AAAI Publications, Workshops at the Thirty-Second AAAI Conference on Artificial Intelligence

Font Size: 
Finding Exogenous Variation in Data
Eliot Abrams, George Gui, Ali Hortacsu

Last modified: 2018-06-20


We reconsider the classic problem of recovering exogenous variation from an endogenous regressor. Two-stage least squares recovers the exogenous variation through presuming the existence of an instrumental variable. We instead rely on the assumption that the regressor is a mixture of exogenous and endogenous observations–say as the result of a temporary natural experiment. With this assumption, we propose an alternative two-stage method based on nonparametrically estimating a mixture model to recover a subset of the exogenous observations. We demonstrate that our method recovers exogenous observations in simulation and can be used to find pricing experiments hidden in grocery store scanner data.


IV; Finite Mixture Models; Nonparametric Estimation

Full Text: PDF