[数量经济学研讨会]Semiparametric Estimation of Panel Count Data Models with Fixed Effects and Dynamic Linear Feedback

报告题目:Semiparametric Estimation of Panel Count Data Models with Fixed Effects and Dynamic Linear Feedback
报告人:魏杰 (华中科技大学经济学院)

内容简介:In this paper, we consider a semiparametric dynamic panel count data model, where the lagged dependent variable enters the model linearly, and the sequentially exogenous covariates together with the individual fixed effect enter the exponential part. Based on the conditional moment restrictions, we propose a two-stage procedure to estimate the model. In the first stage, we consider the simple nonlinear sieve estimation for both the parametric coefficients and the unknown nonparametric function. In the second stage, we improve the estimation for the nonparametric function based on the empirical solution to a nonlinear integral equation. We establish the asymptotic properties for these estimators when N tends to infinity while T is fixed. We also propose a specification test for the linearity for the unknown nonparametric function. The asymptotic properties of the test statistic such as asymptotic distributions under the null hypotheses and local alternatives and global consistency have been established. A small set of Monte Carlo simulations is carried out to investigate the finite sample performance of our proposed model and specification test. Lastly, we apply our model to a health care demand data set and find that the logarithmic income has a significant nonlinear impact on the number of doctor visits for women but not for men. 




人大经济论坛 rdjjlt.org 人大经济论坛 bbs.econ.ruc.edu.cn