Analyst herding – whether, why, and when? Two new tests for herding detection in target forecast prices

Abstract

This study proposes two novel tests for security analyst herding based on binomial correlation and forecast error volatility scaling and applies it to investigate herding patterns in analyst target prices in 2008-2020 in the UK. Analysts robustly herd in their valuations, with results consistent across years, sectors, in panel fixed effect, quantile, instrumental variable regressions, and when controlled for optimism and conservatism. Herding becomes prominent for stocks followed by at least five analysts and towards the long sides of Fama-French sorts, reinforcing its non-spurious and behavioral nature. Analyst herd more strongly subject to low volatility and uncertainty.