Pedro Hurtado

Micro-entrepreneurship programs are a cornerstone of poverty-reduction policy, yet their cost-effectiveness remains uncertain. Building on a landmark six-country randomized controlled trial by Banerjee et al. (2015), this paper applies causal machine learning methods — causal trees, causal forests, and policy learning — to estimate heterogeneous treatment effects and simulate counterfactual allocations in Ghana, Honduras, and Peru. Reallocating resources toward the most responsive subgroups can transform unviable interventions into profitable ones, with welfare improvements reaching up to 6.7 times those achieved under the original random allocation. Overall, interpretable predictive rules based on household and community characteristics yield efficiency gains of up to 6.7 times without altering program design, highlighting the potential of transparent, data-driven targeting to strengthen poverty-reduction initiatives.

Do constitutional reforms promote long-run economic growth? Using a strongly balanced panel of 57 countries observed annually from 1961 to 2018, this paper estimates the aggregate effect of adopting a new constitution on GDP per capita growth, then decomposes it by examining three provisions linked to well-established theoretical channels: central bank authority, free-competition guarantees, and property-rights protections. New constitutions alone produce modest and imprecise effects on growth. However, specific provisions matter and operate at different speeds: central bank authority clauses generate persistent growth gains beginning around year 5, free-competition provisions show significant effects emerging around year 12, and property-rights protections yield positive but statistically imprecise estimates. These different horizons indicate that long-run growth depends not on the act of constitutional replacement itself, but on the specific economic commitments embedded in the constitutional text.