| Asset return predictability and financial anomalies are focus of a large body of finance literature. Two of the anomalies, stock price momentum and long-term mean reversal, have received much attention in recent years. In this paper, we investigate two issues that are still hotly debated in this field: one is the presence and robustness of mean reversion in U.S. equity market, and the other is the sources of profitability of momentum strategies.; In the first essay, we use a parametric two-component stock price model and a panel based methodology to test for mean reversion. For an extensive use of characteristics-sorted portfolios, we do not find strong statistical evidence in support of mean reversion in U.S. stock market. We find that, if it is ever present, mean reversion exists only in the short and unusual pre-WWII period. To investigate whether mean reversion is important in economic sense, we also implement parametric trading strategies based on predictions of our model, and our findings do not support any economic significance of mean reversion for either individual stocks or stock portfolios. Our results are robust to different model specifications, alternative forecast windows and such considerations as systematic risks and transaction costs. In the second essay, we propose a risk adjustment procedure appropriate for momentum portfolios. This procedure takes into account the positive covariation between factor loadings of momentum portfolios and the corresponding factor premia, which is largely ignored in the conventional risk adjustment procedures based on unconditional full-sample regressions. We find that, using our proposed procedure, about 40% of momentum returns from individual stocks and up to 100% of momentum returns from style portfolios can be explained by FamaFrench three factor model. We conjecture that stock price momentum could be at least partially driven by fundamental risks, and irrationality may not be the whole story. |
