Research Workshop on “Difference-in-Differences (DiD): Challenges and Recent Developments”
The course takes place on July 24 – 25 at the University of Bamberg.
Description:
Difference-in-Differences (DiD) is one of the most widely used methods for causal inference in economics and related disciplines. In recent years, the traditional approach to staggered DiD settings—namely, the two-way fixed effects (TWFE) regression—has come under significant critique. This has led to dynamic developments in econometric methods aimed at addressing its limitations.
This course provides an introduction to the recent advances in DiD methodology, with a strong emphasis on practical applications. It adopts an example-driven approach, illustrating all key concepts with real-world research examples. The lectures will be complemented by hands-on exercises using datasets from published papers and the statistical software Stata.
By the end of the course, students will understand the main pitfalls associated with traditional DiD approaches and be able to navigate the rapidly evolving literature on "heterogeneity-robust" DiD estimators. The focus is on the following topics:
- Brief review of the basics of the "canonical" two-period DiD model
- Differential treatment timing: TWFE and its limitations
- Selected "heterogeneity-robust" DiD estimators
The course is built around papers and selected chapters of the (open source) textbook:
Cunningham, Scott (2021): Causal Inference: The Mixtape, Yale University Press https://mixtape.scunning.com/
Much of the material takes inspiration from the Mixtape Sessions https://github.com/Mixtape-Sessions/, especially:
- Causal inference II by Scott Cunningham
- Advanced DiD by Jonathan Roth
- Frontiers in DiD by Brantly Callaway
During the PC session, we will work with the data from the paper by Cheng, C., & Hoekstra, M. (2013). Does strengthening self-defense law deter crime or escalate violence?: Evidence from expansions to castle doctrine. Journal of Human Resources, 48(3), 821-854
Prerequisites:
Laptop with Stata license. Basic knowledge of DiD and Stata.
This course is open for advanced masters students and PhD candidates. If you would like to participate, please send a short email describing your motivation, together with your CV by July 3rd to Xenia Tsoukli at xanthi.tsoukli(at)uni-bamberg.de.
No financial support is available.
About the Instructor:
Prof. Dr. Kamila Cygan-Rehm is a Professor of Quantitative Methods, esp. Econometrics at the Dresden University of Technology (TU Dresden). She earned her Ph.D. in Economics from the Friedrich-Alexander University Erlangen-Nürnberg (FAU). Before joining TU Dresden, she led the junior research group "Outcomes of Education Across the Lifespan" at the Leibniz Institute for Educational Trajectories (LIfBi) in Bamberg. She is affiliated with several research networks, including CESifo, IZA, LASER, and LIfBi, and serves on the Scientific Advisory Board of the Research Data Centre (FDZ) of the Statistical Offices of the Federal and State Governments. Her research focuses on applied microeconomics, with particular interests in education and labor economics, public health, and policy evaluation. For more information, please visit her homepage: https://sites.google.com/view/kamila-cygan-rehm
Schedule:
Thursday, July 24
13:00-14:30 Review of the canonical 2x2 DiD
14:45-16:15 Differential treatment timing & TWFE credibility crisis
16:30-18:00 Selected heterogeneity-robust DiD estimators
Friday, July 25
8:45-10:15 PC session: Application to Cheng & Hoekstra (2013)
10:30-12:00 PC session cont'd
13:00-14:30 Wrap-up and Q&A