A fully Bayesian approach for unit root testing in Chinese stock market

It is well-known that Bayesian approaches can avoid the discontinuity problem of the classical frequent unit root testing statistics. In this paper, the main object is to develop a new fully Bayesian unit root approach based on Kullback-Leibler divergence function and the mixed prior specification. The finite sample behavior of the proposed test statistic is illustrated using Monte Carlo simulation. At last, the developed approach is used to investigate whether the reform of non-tradable shares has long persisted effect on Chinese financial market or not.