Modeling the Spatial-temporal Characteristics of Mutual Funds’ Herd Behavior

Herd behavior in financial markets often leads to unjustified macroscopic phenomena. However, despite existing studies on modeling herd behavior, how it varies across individual agents and over time remains unclear. We show that herd behavior in mutual fund companies can be understood from the functional networks representing interactions inferred from investment similarities. Specifically, in this paper, the spatial characteristics of herd behavior stand for the topology relationships of observations in networks. We analyze the collective dynamics of mutual fund investment from 2003 to 2019 in China using the language of network science and show that herding behavior accompanies this industry’s development but dwindles after the 2015 Chinese market crash. By integrating community detection analysis, we found an increased degree of coherence in the collective herding behavior of the system, even though the localization of herding behavior decreases for clusters of mutual fund companies when the systemic risk builds up. Further analysis showed that herding behavior impacts the payoff of individual fund companies differently across years. The spatial-temporal changes of herding behavior between mutual funds presented in this paper shed light on the debate of individual versus systemic risk and, thus, could interest regulators and investors.