Measuring the disposition effect

Abstract

Despite hundreds of papers confirming the existence of the disposition effect, too little attention has been devoted to the prevailing arguments on the choice of a given method to measure it. This paper fills this gap and compares different measurement approaches.First, based on empirical and simulation-based data, I show how results may differ across measures depending on market trends but, more importantly, on the frequency at which investors make their decisions. Second, the pitfalls in analyzing cross-sectional differences in the disposition effect are illustrated and discussed. Finally, I clearly show that hazard models are quite appropriate to measuring the disposition effect of any investor, be it a day trader or a typical retail investor who monitors his portfolio infrequently.