This paper uses a regression discontinuity design to study the impacts of a noncontributory pension program covering one-third of Bolivian households during the COVID-19 pandemic. Although the program was not designed to provide emergency assistance, it took on additional importance during the...

Access to identification cards (IDs) is often required to claim government benefits. However, it is unclear which policies to increase ID ownership are more effective. We experimentally analyze the effect of two policy interventions to induce the timely renewal of identification cards on access...

Canonical human capital theories posit that education, by enhancing worker skills, reduces the likelihood that a worker will be laid-off during times of economic change. Yet, this has not been demonstrated causally. We link administrative education records from 1987 through 2002 to nationally...

We study the effects of a Bolivian law that introduced benefits and protections for child workers (who are overwhelmingly informal workers) and lowered the de facto legal working age from 14 to 10. We employ a difference-in-discontinuity approach that exploits the variation in the law's...

We exploit an expansion in social protection to middle-income households to provide evidence on how middle-income households cope with economic shocks and how to build their resilience. We use a regression discontinuity design around the eligibility cutoff for a program that delivered monthly...

The push for adopting digital modes of payment rests on three promises: increased efficiency of transactions, increased financial inclusion, and improvements in the financial well-being of low-income individuals. We experimentally test the extent to which these promises are fulfilled. We exploit...

A key challenge for policymakers is how to design methods to select beneficiaries of social programs when income is volatile and the target population is dynamic. We evaluate a traditional static proxy-means test (PMT) and three policy-relevant alternatives. We use a unique panel dataset of a...