PROJECT: A Sensitivity-Based View of Financial Systemic Risk Modeling

A Sensitivity-Based View of Financial Systemic Risk Modeling.

We formulated the inter-bank systemic risk as an optimization problem with equilibrium constraints. An optimal liquidation policy of the central bank to minimize the system’s loss was derived by the Direct-Comparison-Based Approach. One-region, multi-region, and dynamic cases could be modeled.

We apply the direct-comparison based approach to solve the optimization problem and derive an efficient iteration algorithm to this highly nonlinear problem. Our work illustrates the advantages of the direct-comparison based approach, which was developed first for discrete event dynamic systems, and has been applied to solve many theoretical as well as practical problems. The central piece of the approach is the performance difference formula (PDF) that compares the performance under any two policies; we find that surprisingly the PDF for the financial risk contagion problem looks similar to that in MDPs. The research in this work indicates that it can be applied to both static and dynamic problems.