with Diego Ramos-Toro | Journal of Development Economics (2026)
This paper examines whether military service can promote intergenerational mobility in developing countries. We use survey data from 3,037 men exposed to conscription to compare respondents’ socioeconomic and educational outcomes with those of their parents. Exploiting the random assignment of conscription by lottery, we show that serving increases upward mobility for individuals from low-SES backgrounds, with no effects for those from more privileged ones. The positive effects of military service on mobility disappear under non-democratic regimes or wartime, underscoring the importance of stable institutional contexts for the military to enhance mobility.
with Diego Ramos-Toro | Journal of the European Economic Association (2025)
This paper studies conscription’s role in durably shaping attitudes and beliefs consistent with nation-building. We pair original survey data covering 29 cohorts of conscripts in Argentina with random variation in service emerging from a lottery. We find that serving in the military leads to a stronger national identity and social integration several decades after serving but does not affect civic behaviors such as voting or paying taxes. Value inculcation during service helps explain the baseline patterns, while exposure to and interaction with diverse peers reinforce but do not drive the results.
[paper] [doi]
[pre-analysis plan, round 1] [pre-analysis plan, round 2]
[blog post at VoxDev] [video discussion at Faculti] [blog post at Foco Económico]
with Marcos Balmaceda
Generative AI is free to use and can raise productivity on many tasks, yet entrepreneurs and small firms adopt it slowly and use it shallowly. A natural explanation is that entrepreneurs underestimate how much AI could improve their business, so correcting that belief should raise adoption. We test this in a pre-registered field experiment with 736 Chilean entrepreneurs, randomly assigned to vignettes that show a fictional entrepreneur applying ChatGPT to business tasks, or to an uninformative control. The vignettes shift beliefs: profit expectations rise in a Bayesian pattern, concentrated among the most pessimistic participants. Yet behavior does not follow. Willingness to pay for AI coaching and take-up of a free AI course are unchanged in the same session as the update—ruling out belief decay—and a four-month follow-up finds no effect on AI use, business practices, sales, or profits. The null holds even among the strongest updaters. Asked what holds them back, entrepreneurs name material constraints (money, time, and data privacy), not a failure to see AI's relevance. Correcting profit beliefs is not enough to move adoption. It leaves untouched the costs of acting and the firm-specific work of making a general-purpose tool work in one's own business.
[paper]
This paper provides experimental evidence that behavioral frictions can affect operations planning in small firms, thereby reducing firm performance. I analyze a field experiment involving 14,494 e-commerce retailers in Argentina and Brazil conducted during the two weeks before Black Friday, a major sales event. The intervention randomized messages reminding managers that Black Friday was approaching and encouraging them to plan their pricing and advertising strategies. Treated firms shifted from generic to discount-related advertising and increased inventories before the event, but did not vary their pricing relative to the control group. As a result, sales increased by 5.2% for 60 days following the intervention. Machine learning methods reveal larger returns for relatively bigger firms with a stronger social media presence, suggesting strategic complementarities between firm size and light-touch interventions. These findings highlight that psychological constraints may be a meaningful barrier to the performance of small firms.
[paper]
Revise & Resubmit at Economic Development and Cultural Change
This paper examines the evolution of social cohesion in response to salient events that internally divide communities. The analysis focuses on football rivalries in Latin America, a fundamental cultural phenomenon in the region. Derbies induce polarization between the followers of both teams; however, since followers are not significantly segregated, derbies may also diminish the salience of other cleavages, such as socioeconomic status or ethnicity. By combining quasi-experimental variation in the timing of matches with public opinion surveys, I find that attitudes and beliefs conducive to social cohesion tend to improve for up to 30 days following a match. The effect is moderated by the behavior of football players, as captured by the number of red cards awarded during the game, suggesting that players-as-role-models is an important ingredient of such events.
[paper]
with Patricio Dalton, Carla de Oliveira, and Rodrigo Nicolau
CEDLAS Working Paper #145
In conjunction with the recovery of the Argentine economy between 2003 and 2011, income distribution improved considerably. Though it does not provide a quantitative assessment, the relevant literature points to the resurgence of union negotiation as central to this process. This paper provides an account of the evolution of intra-union and inter-union inequality in basic wage agreements signed during this period, and reveals considerable improvements in both. The second part of the paper studies the behavior of the Federación de Camioneros (Truckers’ Federation) and its role in wage negotiations during the same period.
[paper]