Dittmar, Kaul and Lei disagree. They found that “contrary to the common belief in the profession, momentum is not an anomaly.”18 They explain that, if past returns predicted future returns, one would expect portfolios made up of the biggest percent winners over a period of time to outperform less and less, as the portfolio momentum effect of the biggest winner gets diluted by adding the second-place, third-place and subsequent runner-up historical gainers. Because momentum-based portfolio profits do not decrease with the addition of securities of lesser momentum rank, there must be another explanation. They claim that “momentum appears to be a perfectly understandable phenomenon in a world in which different securities have different expected returns and risks.” Put another way, they see momentum as a correlated change in valuation ratios for various subsets of securities. Like many researchers, the authors take comfort in discovering rational underpinnings to the market. “Our results are reassuring because if momentum were to arise solely or predominantly from irrational serial covariance in the firm-specific components of returns, it would not only violate market efficiency in its weakest form but also bring into question the reasonable belief that individuals are risk-averse and therefore demand a (rather large?) risk premium.”
Ang, Chen and Xing have explained momentum as a risk premium required for taking on additional downside risk.19 They look at downside correlations as a specific risk, beyond other risks. Interestingly, they reject skew, kurtosis and both upside and downside betas as expressions of downside risk. According to their paper, “past winner stocks have high returns, in part, because during periods when the market experiences downside moves, winner stocks move down more with the market than past loser stocks.”
McLean and Pontiff suggest that the momentum effect has declined by 30 percent since its initial publication.20 Asness et al. found that “value and momentum become less profitable, more correlated across markets and asset classes, and less negatively correlated with each other over time” (Figure 4).13
Whether one believes it to be a true anomaly, or a set of priced-in risks, there is reason to believe that momentum, as a factor, has not been completely arbitraged away. Jegadeesh and Titman show that the relative returns to high-momentum stocks increased after the publication of their 1993 paper.21