Investigating Time Varying Herd Behavior: A Markov Switching Approach

The main objective of this paper is to investigate the herd behavior in Tehran Stock Exchange. To this end, the daily data for all stocks in the market in the period between December 2008 and December 2017 has been used. For assessing herding, the nonlinear version of Chang, Cheng and Khorana (2000) model has been estimated using Markov Switching method. For comparison, we first estimated Chang, et.al. (2000) linear model and then estimated the nonlinear version (Time Varying). The results of linear model show no herding in the market, but results of time varying model indicates that in some periods, market has experienced herd behavior. Results also show that nonlinear model has much better fit on data and consequently accurate results.