Laplacian-energy-like measure: Does it improve the Cross-Sectional Absolute Deviation herding model?

How to accurately and effectively grasp herding behavior is one of the crucial and challenging issues in the field of financial risk management. Most herding-based paradigms are strictly limited to over-idealized assumptions such as fully connected among individuals or regular structure to local neighborhoods, and the drawbacks are actually obvious. Using a sample of Chinese A-shares from 2006 to 2021, we create a series of daily networks and modify the classical Cross-Sectional Absolute Deviation model by introducing a network topological variable, the Laplacian-energy-like measure. We find that the modified herding model consistently outperforms the traditional model in terms of goodness-of-fit and detection accuracy. This modified model suggests that herding is essentially driven by external contingencies and is more pronounced when stock network interconnectedness is higher. Moreover, the modified herding model is robust to changes in market conditions and also holds in the context of different developed and emerging markets.