摘要
从广义矩估计(GMM)到广义经验似然估计(GEL)的发展,是由于GMM估计量小样本性质的不足,促使人们寻求方法的改进和拓展。通过必要的证明和推导,详细解析GEL类估计量(包括EL,ET,CUE)的逻辑关系和数理结构,认识GEL的内在本质,并运用随机模拟方法证实了在小样本场合GEL类估计量比GMM估计量具有更小的估计偏差和均方误差,即GEL类估计改进了GMM估计的小样本性质。
GMM estimator has poor finite sample properties.It makes people to search the improvement and development.This paper makes clear the GEL estimators in logic relations and mathematical structure through some necessary proofs and derivations.And then we can realize the inner essence of GEL estimators.Also this paper confirms GEL estimators have smaller biases and mean square errors than GMM estimator in small sample case by simulations.This is to say GEL estimation is improvement of GMM estimation.
出处
《统计与信息论坛》
CSSCI
2012年第3期21-25,共5页
Journal of Statistics and Information
基金
教育部社科研究基金西部和边疆地区项目<证券市场动态相关性测度的拓展及应用研究>(11XJC910001)