Herding or Claustrophobia? Using Random Portfolios to Analyze a Measure of Herding

Abstract

The concept of fund manager herding has been studied in depth, and the most widely used measure applied to this market-wide phenomena is the one incepted in Lakonishok, Shleifer & Vishny (1992), LSV. However, this measure has been much criticized, and its validity is still in doubt. This paper presents an in-depth analysis of the LSV measure making use of random portfolios. This methodology allows the LSV measure to be estimated in various conditions, but by construction all are consistent with the null hypothesis of no herding. The main findings confirm the bias inherent in the LSV measure, the non-normality of the resulting distribution of the measure, and its strong dependence on trading intensity, independent of actual herding activity. However, using random portfolio methodology a non-parametric test is proposed that corrects some of the statistical shortcomings of the measure.

Keywords: Herding, herding measures, non parametric measures, random portfolios

JEL Classification: C63, G11, G23