文章摘要
Wang Zengmin,Li Tong.Research on the Test Power of Autocorrelation Test Methods[J].The Journal of quantitative and technical economics,2022,(1):155-189
自相关检验方法的检验功效研究
Research on the Test Power of Autocorrelation Test Methods
  
DOI:
中文关键词: DW检验  LM检验  检验功效  蒙特卡罗模拟
英文关键词: DW Test  LM Test  Test Power  Monte Carlo Simulation
基金项目:本文获得国家自然科学基金重点项目(71850013)资助。
Author NameAffiliation
Wang Zengmin School of Economics and Management, Beijing University of Posts and Telecommunications 
Li Tong School of Economics and Management, Beijing University of Posts and Telecommunications 
Hits:
Download times:
中文摘要:
      研究目标:探究DW检验和LM检验的检验功效及其渐近性。研究方法:运用蒙特卡罗模拟实验方法结合相关影响因素对两种检验方法进行分析与比较。影响因素包括样本容量、解释变量的随机性及自相关性、随机误差项的自相关程度以及分布形态。研究发现:DW和LM检验功效与样本容量和随机误差项的自相关程度正相关,与解释变量的自相关程度负相关;解释变量的随机性对DW和LM检验功效无显著影响;误差项的几种常见分布形态的变化对DW和LM检验功效的影响可以忽略;在误差项存在一阶自相关的情况下,DW检验效果优于LM检验效果。研究创新:以DW检验和LM检验的假设条件为出发点,探究比较不同条件下自相关检验方法的检验功效。研究价值:在实证研究背景下为有效地选择自相关检验方法提供借鉴参考。
英文摘要:
      Research Objectives: To explore the power and asymptotic behavior of DW and LM test. Research Methods: Monte Carlo simulation experiment is used to compare and analyze the influencing factors including sample size, the randomness and autocorrelation of explanatory variables, the autocorrelation degree and the distribution pattern of random error terms. Research Findings: The power of DW and LM test is positively correlated with the autocorrelation degree of random error term and sample size, and negatively correlated with the autocorrelation degree of explanatory variables. The randomness of explanatory variables has no significant influence on the power of DW and LM test. The influence of the variation of error term distribution on the efficiency of DW and LM test can be ignored. When the error term has first-order autocorrelation, the power of DW test is better than that of LM test. Research Innovations: Starting from the hypothesis conditions of DW and LM test, the test power of autocorrelation test methods under different conditions is explored and compared. Research Value: Under the background of empirical research, it can provide reference for the effective selection of autocorrelation test methods.
View Full Text   View/Add Comment  Download reader
Close