation between bureaucrats. These results suggest that in public sector bureaucracies in which the chance of collision is high (Tirole, 1986), an intervention that leads to a lower level of cooperation and trust may still improve performance.

This project documents how national leaders, lacking political backup, use fiscal intergovernmental transfers to build larger coalitions and consolidate power. Fiscal intergovernmental grants are the core operating and investment for local governments in developing democracies, and delays in the disbursements of these grants can undermine local governance, harm provision of public goods, and induce unfair electoral competition at the local level. I posit that national leaders can strategically induce variations in grant disbursements to co-opt opposition-held districts to expand ruling coalition bases and consolidate power. I leverage variations in presidential candidate endorsements by mayors during the 2016 presidential elections in Benin to estimate the effects of party affiliations on the timing of disbursements of intergovernmental grants, and consequently on vote shares for incumbent presidents in subsequent elections. Combining fiscal transfers and electoral data in a difference-in-differences design, I establish that municipalities whose mayors endorsed the losing candidate in 2016 and subsequently became opposition-district experienced shorter delays in intergovernmental grant disbursements and reduced electoral competition in the following presidential election. These effects are primarily driven by a substantial shift in support from the losing candidate to the incumbent president during the reelection campaign.

We investigate conditions under which anti-campaign failure and argue that both voters and politicians face dilemmas. While voters likely want to punish corrupt politicians, they also receive material benefits from potentially corrupt politicians due to poor economic conditions. Using a game-theoretical model, we theorize that anti-corruption campaigns can fail if voters have clientelist relationships with targeted politicians, and convicting them leads to backslash for campaign initiators. The main prediction of our model is that voters can effectively hold corrupt politicians accountable only when alternative sources of materials are provided; e.g., more jobs and/or better economic conditions. We design a field experiment in Ghana to test our model’s prediction by training economically vulnerable young people in entrepreneurship and providing them with grants to start a small business. We study the impact of our project's impacts on economic conditions, voting behaviors, and support for anti-corruption campaigns. 

Career choices are critical to understanding variations in labor-market outcomes. This paper investigates how early career choices vary with political connections in developing countries. We contend that labor-market outcomes depend on individual qualifications and ability for jobs available in an economy; however, political connections can substitute these determinants. Connections reduce costs associated with job search and improve job prospects for connected individuals. Anticipating this, individuals with such connections, we argue, choose careers that political connections can guarantee, which most likely are public-sector jobs. We test this argument by leveraging career choice data from pre-career individuals in Benin. Our empirical analysis yields two main results. First, politically connected individuals are likelier to choose top-tier careers in the public sector and less inclined to start a business. Second, connected individuals tend to drop less out of school and aspire to higher education. Our IV estimations suggest that these results are causal. Overall, the findings suggest that labor-market policies that substitute the insurance effects of connections can increase entrepreneurship through the spillover effects on education outcomes.

In sub-Saharan Africa, secondary school completion rates remain low, especially for girls. The gender gap persists even in countries where secondary education is free. This is not driven by parents’ lack of interest in investing in their daughter’s future: many of them pay for costly apprenticeships with, e.g., hairdressers, beauticians, and fashion designers. Anecdotal evidence suggests that parents expect higher returns from apprenticeships than from formal education, in terms of future employability. This creates a stark tradeoff, where girls miss out on acquiring formal human capital to acquire experience, contributing to the persistent gender gap in labor earnings. This project proposes to reduce this tradeoff by integrating vocational training into formal education in Benin. Specifically, we select five junior-secondary schools in two municipalities in northern Benin, where girls in grade 7th (first year of middle school in Benin) will attend, in addition to formal education, 1-2 hours per week on practical aspects of two vocational training, including fashion designer and hairdressers, for three months between February and April, 2026. We partner with ESPOIR POUR TOUS, an NGO that aims to improve girl education in Benin, and the Benin Ministry of Secondary Education and Vocational Training. The goal is to increase girls’ school participation and achievements while maintaining their ability to transition into the labor market.

In this project, we focus on gender imbalances in STEM fields, where the gap between girls and boys widens each year, limiting girls’ participation in computing and technology—key drivers of innovation and economic growth. The study will (1) introduce gender-responsive pedagogy (GRP) in STEM courses to reduce barriers and create an inclusive, stereotype-free learning environment and (2) organize weekly office hours for STEM courses to encourage girls’ participation and commitment to technology and computing sectors. We define GRP as a set of practices implemented through targeted teacher training sessions and ongoing pedagogical support to help STEM educators identify and mitigate gender-based disparities in their classrooms. As part of the intervention, Grade-8 girls will attend regularly scheduled Mathematics and Physics-Chemistry (MPC) classes with trained teachers in Benin. Additionally, girls in selected schools will participate in 20–30-minute weekly office hours (OH), during which high-achieving young women in MPC, serving as role models, will address clarification questions about class materials and homework assignments. We posit that the OH sessions will demystify STEM subjects, enhance understanding, and increase girls’ confidence in these disciplines. Together, the components of this intervention aim to reduce dropout rates, increase school completion, and help girls develop the critical-thinking skills essential to thrive in STEM fields. Our study aligns with global efforts to innovate teaching practices that respond to growing classroom heterogeneity and support success in STEM disciplines. While we are partnering with ESPOIR POUR TOUS and the Ministry of Secondary Education and Vocational Training in Benin, this project is pending IRB approval, with fieldwork expected to begin in 2026.