The Effect of Beta-Herding on Taiwan’s Market: DCC-MIDAS Approach

The aim of this study is to investigate the herding of beta transmission between return and volatility. We have used the dynamic conditional correlation model with the mixed-data sampling (DCC-MIDAS) model for the analysis. Evidence demonstrates that herding is a key transmitter in Taiwan’s stock market. The significant estimation of DCC-MIDAS explains the herding phenomenon is highly dynamic and time-varying in herding behavior. By means of time-varying beta of herding based on our rolling forecasting method and robustness check of the Markov switching regression approach using four types of portfolios, we find evidence of superior forecasting ability of the model indicates that there are conditional correlations between betas and herding. The evidence also reveals herding formation in Taiwan’s markets during the subprime crisis period.