分位数单位根检验的拓展及其应用研究

发布时间：2018-11-22 20:28

【摘要】：在不同的分位下对时间序列数据是否含有结构性变化进行甄别,对于准确识别数据的动态变化及其分布特征具有重要的意义。本文首次在Koenker和Xiao(2004)研究的基础上提出傅立叶分位数单位根检验模型,并以此捕捉时间序列中存在的结构突变点,进而刻画数据在不同分位下的动态变化特征。本文通过构建傅立叶QKS统计量并采用蒙特卡罗方法对傅立叶分位数模型的临界值、样本容量和检验"势"进行模拟,发现含有傅立叶级数的分位数单位根检验对刻画"尖峰厚尾"特征数据的非线性偏离动态调节特征具有更高的检验"势"。最后,本文利用拓展后的模型对我国通货膨胀的持久性和失业的回滞效应进行再检验,结果发现我国通货膨胀具有平稳的特征,而失业率却包含单位根过程。本研究为分位数单位根检验的拓展及其应用提供了一定的启示。
[Abstract]:It is of great significance to identify whether the time series data contain structural changes under different quantiles for accurately identifying the dynamic changes and their distribution characteristics of the data. On the basis of the research of Koenker and Xiao (2004), a Fourier quantile unit root test model is proposed for the first time in this paper, which can capture the structural mutation points in time series and then depict the dynamic characteristics of data in different quantiles. In this paper, Fourier QKS statistics are constructed and Monte Carlo method is used to simulate the critical value, sample size and test potential of the Fourier quantile model. It is found that the quantile unit root test with Fourier series has a higher "potential" for characterizing the nonlinear deviation from the dynamic adjustment feature of the "peak thick tail" characteristic data. Finally, the extended model is used to re-test the persistence of inflation and the hysteresis effect of unemployment in China. The results show that inflation in China has stable characteristics, while the unemployment rate contains the unit root process. This study provides some enlightenment for the extension and application of quantile unit root test.
【作者单位】：武汉大学经济与管理学院;中国海洋大学经济学院;青岛银行;台湾逢甲大学财务金融系;
【基金】：山东省社会科学规划研究项目“新常态下山东征信体系建设与区域金融风险防控研究”(15DJJJ20) 中国博士后特别资助项目“金融经济周期视角下的金融冲击识别与金融稳定政策研究”(2016T90645) 山东省博士后创新资助项目“以支持大众创业为导向的普惠金融可持续发展路径探索:以山东省为例”(201503002)的资助
【分类号】：O212

【参考文献】