Causal networks reveal the response of Chinese stocks to modern crises.

Changes in stock markets impact the optimal policies for governments, businesses, and even individual households. Causal networks summarizing pairwise directed dependencies between stocks can be used to characterize these changes, as well as to identify influential stocks whose variations have unusually strong effects on the movements of other stocks. In this paper, we apply Granger causality tests in sliding windows to build directed causal networks representing the interactions of stock markets in the world's second largest economy, China, during recent economic, trade and epidemiological crises. We propose identifying influential stocks by using distance correlation and surrogate tests, rather than linear Pearson correlation and asymptotic distributions, to quantify the strength of general nonlinear dependencies between network centrality and financial indicators. We find that, at the onset of a crisis, influential stocks gain their position by themselves being attentive to the stock market. As the crisis progresses, the stock market steadily adjusts to give a stronger leadership role to influential stocks at the expense of non-influential stocks. Interestingly, the industries to which influential stocks belong are closely related to the nature of crisis events. Finally, we show that industries which contained influential stocks represented better investments. [ABSTRACT FROM AUTHOR]