Dynamic Systemic Centrality And Financial Crisis Contagion Source

Stock markets consist a network via information propagation,different markets may playing far different roles in the network.Recognizing systemic vital nodes in the network can help the regulators and the investors to allocate resource more efficiently.Systemic centrality node is a kind of systemic important node in network.Global financial markets have experienced several crises.In process of financial crisis contagion,whether the relevant countries play systemic central node roles?In this paper,we decompose the stock return rate into the combination of good and bad information based on the log return decomposition model.And then we constructs the good information dissemination network and bad information dissemination network.By identifying the systemic centrality of relevant countries in two kinds of information dissemination networks,this paper explores the relationship between dynamic systemic centrality and financial crisis contagion source.Firstly,log return decomposition method is used to extract positive log return and negative log return according to the highest price,the lowest price,the open price and the close price and their appearance order.Secondly,in order to study the systemic centrality nodes features dynamically,the positive log return time series and the negative log return time series are partitioned into rolling windows.Thirdly,in each rolling window,linear Granger causality method is used to construct information propagation network.Fourthly,the Page Rank method is used to identify systemic centrality nodes in information propagation network.Empirical results show that,during recently occurred financial crises usually,concerned countries in crisis do not play systemic centrality node roles in the dissemination of information.In most cases,Chinese mainland market play systemic centrality node roles in both good and bad information propagation networks.