面板数据中方差的共同变点估计  

Common Change Point Estimation of Variance in Panel Data

作　　者：赵军辉 董翠玲[1] ZHAO Jun-hui;DONG Cui-ling(School of Mathematical Sciences,Xinjiang Normal University,Urumqi,Xinjiang,830017,China)

机构地区：[1]新疆师范大学数学科学学院,新疆乌鲁木齐830017

出　　处：《新疆师范大学学报（自然科学版）》2024年第1期22-32,共11页Journal of Xinjiang Normal University(Natural Sciences Edition)

基　　金：新疆维吾尔自治区自然科学基金项目(2023D01A37,2022D01A219);新疆师范大学重点实验室项目(XJNUSYS082018A01)。

摘　　要：文章对面板数据中方差的共同变点提出了一个含有调节参数的CUSUM(Cumulative Sum)型估计量,证明了变点估计量的相合性,并结合二元分割法将其推广到多个方差共同变点的情形。蒙特卡洛模拟发现,调节参数(γ≠0)下CUSUM型估计量的精确度要高于无调节参数(γ=0)下CUSUM型估计量的精确度。应用外汇汇率进行实证分析,结果也表明调节参数(γ≠0)下CUSUM型估计方法是有效的。This paper proposes a CUSUM(Cumulative Sum) estimator with a turning parameter for common variance change point in panel data,proves the consistency of the estimator,and extends it to the case of multiple common variance change points combining with binary segmentation.Monte Carlo simulation shows that the estimation accuracy of CUSUM estimator with turning parameter(γ ≠ 0) is higher than that of CUSUM estimator without turning parameter(γ = 0).The application analysis of foreign exchange rate also shows that the CUSUM estimation method under turning parameter(γ ≠ 0) is effective.

关 键 词：面板数据 方差变点 CUSUM 调节参数 二元分割法 

分 类 号：O212.1[理学—概率论与数理统计]