eThesis - School of Business electronic theses

OBJECTIVES OF THE STUDY In this thesis, I describe the time and return dynamics of the disposition effect. The Disposition effect is an empirical fact that investors hold on to losing stocks and sell their profitable investments more eagerly. While the existence of the disposition effect is an empirical fact, it lacks concise and unifying theory. Previous literature hints of a possible time varying component in the propensity to sell on capital gain. This previous evidence of change over the dimension of the time, and over the dimension of the level of capital gain, introduces an empirical gap in the research. What is the true level of the disposition effect in different periods of the holdings and different level of gains? My research fulfills this gap in the literature.

METHODOLOGY AND DATA My primary dataset is the unique dataset of Euroclear Finland. The official registry called the Book Entry System. This is the complete official registry of holdings of all Finnish publicly quoted companies on an individual investor and daily level. The dataset spans twenty years from January 1st, 1995 to January 1st, 2015 and contains over 280 million initial balance records and changes in ownership. I purposely depart from the conventional hypothesis-testing framework for this study. Instead, I combine many of the approaches of the previous empirical research to simultaneously model the dynamics of the disposition effect over the dimensions of the level of gains and along the duration of the holding period. I apply the Cox proportional duration model modified for estimation for large datasets to disentangle the time and gain dynamics of the disposition effect from the baseline hazard.

MAIN FINDINGS I find that the magnitude of the disposition effect changes asymmetrically over the duration of the holding period and differently in the domain of gains compared to the domain of losses. On short holding period durations, it shows a distinct V-shape over the dimension of the level of gain. I additionally find that the functional form of the disposition effect changes easily, depending on for example, whether the share was acquired by a subscription in an equity issue versus being bought from open markets. When investors acquire shares in an equity issue, the propensity to sell those shares is lower, especially if the holding period return is high. This finding is against the main findings on the disposition effect in the existing literature.