Identification of Dynamic Panel Binary Response Models
2020/02/19
研討會日期 : 2020-02-19
時間 : 14:30
主講人 : Professor Shakeeb Khan
主持人 : Professor Yu-Chin Hsu
地點 : Conference Room B110
演講者簡介 : Professor Khan received his Ph.D. in Economics from Princeton University in 1997. He is currently a Professor at Boston College. His research interests are Econometrics, and Applied Econometrics.
演講摘要 : We consider identification of parameters in dynamic binary response models with panel data under minimal assumptions. This model is prominent in empirical economics as it has been used to infer state dependence in the presence of fixed effects. The main result is a characterization of the identified set which turns out to be a union of convex polyhedrons. Relative to the benchmark result of Honoré and Kyriazidou (2000) (HK) we make several advances: 1) our identified set do not require any restrictions on the support of the observables; for example, unlike HK, we allow for time trends, time dummies, and/or only discrete covariates. 2) our main results are derived under the assumptions that the idiosyncratic error terms are stationary over time conditional on only the fixed effect and the covariates (and without conditioning on initial conditions); this is in contrast to most dynamic nonlinear panel data models including HK that use stronger restrictions on the initial conditions. 3) we show that it is possible to get point identification in some cases even with T =2 (two time periods). For inference in cases with discrete regressors, we provide a linear programming approach to constructing confidence sets for the parameters of interest that is simple to implement. The paper presents simulation based evidence on the size and shape of the identified sets in varying designs to illustrate the informational content of different assumptions. We illustrate our inference technique by analyzing persistence in women’s labor supply decisions.
