[Economics Seminar] Multiple Treatments and Varying Treatment Intensity in Staggered Difference-in-Differences Designs
Radeef Chundakkadan
Dear All,
The Department of Economics, IIT Bombay, welcomes you to a seminar by Prof. Partha Deb on March 11th, 2025 at 2 pm. More details are below:
Speaker: Prof. Partha Deb (City University of New York)
Date and Time: Mar 11th, 2025 (Wednesday); 2 PM
Venue: Room no. 401, Seminar Hall, 4th floor, A91 Eco Hub Building, Department of Economics, IIT Bombay
Title: Multiple Treatments and Varying Treatment Intensity in Staggered Difference-in-Differences Designs
Abstract: In recent years, there have been an extraordinary number of new theoretical papers on how to obtain consistent estimates of treatment effects in difference-in-differences designs when treatment timing is staggered over time across treated units, and treatment effects are heterogeneous across units and time (e.g., Callaway and Sant’Anna, 2021; Sun and Abraham, 2021; Borusyak et al., 2024; Wooldridge, 2025; Deb et al., 2025). These approaches were motivated by the discovery that the so-called two-way fixed effects model, which imposes a constant treatment effect across units and time, does not recover an interesting causal effect; in particular, it does not identify a suitably defined weighted average of causal effects when treatment implementation is staggered (e.g., de Chaisemartin and D’Haultfoeuille, 2020; Goodman-Bacon, 2021).
Most of the new developments in this area assume a canonical staggered difference-in-differences design—that there is only one treatment (policy change), that it can be characterized as a discrete change, and that it is absorbing. In Deb et al. (2025), we develop FLEX—a flexible linear model estimated by OLS with covariates (X)—a pooled regression approach to produce consistent average treatment effects on the treated for staggered designs in repeated cross-sections. In some applications, however, the researcher is interested in multiple events that are related to each other and occur during the sample period with correlated, but not concurrent, timing. In other applications, the same event might recur in an already treated unit. In yet other applications, the intensity of treatment might vary in substantively important ways (continuous treatment). In such circumstances, de Chaisemartin and D’Haultfoeuille (2023) and Tsai (2025) show that both the two-way fixed effects regression with multiple event indicators and the staggered difference-in-differences estimators designed for use with one treatment are biased in the presence of such confounding events. Each develops a new approach that resolves these issues, but both are complex to implement.
In this paper, we extend FLEX to the case of multiple events with correlated timing. Identification of the group-time level effects of each type, level, or intensity of treatment requires the existence of never-treated units. In fact, it turns out that if the FLEX regression developed in Deb et al. (2025) is specified appropriately, then no new estimation is required to obtain average treatment effects on the treated for each type, level, or intensity of treatment. The solution simply requires thoughtful aggregation of the FLEX regression parameter estimates. We first demonstrate the FLEX approach in the context of Monte Carlo simulations, where the correct answer is known. Then, we demonstrate this approach in two policy-relevant applications.
About the speaker: Prof. Partha Deb is Professor of Economics at Hunter College, City University of New York, NY and a Research Associate at the National Bureau of Economic Research. He received his PhD in Economics from Rutgers University in 1991 and has previously held faculty positions at Indiana University Indianapolis and the University of Illinois. His research focuses on the economics of health and healthcare, and applied econometrics. In 2011, he was appointed Senior Advisor at the Center for Medicare and Medicaid Innovation, where he helped design and evaluate large-scale experimental and quasi-experimental healthcare interventions aimed at improving quality and reducing costs. He has an extensive record of mentoring students and collaborating with them on published research.
Regards,