Essays on Return Predictability

Abstract

This dissertation is a collection of three essays that investigate the momentum effect and the short-run predictability in currency carry trade profits. The first essay investigates whether tail risks of momentum strategies make them unattractive within the context of prospect utility. Momentum returns have strongly asymmetric tail risks and that asymmetric tail risk is precisely what makes momentum strategies unattractive. This study is the first to document the undesirable tail risk characteristics of momentum returns. The second essay uncovers economically significant predictability in carry trade profits from shorting the low-yielding currencies. The monthly world equity index return, monthly changes in currency volatility and monthly changes in equity volatility predict carry trade profits from the short leg two months later, while monthly changes in commodity prices, monthly changes in currency volatility and monthly changes in equity volatility predict carry trade profits from the long leg three months later. Investors could have used the discovered leg-specific predictability to time the market and improve their trading outcomes, instead of staying fully invested or predicting carry trade profits from both legs with a single model. Evidence from two tests conducted in this essay points towards the gradual information diffusion model as the most likely explanation for the discovered predictability, while time varying risk premia do not seem to explain this effect. The last essay examines return predictability among carry trades, stocks and commodities in a dynamic vector auto regression setting. The predictive effect goes from commodities to stock, from stocks to low-yielding currencies and from commodities to high-yielding currencies. Variables in these markets are more strongly correlated in the high-risk regime than in the low-risk regime. Drops in the world equity index (commodity prices), but not rises, predict decreases in carry trade profits from low-yielding (high-yielding) currencies. Increases in currency volatility, but not decreases, predict drops in carry trade profits from low-yielding currencies. The in-sample asymmetric effects also exist out-of-sample, but these asymmetric prediction models do not consistently deliver better forecasts than symmetric models.