论文部分内容阅读
在很多情况下个体之间都会存在相关性,如果在利用极大似然估计时假设二元选择面板模型的扰动项是相互独立的,那么模型参数的估计结果将不再是有效的.为此,通过将截面相关性纳入模型,构造了基于Copula函数的似然函数,提出了一种两阶段极大似然估计,并证明了估计量的一致性和渐近正态性,以及方差的渐近形式.蒙特卡罗模拟表明新的估计方法明显改善了估计量的有效性.利用这个方法研究了国内上市公司现金分红行为的影响因素,发现销售增长率由不显著变得较为显著.
In many cases, there will be correlation between individuals. If the disturbance items of the binary selection panel model are assumed to be independent when using the maximum likelihood estimation, the estimation result of the model parameters will no longer be valid , The cross-sectional correlation is included in the model, the likelihood function based on Copula function is constructed, a two-stage maximum likelihood estimation is proposed, and the consistency and asymptotic normality of estimators are proved. The Monte Carlo simulation shows that the new estimation method can significantly improve the validity of the estimator. Using this method to study the influencing factors of the cash dividend distribution behavior of listed companies in China, it is found that the sales growth rate becomes insignificant.