Abstract: From the concept of illiquidity, we calculate seven illiquidity proxies by using samples from Shanghai and shenzhen stock exchange (csi) 300, and use principal component analysis (pca) to obtain the comprehensive illiquidity proxy(prin). We use ARFIMA model to fit the long memory and residual correlation of various variables, monte carlo simulation is constructed to measure the influence of long memory and residual correlation on prediction regression, and Bootstrap sampling method is used to adjust the deviation and screen the illiquidity measure with robust prediction ability. The results show that:the illiquidity measure can predict the short, medium and long term excess returns, R² measures of fht and prin are high, which reflects that the important role of illiquidity factor in liquidity risk pricing; The short-term forecasting ability of most illiquidity indexes to the market excess returns comes from the market volatility, while the medium-term and long-term forecasting ability is dominated by the illiquidity factors which have nothing to do with the volatility.

Key words: illiquidity, long memory, equity premium, market volatility