A variance ratio test of the behaviour of Chinese stock indices

This study utilizes tests based on ranks and signs suggested by Wright (2000) together with the traditional variance ratio test to examine the behaviour of some Chinese stock indices. The results have shown that the null hypothesis of martingale difference behaviour of the Chinese index returns series examined is rejected for the whole samples. However, when the heteroskedastic stochastic disturbance term is used, the finding supports the random walk hypothesis for B shares by end of 1996, the Chinese stock market has become more efficient.