Studies on return predictability in the Malaysian stock market

This thesis comprises three empirical studies that examine stock return predictability in Malaysia. The first study examines whether future returns can be predicted using past returns and market-wide information. This study provides new evidence using the bootstrap automatic variance ratio test and the price delay measure in the Malaysian context. The findings reveal that stock prices deviate from a random walk and tend to respond slowly to market-wide information over the complete sample period. The findings also demonstrate that informational efficiency increases with firm size and trading frequency. This study also examines time variation in informational efficiency using overlapping and non-overlapping subsample windows. It is observed that serial dependence in stock returns and price delays are time-varying. Further, consistent with Lo’s (2004) adaptive market hypothesis, market efficiency is shown to be dependent on market conditions. The second study investigates the profitability of contrarian and momentum strategies for short-, intermediate- and long-term investment horizons. No prior studies have comprehensively investigated whether contrarian and momentum profits can be related to factors such as firm size, monthly seasonality, market states, lead–lag effects and market overreaction in the context of the Malaysian stock market. The findings reveal that momentum strategies do not generate profits in Malaysia. Rather, contrarian strategies realise significant returns over short, intermediate and long investment horizons. Contrarian profits are most pronounced among medium and small stocks, and they are greater following market downturns. Moreover, the previously documented Chinese New Year effect is evident in contrarian profits. Further, the lead–lag effect identified by Lo and MacKinlay (1990) explains the profitability of contrarian strategies. Contrarian returns also diminish after accounting for the Fama–French factors. With the exception of a 48-month strategy, all strategies yield negative risk-adjusted returns after incorporating transaction costs. The final study is a novel attempt to examine the existence of post-earnings announcement drift (PEAD) and its determinants in the Malaysian stock market. This study examines whether the PEAD anomaly is attributable to market risk, arbitrage risk, liquidity, transaction costs, size, book-to-market ratios and investor sophistication. The findings reveal that an earnings surprise has strong predictive power with respect to PEAD. Further, the results show a significant negative effect of turnover on PEAD, suggesting that liquidity encourages arbitrage activities and enhances the extent to which earnings information is efficiently priced. The results are robust to using three different sample periods and alternative measures of liquidity and arbitrage risks. The findings are also robust to constructing standardised unexpected earnings portfolios based on the ex-ante quintile breakpoints and using a sample of stocks that trade every day of the 60-day event window following the announcement date. Moreover, a strategy based on PEAD yields a three-month post-announcement return of approximately 3.35 per cent. However, estimated transaction costs eliminate all profits generated from strategies based on PEAD. This thesis contributes to market efficiency research by providing an in-depth analysis of return predictability in the Malaysian stock market. The findings of the thesis provide new insights into the degree of stock return predictability in Malaysia. In particular, the results have significant implications for market participants to identify mispricing and execute past returns-based and earnings-based investment strategies. It also aids regulators in identifying episodes of mispricing in the Malaysian stock market.