Economerric theory and EmpiricalApplications of Panel Data Analysis

在众多贡献中，主要成果概括如下：1.提出了一种新的估计动态面板数据模型的方法，该方法具有较小的偏差和检验的大小失真。结果表明，对于某些协方差较弱或较多的模型，有限信息极大似然估计比GMM更有效。针对结构动力面板数据模型，提出了一种新的有限信息极大似然方法。提出了一种在时间序列发生结构性变化时估计协整向量的新方法。这对动态面板数据分析具有重要意义。基于日本非上市公司的面板数据，在投资函数中考察了现金约束的影响。使用日本公司的面板数据表明，生产率差异只能部分解释男女工资差异。根据《家庭收入和支出调查》，低收入家庭的消费行为受到可支配收入的限制。使用面板数据对近年来日本婚姻行为代际变化的各个方面进行了调查

Main results, among many contributions, are summarized as follows1. A new estimation method for dynamic panel data models with less bias and size distortion of the test has been developed.2. It has been shown that the limited information maximum likelihood estimator is more efficient than the GMM for some models with weak or many instrumental variables.3. A new limited information maximum likelihood method has been proposed for structural dynamic panel data models.4. A new method has been proposed for estimating cointegrating vectors when there is a structural change in time series. it contains important implications on the dynamic panel data analysis.5. Based upon the panel data on Japanese unlisted firms, the effect of cash constraint has been investigated in the investment function.6. Using panel data from Japanese firms, it has been shown that the male-female wage differential can be only partially explained by the productivity differential.7. Using the Family Income and Expenditure Survey, it has been shown that households with low income face the disposable income constraint in their consumption behavior.8. Various aspects of intergenerational changes in marriage behavior in recent Japan have been investigated using panel data