1.3.5 Event-Study Designs | Masters in Health Economics
Why It Matters
By revealing the timing and dynamics of policy effects, event studies enable more accurate causal inference and better‑informed economic decisions.
Key Takeaways
- •Event studies capture dynamic treatment effects over time for policy evaluation.
- •Baseline period (t‑1) is dropped to avoid collinearity.
- •Pre‑event coefficients test parallel trends assumption directly in data.
- •Placebo and fake‑unit tests stress‑test causal credibility of results.
- •Event‑study plots reveal anticipatory effects and effect persistence.
Summary
The video introduces event‑study designs as a powerful extension of difference‑in‑differences, allowing researchers to trace policy impacts period by period rather than relying on a single average effect. It explains how to convert a static DID model into a dynamic framework by adding event‑time dummies, dropping the t‑1 period to serve as a baseline and avoid perfect collinearity, and estimating separate beta coefficients for each lead and lag. Key insights include the use of unit and time fixed effects, interpreting beta k as the treatment effect relative to the baseline, and checking the parallel‑trends assumption by testing whether pre‑event coefficients differ from zero. Visualization is emphasized: event‑study plots place event time on the x‑axis, estimated betas on the y‑axis, and 95% confidence bands to assess significance. Robustness checks such as placebo dates and fake‑unit tests are presented to ensure the identified effects are not driven by underlying trends. The instructor illustrates the method with a minimum‑wage case study, showing how coefficients near zero before the policy and a sharp jump at t = 0 signal a credible design. Questions reinforce that dropping t‑1 provides a baseline and that significant pre‑event betas signal a violation of parallel trends. The example demonstrates how anticipatory behavior and lagged adjustments become visible only through an event‑study lens. Overall, event‑study designs give analysts granular insight into when and how policies affect outcomes, strengthening causal claims and informing policymakers about short‑term shocks versus long‑term structural changes.
Comments
Want to join the conversation?
Loading comments...