The momentum premium: an economic and behavioural examination

This thesis contains four empirical studies in asset pricing with a focus on momentum strategies. The empirical work in this thesis explores both behavioural and risk-based explanations of the size and persistence of the returns to momentum strategies in equity markets. The first study examines the relationship between return dispersion, a proxy for sectoral shifts, and the United States business cycle. This chapter also examines the relationship between return dispersion and the equity premium and finds that high return dispersion is associated with a negative future equity premium. The results in this chapter show that changes in return dispersion act as a state variable. Motivated by these findings, the second study examines the relationship between return dispersion and the momentum premium across four regional momentum strategies and a global momentum strategy. This chapter finds that return dispersion is negatively related to future momentum returns. This relationship is robust to the inclusion of control variables that have previously been shown to be related to the momentum premium. This chapter proposes a conditional momentum strategy that alters the exposure of the traditional momentum strategy by incorporating the level of return dispersion. This conditional momentum strategy outperforms the unconditional momentum strategy in all regions. To examine a behavioural explanation for the persistence of momentum strategy returns the third study examines the relationship between investor myopia and momentum returns under the framework of myopic loss aversion. This study finds that countries that are predisposed to myopia have a higher momentum premium, a finding which is consistent with the prediction of theories founded on myopic loss aversion. The fourth study investigates investor extrapolation. Using a proxy for trend salience, this chapter finds that momentum strategies that exclude stocks with a non-salient trend outperform the traditional momentum strategy. This result cannot be explained by the Fama and French (1993) factors. These results suggest that trend salience supports the role of extrapolation in the behavioural models of momentum.