Estimating analyst's forecast accuracy using behavioural measures (Herding) in the United Kingdom

Purpose – The purpose of this paper is to identify herding behaviour on financial markets and measure the herding behaviour impact on the accuracy of analysts' earnings forecasts. Design/methodology/approach – Two alternative measures of herding behaviour, on analysts' earnings forecasts are proposed. The first measure identifies herding as the tendency of analysts to forecast near the consensus. The second measure identifies herding as the tendency of analysts to follow the most accurate forecaster. This paper employs the method of The Generalised Method of Moments in order to relax any possible biases. Findings – In both measures employed, a positive and significant relation is found between the accuracy of analysts' earnings forecasts and herding behaviour. According to the first measure analysts exhibit herding behaviour by forecasting close to the consensus estimates. According the second herding measure, it is found that analysts tend to herd towards the best forecaster at the time. Finally, it is concluded that the accuracy of analysts' forecasts increases as herding increases. Research limitations/implications – The present study triggers concerns for further research in the modelling of analysts' forecasting behaviour. Originality/value – This paper proposes that a measure based on human biases is the best way to estimate and predict the analysts' earnings forecast future accuracy.