The influence of volatility and endogeneity in economic time series on cointegration analysis

本研究探讨异方差与内变量对协整分析的影响。我们证明了方差比协整检验在大多数异方差和内变量的情况下产生合理的经验大小，与其他检验相比，包括标准协整检验。虽然阈值调整的协整检验在异方差和内变量下存在大小失真，但在大多数非线性协整关系下通常具有更好的功效表现。

This research investigated the influence of heteroskedastic variances and endogeneity on cointegration analysis. We demonstrated that a variance ratio cointegration test yields reasonable empirical sizes under most cases of heteroskedastic variances and endogeneity, as compared to other tests including standard cointegration tests. While the cointegration test with threshold adjustment has size distortions under heteroskedastic variances and endogeneity, it generally has better power performance under most cointegration relationships with nonlinearity.