Event Study Design (Causal Event Study)
Also known as: dynamic difference-in-differences, event-study DiD, dynamic treatment effects, leads-and-lags model, Event Study Tasarımı (Nedensel Olay Çalışması)
The event study design is a generalised difference-in-differences model that estimates a separate treatment-effect coefficient for each period before and after an intervention, tracing the dynamics of the effect over event time. Its modern, heterogeneity-robust form was developed by Sun & Abraham (2021) and Callaway & Sant'Anna (2021).
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use the event study design with panel data where units are treated at one or more points in time and you want to trace the effect dynamically rather than as a single average. It needs a reasonable panel (at least about 50 units/observations) and several pre- and post-treatment periods. The key assumption is parallel trends: the pre-treatment coefficients should not differ from zero. When treatment is staggered across units, use the Sun-Abraham or Callaway-Sant'Anna estimators, and always report cluster-robust standard errors.
Strengths & limitations
- Traces the full time path of a treatment effect — pre-trend, immediate impact, and fade-out — rather than collapsing it to one number.
- Provides a built-in visual test of the parallel-trends assumption through the pre-treatment coefficients.
- Generalises difference-in-differences and, with the Sun-Abraham or Callaway-Sant'Anna corrections, handles staggered adoption with heterogeneous effects.
- Needs a sufficient panel (about 50+ units/observations) and several periods on each side of the event; otherwise the period coefficients are imprecise and the parallel-trends test is underpowered.
- Plain two-way fixed effects can be biased when treatment timing is staggered and effects are heterogeneous, requiring the Sun-Abraham or Callaway-Sant'Anna corrections.
- Inference relies on cluster-robust standard errors; ignoring serial correlation understates the standard errors.
Frequently asked
How is an event study different from plain difference-in-differences?
Difference-in-differences gives a single average effect of treatment after the event. The event study generalises it by estimating a separate coefficient for each period before and after treatment, so you can see the pre-trend, the immediate impact, and how the effect changes over time.
What do the pre-treatment coefficients tell me?
They are a visual test of the parallel-trends assumption. If the coefficients for the periods before treatment are close to zero, treated and control units were moving together beforehand, which supports the causal interpretation. Coefficients that drift away from zero warn that the assumption may be violated.
Why is the reference period usually -1?
One event-time coefficient must be dropped to give the others a baseline, and the last pre-treatment period (-1) is the natural choice. Every remaining coefficient is then interpreted relative to that just-before-treatment moment.
When do I need the Sun-Abraham or Callaway-Sant'Anna estimator?
When units are treated at different times (staggered adoption) and the effect varies across groups. In that case ordinary two-way fixed effects can be biased, and these estimators recover interpretable dynamic effects.
Sources
- Sun, L. & Abraham, S. (2021). Estimating Dynamic Treatment Effects in Event Studies with Heterogeneous Treatment Effects. Journal of Econometrics, 225(2), 175–199. DOI: 10.1016/j.jeconom.2020.09.006 ↗
- Callaway, B. & Sant'Anna, P. H. C. (2021). Difference-in-Differences with Multiple Time Periods. Journal of Econometrics, 225(2), 200–230. DOI: 10.1016/j.jeconom.2020.12.001 ↗
How to cite this page
ScholarGate. (2026, June 1). Event Study Design (Causal Event Study). ScholarGate. https://scholargate.app/en/causal-inference/event-study-causal
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Interrupted Time SeriesCausal inference↔ compare
- Panel Fixed EffectsEconometrics↔ compare
- Regression DiscontinuityCausal inference↔ compare
- Shift-Share IVCausal inference↔ compare
- Staggered Difference-in-DifferencesCausal inference↔ compare