空间动态面板非线性Logistic模型的稳健LM检验及应用

Robust LM Tests and Applications for Spatial Dynamic Panel Nonlinear Logistic Model

研究目标:探究空间动态面板Logistic平滑转移自回归(SDPD-LSTAR)模型的稳健LM检验的水准和效力。研究方法:重塑得分函数和方差-协方差矩阵,推导识别时间滞后、空间滞后和非线性效应的修正稳健LM检验及其联合效应检验,通过蒙特卡洛模拟和基于STIRPAT的中国284个城市碳排放影响因素实例评估检验的功效和实践性。研究发现:稳健LM检验具有中心卡方的极限分布性质,检验功效好、计算较简便,比一般的LM检验更精确、适用性更广,该优越性随参数局部偏误的出现而显著,应用实例展现了检验良好的实践性。研究创新:提出具有时空依赖和非线性空间区制平滑转换特征的SDPD-LSTAR模型,在ML和GMM框架下推导模型的稳健LM检验。研究价值:探究SDPD-LSTAR模型的选择问题,为空间动态非线性理论和应用提供重要支持。

Research Objectives:Exploring the size and power of the robust LM test for the spatial dynamic panel Logistic smooth transfer autoregression(SDPD-LSTAR)model.Research Methods:The score functions and variance-covariance matrix are adjusted to derive the robust marginal LM test statistics for testing the presence of time lag,spatial lag and nonlinearity effect in the SDPD-LSTAR model along with a joint test for these effects.The tests are analyzed by Monte Carlo simulation and the empirical applications on the STIRPAT-based carbon emission influencing factor model for 284 Cities in China to estimate the size-power and practicality of the tests.Research Findings:The asymptotic distributions of these test statistics are proved to be central chi square distributions.Our robust LM tests have good size-power and simple computation properties.Our robust LM tests are more accurate and applicable than the general LM tests.The advantages of our robust LM tests become increasingly significant as local parametric misspecifications appear.The empirical applications demonstrate that our tests have good practicality properties.Research Innovations:This paper proposes a new SDPD-LSTAR model with spatio-temporal dependence and nonlinear spatial partitioning smooth transformation characteristics.The corresponding robust LM test statistics are derived in the ML and GMM frames.Research Value:This paper explores the selection of SDPD-LSTAR model to provide an important support for the spatial dynamic nonlinear theory and application.