Translating DID advancements to TDID

Investigate how methodological advancements in difference-in-differences (DID)—including multiple time periods, repeated cross-section sampling, time-varying controls, continuous treatments, and compositional changes—translate to conditional triple difference-in-differences (TDID), and clarify the identification challenges and feasible solutions specific to the TDID setting.

Background

The paper develops an identification and estimation framework for conditional triple difference-in-differences (TDID), showing that standard TDID implementations with controls are biased due to differences in covariate distributions across groups and proposing a re-weighted double-robust approach to address this issue.

In the conclusion, the authors outline natural extensions mirroring recent DID developments—such as multiple time periods, repeated cross-section sampling, time-varying controls, continuous treatments, and compositional changes—and note that TDID presents distinct identification challenges compared to DID, explicitly indicating that open questions remain about how these DID advancements carry over to TDID.