Difference-in-differences with endogenous externalities : model and application to climate econometrics

The difference-in-difference (DID) framework is now a well-accepted method in quasi-experimental research. However, DID does not consider treatment-induced changes to a network linking treated and control units. The endogenous network DID methodology we offer here (ENDID) controls for the direct and indirect role of the treatment on any network member. In addition, we control for the sensitivity of the network to the treatment and its endogeneity on the outcome variable. Monte Carlo simulations and a three-step process combining a gravity model with a 2SLS estimation demonstrate the performance of our new estimator. We apply it to the case where drought impacts global wheat trade and, in turn, wheat production. The results show that a DID that disregards the network and its changes leads to significant underestimates of overall treatment effects.