固定效应部分线性可加空间自回归面板模型的截面二次推断函数估计  

Profile Quadratic Inference Functions Estimation for Partially Linear Additive Spatial Autoregressive Panel Model with Fixed Effects

作　　者：丁飞鹏 丁松 DING Feipeng;DING Song(School of Statistics and Data Science,Jiangri University of Finance and Economics,Nanchang 330013;School of Mathematics and Statistics,Jiangci Normal University,Nanchang 330022;School of Economics,Zhejiang University of Finance and Economics,Hangzhou 310018)

机构地区：[1]江西财经大学统计与数据科学学院,南昌330013 [2]江西师范大学数学与统计学院,南昌330022 [3]浙江财经大学经济学院,杭州310018

出　　处：《系统科学与数学》2023年第9期2404-2428,共25页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(71961011);国家自然科学基金(71901191)资助课题;江西高校人文社科基金(TJ19203)。

摘　　要：半参数空间面板模型是空间面板数据分析中的重要工具之一.文章考虑了一类半参数空间面板模型:固定效应部分线性可加空间自回归面板模型.固定效应引起的附带参数问题、模型的空间内生性以及面板数据在个体内的相关性等问题,增加了模型估计的难度.为此,文章提出了截面二次推断函数法,该方法将样条函数近似、马尔科夫链蒙特卡罗(MCMC)模拟法、二次推断函数(QIF)法及变量变换法进行有机结合,可以较好地解决模型的上述问题.进一步,在一些正则条件下,推导了该估计方法的大样本性质,结论表明,参数估计量渐近服从正态分布,非参数连接函数实现了最优收敛速度.与此同时,Monte Carlo模拟结果显示,该方法的表现优于两阶段最小二乘法;最后,该模型和方法被应用于实际数据分析中.Semiparametric spatial panel models are one of important instruments in spatial panel data analysis.This paper considers a semiparametric spatial panel model:Partially linear additive spatial autoregressive panel model with fixed effects.The incidental parameters problem caused by fixed effects,the spatial endogeneity of the model and the correlation within individuals of panel data increase the difficulty of model estimation.In this paper,the profile quadratic inference functions estimation method is constructed by combining spline approximation,Markov Chain Monte Carlo(MCMC)simulation,quadratic inference functions(QIF)and variable transformation,which can solve the above problem of the proposed model.Furthermore,under some wild condition,the large sample properties of the proposed method is derived.The conclusion is that parameter estimators follow asymptotically normal distribution,and nonparametric estimators achieve the optimal rate.Meanwhile Monte Carlo simulation results show that the performance of the proposed method is better than that of the two-stage least squares method.Finally,the proposed model and the proposed method are applied to the real data analysis.

关 键 词：空间自回归模型 部分线性可加模型 空间面板数据分析 

分 类 号：O212.1[理学—概率论与数理统计]