Overview
Stanford Graduate School of Business (Applied Econometrics). Nobel Prize in Economics 2021 (shared with Joshua Angrist and David Card) for methodological contributions to causal inference. Principal architect of the LATE framework and a leading contributor to the regression discontinuity and matching literatures.
Key Contributions / Features
- LATE (Imbens-Angrist 1994, Econometrica): defined the Local Average Treatment Effect as the IV estimand under heterogeneous treatment effects and monotonicity; resolved the longstanding interpretational ambiguity of instrumental variables estimates.
- Compliance-type taxonomy (Angrist-Imbens-Rubin 1996, JASA): compliers, always-takers, never-takers, defiers — now the standard language for IV and RD analysis.
- FRD external validity (Bertanha-Imbens 2014, NBER WP 20773): derived testable continuity conditions for generalising LATE beyond compliers at the threshold; recommended treatment-arm-specific regression plots as a routine diagnostic.
- Optimal bandwidth for RD (Imbens-Kalyanaraman 2012, Rev. Econ. Studies): data-driven bandwidth choice for local linear RD estimator.
- RD practitioner's guide (Imbens-Lemieux 2008, J. Econometrics): canonical reference for applied RD analysis.
- High-order polynomials in RD (Gelman-Imbens 2014, NBER WP 20405): argues against global polynomial fitting.
Related