The momentum effect on stock markets: A literature review and an empirical study

Maria Eslykke Søndergaard

Student thesis: Master thesis

Abstract

In contrast to the traditional finance theory, stating that predictability of future stock prices and identification of continuously profitable trading strategies are impossible, an extensive body of finance literature in the 1980’s documents that stock prices are, at least somewhat, predictable based on past returns. In 1993, Jegadeesh and Titman publish the first article on medium term momentum in stock prices, documenting that, over 3 to 12 months horizons, past winners continue to outperform past losers by around 1% per month. A significant number of subsequent studies document that the momentum effect is a worldwide phenomenon, which is robust across both different types of stocks and time periods. Apart from clarifying what has already been documented on momentum strategies, the thesis tests whether the momentum effect has existed on the Danish stock market over the period from 1996 through 2009, and finds substantial evidence that it has. The momentum returns from a total of 16 examined strategies all turn out to be significantly positive with average monthly returns ranging from 0.61% to 1.55%. Consistent with previously conducted studies, strategies with long formation periods and short holding periods turn out to be the most successful. The momentum returns continue to be positive even after accounting for transaction costs, and also the momentum effect appears relatively robust across different types of stocks and sub-periods. From a presentation of possible explanations of the momentum effect, it is found that none of the three risk measures from the Three-Factor Model (i.e. beta, size, and book-to-market values) are able to account for the observed return pattern. Also explanations related to data snooping and flawed methodology seem unfounded due to the many studies documenting the momentum effect. The models from behavioural finance, using psychological biases and interaction between different investor types, thus appear to provide the best explanations for the momentum phenomenon. The behavioural models are, however, many and no single model is found to be superior. One possibility could be that all the models individually contribute to explaining the momentum effect, but that different models turn out to be superior in different markets or different types of stocks. This hypothesis, however, remains to be tested.

EducationsMSc in Finance and Accounting, (Graduate Programme) Final Thesis
LanguageEnglish
Publication date2010
Number of pages97