A new method of measuring herding in stock market and its empirical results in Chinese A-share market

- We propose a new model of measuring herding behavior of stock markets.
- The model can test strong herding patterns while filtering out weaker ones.
- Market turmoil of 2007-2008 caused long-lasting and decaying herding in China.
- Empirical results show that the multivariate models fit well.
- Herding is a relative concept, which depends on long-term data as its benchmark.

We propose a new method based on Arbitrage Pricing Theory to test herding pattern. With the innovative WCSV model, we theoretically prove that this method can test strong herding patterns while filtering out weaker ones, showing better discriminating power than previous methods; also, it can detect time points that affect the overall herding level. Empirically, we apply this method to Chinese A-share market and conclude that the market turmoil of Year 2007-2008 caused long-lasting herding with a decaying trend. Fama-French Three-Factor Model is proved to be an applicable underlying APT model in WCSV model with good model fitting and significant improvement compared to univariate CAPM. Also, we suggest that herding is a relative rather than absolute concept, which depends highly on the chosen benchmark.