Date of Award

12-2005

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Business Administration

Major Professor

Phillip Daves

Committee Members

James W. Wansley, Michael C. Ehrhardt, Halima Bensmail

Abstract

One of the most controversial topics in recent investment literature has been stock return momentum. If an investor buys past winners and sells past losers, he will earn positive profits in the intermediate-term horizon (3 to 12 months). While behavioral theories seem to dominate as an explanation for the momentum phenomenon since momentum has been regarded as direct counter evidence for the efficient market hypothesis, Chordia and Shivakumar (2002) find that momentum can be explained by a set of macroeconomic variables. Chordia and Shivakumar argue that momentum is caused by time-varying expected returns that can be predicted by a set of macroeconomic variables, which might be associated with time-varying risk.

However, the first essay of my dissertation shows that even if the macroeconomic variables are independent of stock returns, they can appear to predict momentum profits if they exhibit high persistence and the momentum portfolio period overlaps with the parameter estimation period. I am able to produce results similar to those of Chordia and Shivakumar with randomly generated variables, while I show that once the parameter estimation periods are changed, the predictive power of the macroeconomic variables for momentum disappear. My results provide evidence that the predictive power of the macroeconomic variables comes from a spurious relation between stock returns during the momentum portfolio formation period and predicted returns from the macroeconomic variables. My results further suggest that Chordia and Shivakumar’s argument that the predictive power of macroeconomic variables for momentum is a challenge to behavioral theories is indeed premature.

The second essay shows that the ratio of the 50-day moving average to the 200-day moving average has significant predictive power for future returns. Stocks with a high moving average ratio tend to outperform stocks with a low moving average ratio for the next six months. This predictive power is distinct from that of the nearness of the current price to the 52-week high, which was first documented by George and Hwang (2004). The moving average ratio, combined with the nearness to the 52-week high, can explain most of the intermediate-term momentum profits. This suggests that an anchoring bias in which investors use moving averages and the 52-week high as their reference points for estimating fundamental values is the main source of momentum effects. Momentum profits caused by the anchoring bias do not disappear in the long-run, confirming George and Hwang’s argument that intermediate-term momentum and long-term reversals are separate phenomena.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS