Working Papers

Abstract: This paper studies how exposure to organized crime affects teenagers' educational choices and how public policy responds in such contexts. We focus on the 2008 extradition of Medellín's top gang leader, after which a gang war erupted and created neighborhood-level variation in criminal activity. Using rich administrative data from Medellín, Colombia, we find that dropout rates among boys increased in affected neighborhoods, while girls were unaffected,consistent with heightened gang recruitment rather than general violence. We then examine how Conditional Cash Transfer (CCT) programs shaped educational outcomes and find that they effectively reduced dropout rates, with stable impacts even amid increased gang activity. To further understand criminal involvement and evaluate policy responses, we develop and estimate a dynamic discrete choice model of schooling, crime, and labor market decisions from middle school through early adulthood. Leveraging design-based results, we identify shifts in the criminal payoff and assess the effectiveness of CCTs in offsetting dropout. The model evaluates three policy interventions: (i) expanding CCT coverage, (ii) increasing law enforcement, and (iii) combining both. Expanding subsidies reduces teenage crime and arrests during school years and raises higher-education enrollment, though with limited post-school effects. Stronger law enforcement curbs post-school crime but raises arrests among teenagers. Combining both policies yields the largest crime reduction but smaller educational gains than CCTs alone. A back-of-the-envelope cost-benefit analysis shows that while the combined policy delivers the highest net benefits, expanding CCT coverage achieves nearly comparable gains at lower implementation costs and with clearer fiscal implications.

Abstract: Foreign military aid can strengthen weak states' capacity to confront internal conflicts, but it may also generate unintended consequences. This paper examines how U.S. military aid to Colombia affected the recruitment of child soldiers. I exploit variation in aid allocation across municipalities using a shift-share instrumental variable to identify the causal impact of increased military investment. The results show that higher U.S. military aid led to greater demand for child soldiers by both paramilitary and guerrilla groups and reduced secondary school enrollment in affected municipalities. To interpret these findings, I develop a model of strategic interaction between the government, paramilitary organizations, and guerrilla forces. The framework highlights distinct mechanisms through which military aid influences armed-group behavior. For paramilitaries, the increased recruitment of children is driven by heightened violence production, consistent with complementary attack strategies and possible resource sharing with the state. For guerrilla groups, the effect arises through a substitution mechanism: as the opportunity cost of adult labor increases, they rely more heavily on child soldiers.

Abstract: Alzheimer's disease and related dementias (ADRD) impose substantial health and economic costs, yet their effects on labor market outcomes remain insufficiently understood. We construct a novel administrative dataset for Colombia linking national health records (RIPS), formal employment and earnings data (PILA), and household survey data (SISBEN) to study the consequences of an early-onset ADRD diagnosis for patients' healthcare utilization and formal labor market outcomes. Using a staggered difference-in-differences event-study design with matched controls, we focus on working-age individuals with early-onset ADRD who were formally employed before diagnosis. We find that diagnosis is followed by large and persistent increases in healthcare utilization, particularly outpatient primary care, and by substantial declines in formal labor market attachment. Within two years of diagnosis, formal employment falls by 10-20 percentage points, with earnings losses of approximately 15 percent for women and nearly 30 percent for men. Overall, the evidence suggests that early-onset ADRD imposes sizable labor-market costs during prime working years and may also affect household labor allocation, underscoring the importance of future work that examines household structure and health care in shaping economic responses to major health shocks.

Abstract: This document uses the Phillips and Sul (2009) test to assess the hypothesis of convergence on educational achievement of public and private schools, at the municipal level, for the period 1980-2014. Even though the normativity and provision of public education is homogeneous across the country, we find that the educational attainment of the municipalities do not show evidence of global convergence. However, we find the existence of six convergence clubs. The first two clubs, which include the municipalities with the best performance, are mainly private institutions in the major metropolitan areas, while the latter three clubs are composed primarily for public institutions located on the Pacific coast and in the Caribbean region.