Stochastic Regression Models and Their Applications

点击翻译按钮获取中文摘要

I have found some properties of the maximum likelihood estimator and its likelihood function in the conditional t distribution model, where heteroskedasticity occurs as a conditional variance. In particular, I have discovered the degrees of freedom is estimable, contrary to the previous claim in the literature, and its estimation is most successful when estimated using the joint likelihood function. I have also obtained a property related to measures of nonnormality for the asymptotic variances of the least squares estimator of linear asset pricing models when the data generating process is dependent. In addition, I have estimated the effect of the Fukushima nuclear disaster on stock returns of electricity power and gas utilities in Japan.