research

Job Market Paper

1. Incentives, Management, and Public Service Delivery: Evidence from a large field experiment in Peru

   Governments often struggle to improve service quality when critical dimensions of effort are difficult to observe. In decentralized systems, central authorities can verify service provision but not the quality of frontline interactions, creating a classic asymmetric information problem. This paper examines whether incentive design can address this challenge by strengthening management practices. I study the effects of two institutional incentive schemes on the implementation and effectiveness of a municipal service delivery intervention aimed at reducing childhood anemia in Peru. Specifically, I compare a standard input-based contract with a mixed incentive contract that conditions rewards on both inputs and outcomes. Leveraging a clustered randomized controlled trial across 539 rural municipalities, I estimate the impact of these contracts on managerial practices, health service utilization, and child health outcomes. Municipalities assigned to the mixed contract increased managerial intensity: they strengthened monitoring, adopted new coordination practices with health facilities, and more actively motivated frontline workers. These changes led to higher utilization of preventive health services, increased uptake of iron supplementation, and greater engagement with the health system. However, despite these improvements in implementation, the intervention did not generate short-run reductions in childhood anemia.

   Presented at: Workshop El mundo rural en América Latina y el Caribe - CAF (2025), LACEA (2025), APPAM (2025), NEUDC (2025), EHEW (2026), AFE (2026), LACEA (2026*).

   Funded by: CAF, The Weiss Fund, JPAL Governance Initiative, Columbia University.

   Governments often struggle to improve service quality when critical dimensions of effort are difficult to observe. In decentralized systems, central authorities can verify service provision but not the quality of frontline interactions, creating a classic asymmetric information problem. This paper examines whether incentive design can address this challenge by strengthening management practices. I study the effects of two institutional incentive schemes on the implementation and effectiveness of a municipal service delivery intervention aimed at reducing childhood anemia in Peru. Specifically, I compare a standard input-based contract with a mixed incentive contract that conditions rewards on both inputs and outcomes. Leveraging a clustered randomized controlled trial across 539 rural municipalities, I estimate the impact of these contracts on managerial practices, health service utilization, and child health outcomes. Municipalities assigned to the mixed contract increased managerial intensity: they strengthened monitoring, adopted new coordination practices with health facilities, and more actively motivated frontline workers. These changes led to higher utilization of preventive health services, increased uptake of iron supplementation, and greater engagement with the health system. However, despite these improvements in implementation, the intervention did not generate short-run reductions in childhood anemia.

Working Papers

1. Gender-Based Violence in Schools and Girls’ Education: Experimental Evidence from Mozambique

   With Sofia Amaral , Aixa Garcia , Selim Gulesci , Alejandra Ramos , Micaela Sviatschi

   Status: Submitted, under review.

   School-related gender-based violence (GBV) is pervasive, yet evidence on effective interventions to curb violence is limited, particularly in settings where perpetrators face weak accountability and victims and bystanders have limited agency for reporting. We evaluate a policy in Mozambique that combined teacher training to strengthen schools’ capacity to respond to GBV with guided student discussion sessions to increase awareness around violence and reporting of abuse. To examine differential effects and complementarities across genders, student training was randomly targeted to girls, boys, or both. The intervention reduced sexual violence perpetrated by teachers against girls by 67 percent in all treated schools, regardless of the gender of students trained. Yet when girls received the student training, girls’ school enrollment increased by 10 percent. The evidence is consistent with victim engagement playing a role in translating reductions in violence into educational gains: training girls increased reporting of GBV to trained school staff, who were more likely to formally report cases to school authorities. Administrative data from the national child helpline further show increases in GBV-related calls and investigations after our intervention. Our findings highlight the importance of institutional accountability in constraining abuse by powerful actors and of engaging victims in reporting mechanisms to translate safer school environments into human capital investment.

   Media/Policy Mentions: VoxDev, VoxUNICEFDev, Development Impact WB Blog.

   Presented at: PacDev; UNU-WIDER; RIDGE.

   Funded by: UNICEF, JPAL Post Primary Education Initiative, Princeton University, Weiss Foundation, Leibniz Association, GIZ.

   DOI Bib PDF

   School-related gender-based violence (GBV) is pervasive, yet evidence on effective interventions to curb violence is limited, particularly in settings where perpetrators face weak accountability and victims and bystanders have limited agency for reporting. We evaluate a policy in Mozambique that combined teacher training to strengthen schools’ capacity to respond to GBV with guided student discussion sessions to increase awareness around violence and reporting of abuse. To examine differential effects and complementarities across genders, student training was randomly targeted to girls, boys, or both. The intervention reduced sexual violence perpetrated by teachers against girls by 67 percent in all treated schools, regardless of the gender of students trained. Yet when girls received the student training, girls’ school enrollment increased by 10 percent. The evidence is consistent with victim engagement playing a role in translating reductions in violence into educational gains: training girls increased reporting of GBV to trained school staff, who were more likely to formally report cases to school authorities. Administrative data from the national child helpline further show increases in GBV-related calls and investigations after our intervention. Our findings highlight the importance of institutional accountability in constraining abuse by powerful actors and of engaging victims in reporting mechanisms to translate safer school environments into human capital investment.

   @techreport{NBERw33203,
     keywords = { Working Papers},
     title = {Gender-Based Violence in Schools and Girls’ Education: Experimental Evidence from Mozambique},
     author = {Amaral, Sofia and Garcia, Aixa and Gulesci, Selim and Ramos, Alejandra and Sviatschi, Micaela},
     institution = {National Bureau of Economic Research},
     type = {Working Paper},
     series = {Working Paper Series},
     number = {33203},
     doi = {10.3386/w33203},
     url = {http://www.nber.org/papers/w33203},
     presented_at = {PacDev; UNU-WIDER; RIDGE.},
     ordering = {  }
   }

2. Spillover Effects of School Infrastructure Upgrade: Evidence from Peru

   With Fabiola Alba-Vivar

   This paper examines a school infrastructure upgrading program and its direct and indirect impacts on enrollment, school dropout rates, and student achievement. We exploit the program’s rollout over four years before an abrupt redesign left some schools without the upgrade. Our findings suggest that schools receiving the upgrades experienced a 0.58 percentage point reduction in dropout rates. However, we observe spillover effects on nearby schools: an increase in private school closures and a decline in enrollment, predominantly affecting the public sector. Notably, reading test scores in nearby private and public schools also dropped, suggesting that the students leaving these schools were likely those with higher human capital.

   Media/Policy Mentions: Podcast Perú Debate.

   Presented at: Conferencia Internacional de Desarrollo Económico at PUCP (2023), Crime and Education Labs Seminar at University of Chicago (2025).

   Funded by: Consortium for Economic and Social Research.

   PDF

   This paper examines a school infrastructure upgrading program and its direct and indirect impacts on enrollment, school dropout rates, and student achievement. We exploit the program’s rollout over four years before an abrupt redesign left some schools without the upgrade. Our findings suggest that schools receiving the upgrades experienced a 0.58 percentage point reduction in dropout rates. However, we observe spillover effects on nearby schools: an increase in private school closures and a decline in enrollment, predominantly affecting the public sector. Notably, reading test scores in nearby private and public schools also dropped, suggesting that the students leaving these schools were likely those with higher human capital.

Work in Progress

1. Are Cash Transfer Conditionalities Still Important? Evidence from Colombia on Health Outcomes

   With Bridget Hoffmann , Diego Vera

   Status: Draft coming soon.

   Conditional cash transfer programs tie payments to families’ use of preventive health care, but enforcing these conditions is costly. After decades of such programs, are the conditions still needed, or can habits sustain the behavior they were designed to encourage? We study Colombia’s flagship CCT, Familias en Acción, using an unexpected natural experiment: the suspension of the program’s health conditionality during the COVID-19 pandemic and its later reinstatement. We use nationwide administrative data that track every visit to the health system from 2019 to 2023, linked to household socioeconomic records and social program payments.
2. Policing, Insecurity, and Human Capital: Experimental Evidence from Peru.

   With Isabel Hincapie , Daniela Paz

   Status: Pilot underway.

   This project examines how insecurity in public spaces affects investments in education and health. Attending school and visiting health facilities require repeated trips that expose students, patients, and their families to risks on the street. We test whether unsafe streets reduce these investments. In collaboration with the Government of Peru, we randomize a performance-based transfer from the central government to police precincts in the country’s highest-crime areas, rewarding them for reorganizing patrols around high-traffic sites, including schools, universities, and health facilities. We combine georeferenced crime records with administrative education and health records to measure effects on school attendance and the use of health services.
3. Staying in School: A Large-Scale RCT on Behavioral Interventions to Reduce Crime and Dropout in Peru

   With Natalia Guerrero , Micaela Sviatschi

   Status: Intervention design.

   Criminal organizations depend on a steady supply of young recruits, and schools have become central recruitment sites. Adolescents who join forgo schooling for short-term payments, misjudging the costs of crime and the returns to education. We test whether correcting these beliefs raises the cost of recruitment on the labor supply side. We redesign a police-led prevention program in Peru as a school-based intervention, embedded in a national incentive scheme across the highest-crime police precincts. A two-stage cluster-randomized design identifies direct and spillover effects on beliefs, recruitment, dropout, and local crime.