作　　者： ;

机构地区： 广州大学

出　　处： 《预测》 2008年第6期 67-72,共6页

摘　　要： 经济理论认为许多经济变量具有非对称的阈值自回归调整行为，而标准的单位根检验ADF和PP法都是基于线性自回归模型而构造的，因此对非对称单位根检验并不适用。本文应用Monte-Carlo模拟方法对ADF和PP检验在TAR与M-TAR模型下的检验势进行系统研究，并对模拟结果的产生原因进行了深入分析。模拟结果表明：数据的非对称性程度和数据的均值回复时间是影响ADF和PP检验的最主要因素；当非对称性增强和均值回复时间增大时，ADF和PP的检验势都具有较大幅度的下降，但是PP比ADF法具有更大程度的下降。 Economic theory often predicts that some economic variables display asymmetric threshold autoregressive adjustments towards their long-run equilibrium, but standard ADF and PP test methods are misspecified in this case, consequently, may suffer from a lack of power against such alternatives. This paper aims to study the power of ADF and PP against such alternatives as TAR or M-TAR, and analyse some reasons for that. Mont-Carlo experiments demonstrate that the Asymmetry and mean-reversion of the data plays a important role in the power of ADF and PP methods ; and when the Asymmetry and mean-reversion is very strong, the power of ADF and PP test falls dramatically, but the PP method is less powerful than the ADF test.

关 键 词： TAR与M-TAR模型 MONTE-CARLO模拟 非对称性 检验势

分 类 号： [F224]

领　　域： []