SAFE: An Early Warning System for Systemic Banking Risk

Abstract

From the financial supervisor’s point of view, an early warning system involves an ex-ante approach to regulation, targeting to predict and prevent crises. An efficient EWS allows timely ex-ante policy action and can reduce the need for ex-post regulation. This paper builds on existing microprudential and macroprudential early warning systems (EWSs) to propose a hybrid class of models for systemic risk incorporating the structural characteristics of the financial system and feedback amplification mechanism. The models explain financial stress using data from five largest bank holding companies, regressing institutional imbalances using an optimal lag method. Z‐scores of institutional data are justified as explanatory imbalances. The models utilize both public and proprietary supervisory data. The SAFE EWS monitors micro-prudential information from systemically important institutions to anticipate build-up of macro-economic stresses in the financial markets at large. To the supervisor, SAFE presents a toolkit of possible institutional supervisory actions that can be used to diffuse the build-up of systemic stress in the financial markets. A hazard inherent for all ex-ante models is that the model uncertainty may lead to wrong policy choices. To mitigate this risk, SAFE develops two modeling perspectives: a set of medium term (six-quarter) forecasting specifications to allow the policymakers sufficient time for ex-ante policy action, and a set of short term (two-quarter) forecasting specifications for verification and adjustment of supervisory actions. Individual financial institutions may utilize public version of SAFE EWS to enhance systemic risk stress testing and scenario analysis. The paper shows econometric results and robustness support for the SAFE set of models. Discussion of results addresses usability and tests of usefulness of supervisory data. In addition, the paper investigates and suggests levels for action thresholds appropriate for this EWS.

Keywords: systemic risk, early warning system, financial stress index, microprudential, macroprudential, structural characteristics, feedback, liquidity amplification, contagion

JEL Classification: G01, G21, G28, C25, C53