2026–27 Economics Job Market Candidate
Juan C. Yamin
I develop econometric methods for translating empirical evidence into policy decisions and experimental designs.
Statistical Decision Theory · Experimental Design · Causal Inference
I am an econometrician and Ph.D. candidate in Economics at Brown University.
Job Market Paper
Poverty Targeting with Imperfect Information
Better targeting requires not only better income predictions, but better decisions about how to act on them.
Cash transfer programs often target households using income estimates from surveys, censuses, or proxy means tests. The standard approach takes those estimates at face value and allocates transfers to the households that look poorest. But extreme estimates can reflect noise as well as extreme poverty, causing programs to concentrate scarce transfers on too few households and miss others who are genuinely poor.
I show that this plug-in rule is inadmissible: another allocation rule does at least as well in every case, and strictly better in some. I develop a nonparametric empirical Bayes rule that uses the same data, budget, and prediction model but first learns how much to trust each estimate. In simulations using household survey data from nine African countries, the rule reaches 45.6 poor households per 1,000 people, compared with 25.5 under standard targeting—nearly 80 percent more.
Research Agenda
Current Research
Beyond poverty targeting, my current work develops econometric tools for experimental design and empirical Bayes estimation, alongside a collaborative applied publication on corruption and accountability.
When and How to Pilot
Working paper
Design rules for field pilots, staged experiments, and A/B tests, studying how small first-wave estimates should guide later allocation while guarding against overreaction to small-sample variation.
Two-Way Effects Models
Working paper
with Cole Davis
A nonparametric empirical Bayes approach to estimating worker and firm effects when the distribution of individual effects may vary with latent firm components.
Subnational Alignment and Corruption
Conditionally accepted at the American Political Science Review
with Leopoldo Fergusson, Arturo Harker, and Carlos Molina
Evidence on how partisan alignment shapes corruption and accountability in Colombian local government.
Software
cmrdesign
cmrdesign is open-source R and Python software for pilot-informed experimental design. It uses pilot outcomes and treatment labels to recommend a main-wave treatment allocation and report a finite-sample regret certificate.