半参数变系数空间误差回归模型的估计

Estimation of Semi-parametric Varying Coefficient Spatial Error Regression Model

研究目标:克服半参数变系数回归模型中误差项可能存在的空间相关性问题。研究方法:提出一类新的半参数变系数空间误差回归模型,并构造其截面似然估计。研究发现:在小样本条件下,模型估计量具有良好的表现,其精度随着样本容量的增加而提高;应用该方法分析我国资源禀赋与地方公共品供给之间的相互关系,进一步证实了模型较强的适用性。研究创新:证明了估计量的一致性与渐近正态性,并通过蒙特卡洛模拟考察了估计方法的小样本表现。研究价值:新方法对于其他结构的半/非参数空间计量模型理论研究具有推广价值,其估计技术在经济、管理等学科中具有应用价值。

Research Objectives： Solving the spatial correlation of random error terms for the semiparametric varying coefficient regression model. Research Methods： Proposing a new class of semi-parametric varying coefficient spatial error regression model and constructing a profile likelihood estimation method is constructed. Research Findings： The research results show that all estimators have good performances under finite sample and their accuracy is improved with the increasing of sample size. We apply the method to analyze the relationship between natural resources and local public goods supply in China, which confirms the model＇s wide applicability. Research Innovations： The consistency and asymptotic normality of all estimators are proved, and the finite sample performance is studied by Monte Carlo simulation. Research Value： The research methods have popularized values for other semi-parametric and non-parametric spatial econometric models, and the estimation techniques have good application prospects in the fields of economy and management.