Does Time Variation in Systematic Risk Affect the Profitability of Contrarian Investment Strategies?

Abstract

Recent empirical studies seem to suggest that contrarian strategies make substantial profits that may be inconsistent with informationaly efficient markets. This paper examines whether the detection of contrarian profits is sensitive to the definition of abnormal returns and the duration of the formation and testing periods and investigates the effect of well-known monthly seasonals on the empirical results. Furthermore, in view of recent evidence that beta risk is not constant, the paper employs a procedure that allows for time-variation in systematic risk, and also examines whether abnormal returns of contrarian strategies are normal compensation for changes in risk between portfolio formation and portfolio testing period. No empirical evidence on contrarian profits exist so far for the Athens Stock Exchange, an emerging market for which one would expect more return predictability. To anticipate the results, we find that longer horizon contrarian strategies are more profitable than shorter horizon strategies. However, contrarian profits are very sensitive to the specification of abnormal returns, i.e. when we allow beta risk to vary over time most of the profits from contrarian strategies disappear. Furthermore, even for the case where we do detect arbitrage profits the results indicate that they may be due to changes in market risk.