Assessment of Financial Systemic Crisis on a Causal and Reliable Perspective

In financial network models, to assess systemic risk, a general understanding and prediction of precisely how a single financial institution is associated with systemic risk from the network perspective remains lacking. This paper proposes a framework for predicting and assessing system crises through inferring the cause–effect relationships between financial institutions and system state, which is structured in three steps: the assessment stage for system state based on the mean-variance framework, the prediction stage based on a Bayesian network and the reliability stage based on the Markov process. By applying them to monthly returns of financial institutions, it implies the need to pay attention to insurance and Broker sectors while regulating the banking system on the Bayesian network theory. Moreover, we find that the measure contains predictive power both during tranquil periods and during financial crisis periods. The results can be applied to derive interventions in financial crisis management with regard to systemic risk prediction and system state reliability.