面板半参数空间自回归模型的变量选择——基于STIRPAT模型的碳排放影响因素分析

发布时间：2018-04-11 06:22

本文选题：半参数空间自回归模型 + 变量选择　；参考：《数理统计与管理》2017年05期

【摘要】：变量选择有助于简化模型,提高估计和预测的精度,但目前鲜有涉及面板半参数空间自回归模型变量选择的研究。本文在ALASSO的基础上提出了SSAR-ALASSO法,该法的核心在于惩罚函数的选择和目标函数的构建。SSAR-ALASSO在变量和参数的对应关系、惩罚函数的选择、特殊参数的取值区间以及适用模型等方面与ALASSO存在差异。模拟结果显示,SSAR-ALASSO法在变量选择的准确性和参数估计的精度两方面均表现良好,随着样本容量的增加表现效果更佳。本文在碳排放量影响因素实证中采用SSAR-ALASSO法对STIRPAT模型进行变量选择。研究结果表明人均财富、技术水平、产业结构、所有制结构和产业集聚显著影响碳排放量,城市化、对外开放、能源价格和环境政策对碳排放量无显著影响。
[Abstract]:Variable selection is helpful to simplify the model and improve the accuracy of estimation and prediction, but there are few researches on variable selection in panel semi-parametric space autoregressive model.On the basis of ALASSO, the SSAR-ALASSO method is proposed in this paper. The core of the method is the choice of penalty function and the construction of objective function. The corresponding relation between variables and parameters, the choice of penalty function, and the choice of penalty function are presented in SSAR-ALASSO.The value range and applicable model of special parameters are different from ALASSO.The simulation results show that the SSAR-ALASSO method performs well in both the accuracy of variable selection and the accuracy of parameter estimation, and the performance is better with the increase of sample size.In this paper, the SSAR-ALASSO method is used to select the variables of the STIRPAT model in the empirical analysis of the influencing factors of carbon emissions.The results show that per capita wealth, technological level, industrial structure, ownership structure and industrial agglomeration significantly affect carbon emissions, urbanization, opening to the outside world, energy prices and environmental policies have no significant impact on carbon emissions.
【作者单位】：中山大学岭南学院;暨南大学经济学院;
【基金】：国家哲学社会科学基金重点项目(15ATJ001,15AJL005)的资助
【分类号】：O212.1

【相似文献】