Herding Behavior In Stock Markets

Traditional finance theory is based on efficient market hypothesis, while empirical research in these years found that there is seriously irrational behavior in stock markets. Behavioral finance becomes a new hot topic, and herding behavior is a typical case. Herding behavior in stock markets is the phenomenon that market traders are influenced by others, ignore their own information, make similar decisions, which is reflected in asset price.We used residuum not returns to research herding behavior, which can avoid pseudo herding. By using multielement statistic method, we got necessary and sufficient condition of existence of herding behavior. At the same time, we used complex networks to describe herd. We used average length of all the shortest path and clustering coefficient to find the optimal threshold correlation coefficient. We also build a herding network, calculating the distribution of degree. Then we extended CAPM model, adding herding unit, and dividing the risk. Finally, we forecast ed the direction of herding evolvement, checked its fat tail.