Bayesian Unit Root Test for Chinese Residents' Consumption Models

摘要

Economic time series may be generated by a vector autoregressive model, and there may be a unit root in the process. It is well known that the classical methods for unit roots test require large samples. This paper deals with the unit roots on the VAR (p) model from a Bayesian perspective, including the parameters' Bayesian statistical inference on designed prior distribution, and devises MCMC computational procedure. With the MCMC simulation plus Savage-Dickey density ratio method, stationarity analysis of Chinese rural residents' consumption and urban residents' consumption are conducted, respectively. The results indicate that the Bayesian method is an effective tool to analyze economic time series.