Estimating risk-return relationships: An analysis of measures

Abstract

We show that the risk-return paradox can be partly explained by the choice of accounting risk and return measures. Returns computed with equity or assets from End-of-Period (EOP) annual reports produce negative risk-return associations, while measures calculated using Beginning-of-Period (BOP) equity or assets yield more positive relationships. The likelihood of reporting negative relationships using EOP methods is accentuated by dividing samples at median returns. Below-median firms suffer losses and may appear to have lower and more variable returns than above-median firms, simply because of EOP methods. Our results show that mean and variance measures are unstable and risk-return relationships vary inversely the number of firms reporting mean losses.