空间面板平滑转移门槛模型的设定与估计

研究目标:探究具有个体效应的三类空间面板非线性模型:ARAR-STAR模型、SAR-STAR模型和SEM-STAR模型的设定与参数估计。研究方法:对于模型的设定,利用LM检验法来检验是否存在非线性和空间依赖性,用拟极大似然法和马尔科夫链蒙特卡洛(MCMC)方法来估计三类空间面板非线性模型中的参数,并通过蒙特卡洛模拟来评估参数估计的准确性。研究发现:蒙特卡洛模拟实验表明LM检验法在小样本下具有很好的表现,计算得到的拒绝率基本都落在95%的置信区间内,检验的功效都很理想。空间非线性模型中参数的准确估计需要的数据较多,一般来说,当T大于10、N大于200时,拟极大似然法和MCMC法都能够得到参数的准确估计,平滑参数γ的估计偏差较大,但对γ估计的误差不会影响到其他参数的准确估计。研究创新:探究了空间面板非线性模型的选择和两种参数估计方法。研究价值:丰富了空间非线性面板的文献,拓展了现有空间面板非线性模型的形式,为应用研究提供理论依据。

Spatial Panel;STAR Models;LM Test;Quasi-maximum Likelihood Estimation;MCMC

Research Objectives:Exploring the setting and parameter estimation of three types of spatial panel nonlinear models with individual effects:ARAR-STAR model,SAR-STAR model and SEM-STAR model.Research Methods:For the setting of the model,the LM test is used to test whether there is nonlinearity and spatial dependence.The quasi-maximum likelihood method and the Markov chain Monte Carlo(MCMC)method are used to estimate the three types of spatial panel nonlinear models.The parameters are evaluated by Monte Carlo simulation to estimate the accuracy of the parameters.Research Findings:Monte Carlo simulation experiments show that the LM test has a good performance in small samples,and the calculated rejection rate basically falls within the 95%confidence interval,and the test has great efficacy.Accurate estimation of parameters in spatial nonlinear models requires more data.Generally speaking,when T is greater than 10 and N is greater than 200,both quasi-maximum likelihood method and MCMC method can obtain accurate estimation of parameters.The estimation deviation of smoothing parameterγis large,but the error ofγestimation will not affect the accurate estimation of other parameters.Research Innovations:It explores the selection of nonlinear models for spatial panels and two methods of parameter estimation.Research Value:This paper enriches the literature of spatial panels,expands the form of existing spatial panel nonlinear models,and provides theoretical basis for applied research.