一类半参数GARCH-M模型的统计检验  

The Statistical Test for A Semiparametric GARCH-M Model

作　　者：赵海清 刘惠 麦继芳 ZHAO Hai-qing;LIU Hui;MAI Ji-fang(School of Mathematics and Statistics,Lingnan Normal University,Zhanjiang 524048,China)

机构地区：[1]岭南师范学院数学与统计学院,广东湛江524048

出　　处：《数理统计与管理》2021年第1期51-60,共10页Journal of Applied Statistics and Management

基　　金：岭南师范学院人才专项(ZL2037);广东省科技计划项目(2017A030303085);教育部产学研合作协同育人项目(201802151033)。

摘　　要：半参数GARCH-M模型能够对资本市场的相对风险厌恶进行度量,本文利用经验似然方法对风险厌恶是否受到外生变量的影响进行检验。论文首先讨论了模型的参数估计及其渐近性质,其次利用估计方程构造检验统计量并证明其渐近服从卡方分布,最后进行了数值模拟。结果表明估计精度较高,检验统计量对备择假设比较敏感。With empirical likelihood methods,the statistical test for a semiparametric GARCH-M models is investigated in this paper.Firstly,the quasi maximum likelihood estimators for parametric parts are proved to be consistency and asymptotical normality.Secondly,the empirical log-likelihood ratio test statistic is constructed and the asymptotic null distribution is proved to be chi-squared.The new method has the advantage that the tests are self-scale invariant,and simulation shows that the test statistics are quite sensitive against the alternative hypothesis.

关 键 词：GARCH-M模型 相对风险厌恶 假设检验 经验似然 

分 类 号：O212[理学—概率论与数理统计]