We apply the Kalman filter method to estimate nine Asian markets and find evidence that stock return dispersions decline as markets experience stress conditions, supporting the existence of herding. This paper finds that herding behavior is time-varying and comoving across markets. Both linear and nonlinear Granger causality tests conclude that there is strong bilateral causality between herding and returns for all nine Asian markets. For markets in Japan, South Korea, and Thailand, we consistently find strong two-way causality exists in pairwise variables among herding, stock returns, and illiquidity. No consistent evidence can be drawn from other markets for other pairwise variables. © 2014 World Scientific Publishing Co. and Center for Pacific Basin Business, Economics and Finance Research.