二次趋势模型的误设检验与仿真分析

在介绍两种生成二次趋势模型的基础上,指明两者具有某种内在的关系,并以隐性趋势模型为数据生成过程,使用显性趋势模型作为估计对象,进行参数估计和相应的假设检验。理论分析结果表明:显性趋势模型的参数、t检验统计量和联合F检验统计量的极限具有非标准的分布,且高度显著;以显性趋势模型为数据生成过程,使用隐性趋势模型作为估计对象,结果表明隐性趋势模型是带趋势项的单位根过程;采用LLR检验统计量对两类模型进行区分检验,使用仿真技术进行模拟,仿真结果支持上述理论分析结论和LLR统计量能够区分两种模型。

Misspecification Tests and Simulation Analysis for Quadratic Trend Model

The paper first introduces two kinds of model generating the quadratic trend and concludes that there are some relations between them through literatures.Then with the recessive trend model for the data generation process,the dominant trend model is used to estimate parameters and execute hypothesis testing,concluding that both the t-test statistic and combined F-test statistics have non-standard limit distribution and are highly significant through theoretical analysis.When the dominant trend model is selected as data generation process and the recessive trend model is used as estimated object,the results show that the recessive trend model is a unit root process with trend.After that,the LLR statistics is introduced to distinguish these two models;Finally the simulation technology is adopted and the simulation results support the theoretical analysis conclusion and indicate that the LLR statistics can distinguish these models.