Empirical investigation on CAPM anomalies in Hong Kong stock market

Abstract

This research studies return patterns in the Hong Kong stock market: Fama&French’s 25 portfolios, LSV deciles and double-sort portfolios formed on accounting ratios, momentum portfolios formed on George and Hwang (2004)’s 52-week-high sorts (GH2004) and portfolios formed using Jegadeesh and Titman (1993) (6-6) strategy (JT93). These are well documented CAPM anomalies in the literature. I find these patterns are strong using Hong Kong data retrieved from PACAP Hong Kong database during the period of 1980-2002. I try to use the FF three-factor model to explain these anomalies and find that the FF three-factor model can explain the portfolios’ return in the FF 25 portfolios, LSV deciles and LSV double-sort portfolios but can not explain either the short-term momentum or long-term reversal of average return of portfolios formed from 52-week high sorts and JT93 sorts. In particular, the FF three-factor model fails to explain the long-term portfolio return formed on past 48-month performance, which is one of the results in chapter five. These findings present a challenge to Fama and French (1996)’s view that the CAPM average-return anomalies could be captured by the three-factor model. Moreover, the results in chapter five is different from George and Hwang (2004)’s findings. First, they find that there is 52-week high sorted short-term momentum but not 52-week high sorted long-term reversal in the US market. In contrast, I find both short-term momentum and long-term reversal exist in Hong Kong market. Second, they find 52-week high strategy could explain JT93 strategy. But I find this is not true in the Hong Kong market. My result shows that 52-week high strategy can not explain the JT93 strategy and vice versa. Also, in a regressive framework, I find 52-week high strategy and JT93 (6, 6) strategies show the same power in predicting the future returns. Third, I also consider the liquidity effect which is missing in George and Hwang (2004). I find a strong liquidity effect independent to the momentum effect.