Abstract In 1995, Benartzi and Thaler introduced the concept myopic loss aversion to explain the equity premium puzzle. They provided empirical evidence to support their arguments. Recently, Durand et al. criticized this empirical analysis. They propose an approach which not only rejects the significance of the earlier findings but also suggests a reversal of the original findings. In contrast to their approach, we implement a bootstrap approach and find results in line with the results of Benartzi and Thaler. We further show that the significance of the effect strongly depends on somewhat arbitrary assumptions about the length of data history.