Higher Realized Moments and Stock Return Predictability

This study exploits information contained in high frequency sample data by computing higher realized moments of individual firms in the emerging stock market of Pakistan. Furthermore, the relation of higher moments with future stock returns is examined by constructing decile portfolios based on weekly realized volatility, skewness and kurtosis to predict the next week return of the trading strategy that takes long position for portfolio of stocks having high realized moment and takes short position for portfolio of stocks having low realized moment. The long short spread is significant for equal weighted weekly returns based on realized volatility. The long short weekly return is positive and highly significant for realized skewness, 1.659 and 1.969 (in bps) with t-statistics of 7.92 and 14.027 for value and equal weighted portfolios respectively. The result for realized skewness is also supported by Carhart’s Alphas. Similar results are obtained for realized kurtosis, 0.427 and 0.664 (in bps) of long short return, with t-statistics of 2.079 and 4.049 for value and equal weighted portfolios respectively. The evidence suggests that realized skewness and kurtosis can predict the next week’s moment based on cross sectional stock returns.