Identifications of the Distribution of Treatment Effect with Duration Outcomes

Abstract

Identification in econometric models maps prior assumptions and the data to information about a parameter of interest. However, there are two important features characterize duration data. The first feature is that the data may be censored, and the second characteristic of duration data is that exogenous determinants of the event times characterizing the data may change during the event spell. The two features induce some well-known identification issues for the duration models. Following the recent literature in partial identification, we provide the conditions when the duration models could be identified and provide several suggestions for the confidence bounds of partial identifications.