Ana Armendariz
PhD Candidate in Economics & Econometrics
University of St Gallen
Working papers
- Do Drug Consumption Rooms Reduce Drug-Related Hospitalizations? Evidence from Switzerland
This paper estimates the causal effect of drug consumption room openings on drug-related hospitalizations in Switzerland, exploiting their staggered introduction across Swiss cities between 1998 and 2022 with individual-level hospital records and a difference-in-differences design. Openings reduce drug-related hospitalizations by roughly 35% within 5 km of a facility, with the largest reductions in fatal outcomes.
- Testing Effect Homogeneity and Confounding in High-Dimensional Experimental and Observational Studies
We propose a framework for testing the homogeneity of conditional average treatment effects (CATEs) across multiple experimental and observational studies, using double machine learning to handle high-dimensional covariates.
Work in progress
- When Do Common Trends and Unconfoundedness Agree? An Overidentification Test Across Studies
Applied economists with panel data must often choose between difference-in-differences, which relies on parallel trends, and selection on observables, which relies on unconfoundedness. We use a Hausman-type overidentification test with machine-learning-based covariate control to assess whether the two strategies are jointly consistent, and apply it to the outcomes of 25 published DiD studies spanning labor, health, development, trade, and political economy.
- Precisely Zero: Migration's Null Effect on Crime
We examine the effect of migration on crime in Switzerland between 2009 and 2025, leveraging the quasi-random allocation of asylum seekers across cantons and detailed administrative data. We find precisely estimated null effects on overall, sex, violent, and property crime, human trafficking, and unauthorized prostitution, robust across migrant statuses, defendant nationalities, and crime specifications.
- Overfitting in Policy Learning
We propose a method to detect overfitting in policy learning models and develop tools to help applied researchers mitigate this issue.