Does cross-shareholding lead to China's stock returns comovement? Evidence from a GMM-based spatial AR model

This investigation investigates how the cross-shareholding network leads to stock returns’ comovement and makes an empirical test using a spatial autoregressive model with multiple weighted matrices. The cross-shareholding network leads to a significant and considerable positive spatial dependence of stock returns, which is roughly equivalent to the local dependence. Besides, we provide evidence that the strength of comovements is closely related to the network density. Vertices with higher out-degree (investing more firms) take on more spillover effects. In contrast, the cross-shareholding ratio seems not to play a critical role because of its low level and the exponential decline influence mechanism. Changes of the cross-shareholding network measurement suggest the robustness of our result. The result suggests that the cross-shareholding could strengthen the stock comovement effect. The policy implication is that the control of this comovement should focus on the network structure: reducing network density by finding vertices with higher degrees rather than arbitrary restrictions of shareholding ratios.