Template-Type: ReDIF-Paper 1.0
Author-Name: Parush  Arora
Author-Name-First: Parush 
Author-Name-Last: Arora
Author-Email: parush.arora@ashoka.edu.in
Author-Workplace-Name: 
Author-Name: Rishabh Bijani
Author-Name-First: Rishabh
Author-Name-Last: Bijani
Author-Email: rishabh.bijani_ug2023@ashoka.edu.in
Author-Workplace-Name: 
Title: Efficient Estimation of Treatment Effects under Staggered Adoption: GMM Approach
Abstract: The two-way fixed effects (TWFE) estimator is biased for the average treatment effect on
the treated (ATT) when treatment is adopted in a staggered manner and effects are heterogeneous
across cohorts and event times. A growing literature has responded with
heterogeneity-robust estimators that restore unbiasedness and attain efficiency under the
assumption of spherical errors. However, the spherical errors assumption is rarely true as
panel data is typical modeled under the assumption of serial correlation. In light of this,
we develop a generalized method of moments (GMM) framework that delivers an unbiased
and efficient estimator of the ATT under arbitrary within-unit serial correlation. The
framework is built on a modified incidence matrix that maps each 2 × 2 DiD comparison to
its corresponding cohort-and-time-specific ATT (CATT) and explicitly debiases “forbidden”
comparisons. Optimal weighting based on the inverse moment-covariance matrix yields the
minimum-variance estimator within the class of unbiased estimators and consistency and
asymptotic normality follow from standard GMM arguments. Monte Carlo evidence confirms
substantial efficiency gains over Callaway and Sant’Anna (2021) and Sun and Abraham
(2021), and modest but systematic gains over Gardner et al. (2024), Wooldridge (2025) and
Borusyak et al. (2024), with the largest gains concentrated at short post-treatment horizons.
Replicating Beck et al. (2010) and Cheng and Hoekstra (2013), we show that the choice of
estimator is not only a technical decision but can potentially change the conclusion of the
paper.
length: 30
Creation-Date: 20260504
Revision-Date: 
Publication-Status: 
File-URL:/www/wwwashokaeduin_628/public/dp/RePEc/ash/wpaper/paper163_0.pdf
File-Format: Application/pdf 
Number: 163
Handle: RePEc:ash:wpaper:163