变系数部分线性EV面板数据模型的效应检验

研究了带测量误差的变系数部分线性面板数据模型中个体效应的检验问题。 基于截面最小二乘方法分别对固定效应模型和随机效应模型中参数部分和非参数部分进行估计,经过纠偏处理后分别得到修正后的估计量,并构造出参数 Hausman 检验统计量和非参 Hausman 检验统计量;进一步,在小样本下结合 Bootstrap 抽样方法和Hausman 检验统计量得到检验的拒绝域,并通过模拟来检验统计量的功效;最后利用本文方法实证分析研究了国内 16 家上市银行绩效影响因素,并给出了实质性建议。

Effects test of varying coefficient partially linear EV models with panel data

This paper considers the hypothesis testing of individual effects in varying coefficient partially linear EV models with panel data. It proposes parametric and nonparametric estimators in the fixed effects model and random effects model, respectively, based on the profile least squares method. After correction for the covariates, we derive the modified estimators and construct parametric Hausman test statistics and nonparametric Hausman test statistics.We further obtain the rejection region of the tests by combining the Bootstrap sampling method and the Hausman test statistics for small samples, and calculate the test power function by simulation experiments. Finally we carry out an empirical analysis by analyzing the factors influencing the performance of 16 domestic listed banks, and some concrete proposals are presented.