Welcome to the materials for the St. Gallen Causal Panel Methods Workshop, a weeklong intensive course (10:15–12:00 & 14:15–16:00 daily) held at the University of St. Gallen from June 30 to July 4, 2025. We focus on estimation, interpretation, application, and troubleshooting of cutting-edge causal inference methods for panel data.
This workshop is a sequel to previous courses on panel methods and is designed for researchers ready to deepen their understanding of modern techniques for causal inference in panel settings. We will cover:
- Difference-in-Differences (DiD)
- Violations of parallel trends and use of covariates
- Event studies, leads/lags, and triple differences
- Differential timing (GxT) and decomposition methods (e.g., Bacon, CS, SA)
- Imputation estimators (e.g., Borusyak et al.)
- Continuous DiD (DxT) and synthetic control (if time permits)
- Practical implementation, checklists, and interpretation issues
You will replicate recent empirical papers and work on your own applied project. This workshop builds both coding fluency and conceptual clarity.
There are 31 students registered for the course, and 13 are taking the exam. Evaluation is based on three components:
- Daily Podcasts: Each day there will be a podcast assigned for listening. You do not need to turn in write-ups daily, but all podcast reflections will be due alongside the final exam.
- Coding Exercise: This will be conducted together in class on the final day (Friday, July 4).
- Final Exam: This is a take-home exam, due by August 13, submitted via Canvas. The exam will be made available shortly after the course ends. Grades (pass/fail for PhD students) must be submitted by August 15.
Daily Time Block (Monday through Friday)
- Morning Lecture: 10:15–12:00
- Lunch at Cafeteria: 12:00–13:00
- Afternoon Lecture: 14:15–16:00
- 2x2 DiD: Four averages, three subtractions
- Potential outcomes framework and ATT
- Three-regression equivalence
- Population weighting in 2x2 DiD
- Introduction to event studies and falsification strategies
- Repeated cross sections and compositional changes
- Conditional parallel trends and selection on observables
- Imputation vs MAR
- Estimation methods: IPW, OR, DR, TWFE with covariates
- Diagnostics: PS score balance, NDIM
- Bacon decomposition of TWFE
- Callaway & Sant’Anna: ATT(g,t) and interpretation
- Event studies using both long differences and short gaps
- Sun & Abraham decomposition
- Real-world GxT example from current research
- Imputation estimator: Borusyak, Jaravel, and Spiess
- Building a diagnostic checklist for your project
- Continuous treatment: Estimation, interpretation, and issues
- Replication: Abortion clinic closures as DxT case
- (If time) Intro to synthetic control
- In-class coding exercise
- Address: House Washington, Rosenbergstrasse 20–22, 9000 St. Gallen
- Lecture Room: 83-2229 in House Washington
- Office Space: Room 83-2217 (Scott Cunningham)
- Cafeteria: Rosenbergstrasse 59 (group walk daily at 12:00)
- Guest Card: Provided on Monday to access building areas
- Monday 4:00 PM: Welcome Apero in Break Room (House Washington)
- Wednesday 4:30–6:30 PM: Bilateral Meetings (sign up with Sam)
- Wednesday 7:00 PM: Faculty Dinner
- Friday 4:00 PM: Farewell Drink @ Restaurant Brauwerk (across from House Washington)
/data/— Sample datasets/labs/— Stata and R code/notes/— Misc. notes/references/— Papers and readings/slides/— Updated lecture slides (redownload each day)
GitHub Repository: https://github.com/scunning1975/St-Gallen
Canvas Course Page (HSG login required): https://learning.unisg.ch/courses/22479
Doctoral students or early-stage researchers using diff-in-diff methods seeking to:
- Understand key identification assumptions
- Learn how modern estimators adjust for violations
- Gain fluency with applied implementation
- Troubleshoot strange or contradictory findings
- Stata:
csdid,drdid,eventstudyinteract - R:
did,fixest,synth,honestdid - Google Sheets (for shared exercises)
Comfort with basic panel data regressions is assumed. Prior DiD experience is not.
- Cunningham (2021), Causal Inference: The Mixtape
- Baker et al. (2025), Working Paper
- Callaway & Sant’Anna (2021), JoE
- Goodman-Bacon (2021), JoE
- Sun & Abraham (2021), JoE
- de Chaisemartin & D'Haultfœuille (2020), AER
- Borusyak et al. (2024), Restud
- Roth & Rambachan (2023), Honest DiD
- Callaway, Goodman-Bacon & Sant’Anna (2025), AER R&R
See /references/ for more.
MIT License. All materials are open for educational use.
Use GitHub Issues or contact Scott Cunningham directly.