Website CV Paper
Fields
Econometrics, Applied Micro
Job market paper
Changes-in-Changes with a Continuous Treatment
This paper extends the changes-in-changes framework to continuously distributed treatments in a two-period panel without requiring exact stayers. Existing approaches typically identify effects for subpopulations associated with realized treatment changes. We instead establish point identification of counterfactual outcome distributions across treatment levels for the whole population, allowing for endogenous selection into treatment. We model selection through a copula between potential outcomes and treatment, assuming it is stable over time and belongs to a one-parameter family. We then exploit cross-period shifts in treatment ranks to recover the counterfactual distributions. We also establish identification under more flexible local Gaussian restrictions that relax the one-parameter copula restriction. We propose a semiparametric estimator based on distribution regression and establish its asymptotic properties. An application to extreme heat exposure and U.S. corn yields finds that greater heat exposure shifts the yield distribution downward, with larger losses in the lower tail.