What’s Trending in Difference-in-Differences? A Synthesis of the Recent Literature

Causal Question / Estimand

A synthesis, not a single estimand. Organizes the modern DiD literature around the group-time / dynamic treatment effects (Causal-Estimand) and the assumptions under which they are identified, providing a unified framework and practical roadmap.

Identification Strategy

Lays out the assumptions DiD rests on — Parallel-Trends (unconditional vs Conditional-Parallel-Trends) and No-Anticipation — and then surveys: (i) the TWFE-under-staggered-timing problem and Negative-Weighting; (ii) heterogeneity- robust estimators; (iii) tools for assessing/relaxing parallel trends (pre-tests done right, sensitivity analysis); and (iv) inference/clustering. A decision-oriented map of the field.

Key Assumptions

Parallel-Trends / Conditional-Parallel-Trends, No-Anticipation, and an explicit accounting of Treatment-Effect-Heterogeneity; valid comparison groups and clustering for inference. SUTVA throughout.

Threats to Validity

n/a — survey. Its through-line of cautions: don’t trust unsaturated TWFE under staggered timing; don’t treat pre-trend tests as sufficient; cluster appropriately; match the estimator to the design.

Setting / Data

n/a — methodological synthesis with worked guidance and references to software (did, HonestDiD, did_multiplegt, etc.).

Key Claims

  • A single framework ties together the staggered-timing, parallel-trends, and inference strands of the recent DiD literature.
  • Practical recommendations: pick a heterogeneity-robust estimator, use honest sensitivity analysis for parallel trends, and cluster inference correctly.
  • Serves as the field’s orientation map circa 2023.

Connections

Citation

Roth, J., Sant’Anna, P. H. C., Bilinski, A., & Poe, J. (2023). What’s Trending in Difference-in-Differences? A Synthesis of the Recent Econometrics Literature. Journal of Econometrics, 235(2), 2218–2244. https://doi.org/10.1016/j.jeconom.2023.03.008