Abstract: Classical tests for unit roots have been criticized for their unusual asymptotic theory leading to disconnected confidence intervals and their lack of power in small samples. We develop a simple and efficient MCMC algorithm for fitting the dynamic quantile regression model based on a location-scale mixture representation of the asymmetric Laplace distribution. Moreover, we exploit the maximum posterior interval for different prior distribution. The simulation result shows that Bayesian quantile regression method is a complete and robust method to test non-stationary for time series. An empirical application of the method modeling the China CPI index dynamic illustrates that there exists stock on upper tail of the distribution but not lower tail.

Key words: quantile, AR models, unit root, bayes factor

摘要： 针对频率统计方法存在不连续的置信区间以及在小样本情况下检验势比较低的问题。把非对称Laplace分布表示成正态分布和指数分布的线性组合,推导了不同先验分布情况下参数的最大后验密度置信区间,并构造了分位回归单位根检验的贝叶斯因子,实现了对非平稳时间序列的局部单位根检验。仿真分析表明贝叶斯分位回归方法是一种稳健全面的单位根检验方法。对我国居民消费价格指数的实证研究发现,我国居民消费价格指数表现出局部的持续性,在分布的下尾部不受普通冲击的影响,但在分布的上尾部受普通冲击的影响。

关键词: 分位数, AR模型, 单位根, 贝叶斯因子