Exploring New Statistical Methods: Inference with Few Treated Units

Bruno Ferman
Professor of Economics
Fundacao Getulio Vargas (FGV) - Sao Paulo

In many causal inference applications, only one or a few units (or clusters) receive treatment. A key challenge in such cases is that standard inference methods based on asymptotic theory may be unreliable, even when the total number of units is large. This workshop explores inference techniques specifically designed for settings with few treated units, covering both cross-sectional and panel data approaches.

The first part of the workshop provides an overview of why standard inference methods often fail in these scenarios. It then introduces alternative approaches, discussing their trade-offs and connections. While different methods rely on different assumptions and data requirements, a common theme is that the limited information on treated units necessitates stronger assumptions compared to standard inference methods.

In the second part of the workshop, participants will be invited to implement different inference methods through simulations, allowing them to experience firsthand both the challenges and potential solutions for settings with few treated units. If possible, participants are encouraged to bring a laptop with either Stata or R installed. Only basic proficiency in one of these programs is required. However, those without a laptop or prior experience with these tools are still welcome to participate.

Workshop Schedule:
11am to 12pm - Session 1
12pm to 1pm - Lunch
1pm to 2pm - Session 2

Bruno Ferman is a professor at the São Paolo School of Economics and an expert on developing and using new statistical methods. His own research focuses on behavioral economics and finance, on themes including microcredit product design and the ways in which financial incentives can influence or reveal behavior preferences. His work has explored topics such as the impact of peers’ social influence on individuals’ financial decisions, and the impact of information disclosure on consumers’ borrowing decisions. Ferman holds a PhD from MIT, as well as an MA and BA in economics from the Pontifícia Universidade Católica do Rio de Janeiro.