Stock return predictability and the winner-loser reversals : analysis of the arguments and the Singapore evidence.

Abstract

This dissertation examines the winner-loser return reversals in the Stock Exchange of Singapore. Three major arguments that have been proposed to explain this phenomenon are analyzed. It's shown that there are significant return reversals in SES after considering several possible methodological flaws and biases. The reversals then are not simply the outcome of an elaborate data snooping process. There is no evidence that loser portfolios are fundamentally riskier than winner portfolios either in terms of timevarying risks or performance in adverse economic states. Investors do not make systematic expectational errors for both winner and loser portfolios if E/P or C/P are used to proxy investor's expectation about the stock's future growth rates. Evidence is also provided to reject the size effect and the January effect as explanations for the reversals. The apparent anomaly of winner-loser reversals in SES therefore can not be explained by existing arguments such as methodological flaws and biases, time-varying risks and investors' extrapolation. The results do not rule out the possibility of the existence of other expectational errors. A possibility consistent with the data is mispricing due to a variety of unidentified causes which is most likely to surface among stocks with extreme price changes.