differenceindifferences

Difference-in-differences (DiD) is a quasi-experimental econometric method used to estimate causal effects of policies or interventions. It compares changes in an outcome over time between a treatment group that experiences the intervention and a control group that does not, aiming to isolate the policy’s impact from common trends affecting both groups.

Implementation relies on panel data across units and time. A common specification regresses the outcome on unit and time fixed effects, a treatment indicator, a post-treatment indicator, and an interaction term. The coefficient on the interaction (treatment group times post-period) is the DiD estimate of the treatment effect. With multiple periods or staggered adoption, two-way fixed effects or event-study specifications are often used.

Key identification assumption is parallel trends: in the absence of treatment, treated and control groups would have followed similar trajectories. Violations, spillovers across groups, anticipation effects, or changing group composition can bias results. In settings with heterogeneous effects, the DiD estimate may reflect a weighted average of time-specific effects rather than a single causal parameter.

Inference and extensions: standard errors should be clustered at the experimental unit level to account for serial correlation; pre-treatment trend checks and placebo tests improve credibility. Extensions include dynamic effects via leads and lags, multiple time periods, and alternative approaches such as synthetic control methods when parallel trends are questionable. DiD remains a foundational tool for evaluating policy impact using observational data, provided its assumptions are carefully assessed.