Abstract
Staggered difference-in-differences (DiD) designs—where different units receive treatment at different times—are increasingly common in accounting research. However, recent econometric advances have revealed fundamental challenges with standard two-way fixed effects (TWFE) estimation in staggered settings, potentially biasing results in many accounting studies that rely on this methodology.
We provide a comprehensive primer for accounting researchers on staggered DiD designs. We explain the sources of bias in TWFE estimation when treatment timing is staggered, demonstrate how these biases manifest in typical accounting applications, and review alternative estimation strategies that address these issues. Our goal is to help researchers understand when standard approaches are appropriate and when alternative methods are necessary.
Key Findings
TWFE Bias in Staggered Settings
Standard two-way fixed effects estimation can produce biased and inconsistent estimates when treatment timing varies across units, a problem often ignored in accounting applications.
Negative Weights Problem
TWFE estimation in staggered designs creates negative weights on treatment effects for some groups, implicitly comparing treated and control units in ways that violate researcher intent.
Alternative Approaches
Newer methodologies like local treatment effects, heterogeneity-robust estimators, and event-study approaches provide valid alternatives to TWFE in staggered settings.
Practical Guidance
We provide empirical examples demonstrating how different estimation approaches yield different results, and offer practical guidance for researchers on method selection.
Research Contribution
This primer addresses a critical methodological issue facing accounting research. As staggered DiD designs become more prevalent in studying regulatory changes, tax reforms, and other policy variations, understanding their statistical properties is essential for drawing valid inferences.
Our paper is not primarily a methodological contribution, but rather a comprehensive introduction to important econometric challenges that many accounting researchers may not be aware of. By raising awareness of these issues and providing practical guidance on alternative approaches, we aim to improve the methodological rigor of future accounting research using staggered designs.