部分线性变系数空间自回归面板模型的有效估计及应用

Efficient Estimation and Application of Partially Linear Varying Coefficient Spatial Autoregressive Panel Model

借鉴二次推导函数法(QIF)的思想,结合变量变换法,为部分线性变系数空间自回归固定效应面板模型提供了一种有效的估计方法.在一些正则条件下,建立了估计量的渐近理论,结果表明,参数估计量满足渐近正态性,非参数估计量具有最优收敛速度.进一步,采用Monte Carlo模拟评价了估计方法在有限样本下的表现.最后,将所述模型和方法应用于实际数据分析中.

In this paper,based on idea of quadratic inference functions method(QIF) and combining variable transformation,an efficient estimation method is proposed for partially linear varying coefficient spatial autoregressive panel model with fixed effects.Under some regular conditions,asymptotic theory of estimator is established,it turns out that the parametric estimators satisfy asymptotic normality and the nonparametric estimators achieve the optimal convergence rate.Furthermore,the performance of the estimation method under finite samples is evaluated by Monte Carlo simulation.Finally,the model and method are illustrated by practical data analysis.