Are traditional timing models well specified?

Abstract

Traditional timing models are affected by several biases, which generate spurious timing and stock-picking coefficients. Academics have appointed different causes as the possible sources of these biases. A negative correlation between timing and stock-picking abilities arises as a consequence of the biases in traditional timing models. This article provides evidence for one bias commonly found in traditional timing models, which is related with options. We focus on this bias in view of the scant attention it has so far received in the literature. We believe one possible cause for this bias is the failure to include the cost of the option implicit in timing activities in the timing models, and on this basis, we opt for a corrected version of the Merton and Henriksson model (1981). This study therefore is a pioneer in the assessment of the magnitude of this bias and in the measurement of the impact of its correction on fund managers’ results. Our results confirm both the existence of the bias and the correction of the problem when the cost of the option is included in timing models. The modified version of the Merton and Henriksson model, unlike the traditional model, reports positive timing and stock-picking coefficients, supporting the good performance by managers.