An Innovative Approach to Analyze Financial Contagion Using Causality-Based Complex Network and Value at Risk

In this paper, we propose a new approach to analyze financial contagion using a causality-based complex network and value-at-risk (VaR). We innovatively combine the use of VaR and an expected shortfall (ES)-based causality network with impulse response analysis to discover features of financial contagion. We improve the current research methods by building a Granger causality network on VaR and ES and using conclusions drawn from network analysis as a foundational step before impulse response analysis. First of all, we select 30 stock indices that are very well-known globally and collect their trading data. After calculating the risk indicators of VaR and ES, we perform the Granger causality test on them and then build networks based on their respective Granger causality square matrix. Next, we examine the networks’ topological features to discover different degrees of risk transmission among all stock indices in the system. Lastly, we identify the most and the least active stock indices in the risk transmission network and conduct impulse response analysis on them. We discover that BSESN (India S&P BSE SENSEX) is the most risk-sensitive stock index as its VaR significantly increases by 0.03–0.04% and its ES jumps even more, by 0.07–0.08%, in response to an impulse from a few key stock indices. We also find that either PSI20 or XU100 is the most risk-proof stock index, depending on whether we choose VaR or ES as a risk indicator.