| 摘要: | 研究期間：10108~10207;Among all stock market anomalies, the co-existence of short-term (or intermediate- term) momentum and long-term reversals in stock returns is perhaps one of the most puzzling phenomena that cannot be fully explained by common rationality-based theories. Famous behavioral models, such as those proposed by Barberis, Shleifer, and Vishny (1998), Daniel, Hirshleifer, and Subrahmanyam (1998), and Hong and Stein (1999), all attempt to resolve the puzzle of short-term “underreaction” and long-term “overreaction” based on different behavioral assumptions. A common feature of the three theories is that they all treat short-term momentum and long-term reversals as inseparable phenomena. George and Hwang (2004, 2007), however, argue that the two phenomena are separate ones by showing that they can be explained by a 52-week high ratio and a five-year-low ratio, respectively. As the short-term momentum can be explained by the 52-week high ratio, whose returns do not reverse in the long term, George and Hwang (2004) attribute the short-term momentum to an underreaction-only story. By contrast, George and Hwang (2007) attribute long-term reversals to a capital-gains lock-in effect captured by the five-year low ratio. I am, however, skeptical of the competing hopotheses proposed by George and Hwang (2004, 2007). First, if 52-week high really subsumes the short-term price momentum, and is attributed to underreaction, would its predictability also persist in markets where there exists no short-term momentum like those of Japan? Second, even if it is capital gains lock-in that drives long-term reversals, George and Hwang (2007) do not explain why it takes up to three to five years to reflect the changes in reservation prices. In this three-year project, I will re-examine the robustness of George and Hwang’s (2004, 2007) results in two major ways. 1. I will first investigate whether the 52-week high and 5-year low in combina- tion really fully explain the anomaly of short-term momentum and long-term reversals using Japanese data. This is important because George and Hwang’s (2004) analysis is based on the U.S. data, which is known to suffer from serious data-mining problem. Using Japanese data has two major advantages: it is free from the data-snoopying problem, and it is known to exhibit no short-term momentum (see, e.g., Chou et al (2007) and Chui et al (2010)). I will also investigate if the explanatory power of 52-week high (and 5-year low) is really behaviorally driven by examining if it is related to arbitrage risk. 2. Based on the U.S. data, I will examine if long-term reversals are really attributed to the capital gains lock-in effect. As an alternative hypothesis, I propose that it is investor sentiment that causes long-term return reversals. |
