STAR误差修正模型中协整关系检验的inf-t统计量

An inf-t Test for Cointegration in STAR Error Correction Models

在将误差修正过程设定为全局平稳的指数平滑转换函数的情形下,本文建立了一个新的检验统计量inf-t对非线性STAR误差修正模型中的协整关系进行检验;推导了inf-t统计量的渐近分布,并通过Monte Carlo模拟的方式给出了其渐近临界值。inf-t统计量取未识别参数空间上的下确界,有效避免了原假设下的未识别参数问题。Monte Carlo数值模拟研究的结果表明,inf-t统计量相对E-G两步法的t_(EG)和Kapetanios等(2006)的t_(NLECM)统计量具有更高的检验势。将inf-t统计量应用于对我国货币需求稳定性进行检验,发现我国狭义货币需求量长期稳定,短期存在指数平滑非线性机制转换特征。

This paper proposes a new statistic inf-t testing cointegration rela- tionship in a smooth transition vector error correction model that follows a globally stationary ESTAR error correction process. We derive the relevant asymptotic dis- tribution of the proposed test, and obtain critical values by Monte Carlo simula- tions. The inf-t statistic is the infimum over the unidentified parameter space, and overcome the unidentified parameter problem under null hypothesis. Monte Carlo simulation exercises confirm that our proposed test has much better power than E-G approach and tNtCMStatistic in Kapetanios et al. （2006） against the alternative of a globally stationary STAR cointegrating process. By testing the stability of money demand in China, we find that narrow money demand in China is stability in long term, but existence ESTAR nonlinear dynamic in short term_