Can Systemic Risk Measures Explain Bank Defaults?
55 Pages Posted: 16 Jan 2018 Last revised: 7 Jul 2020
Date Written: July 10, 2018
Abstract
This paper assesses the predictive power of systemic risk measures (SRM) for bank defaults by applying a two-staged probit model. Initially, bank defaults are predicted using only idiosyncratic variables. The model is then amended with SRM, which should improve the forecasting accuracy under the hypothesis that bank failure can be either idiosyncratically or systemically induced. Surprisingly, this is not the case. Only one of the three tested SRM benefits the model in statistically significant terms. In investigating the reasons for this finding, a plurality of robustness checks is conducted, controlling for the bank size and the time of the default. It was found that bank size is a strong determinant of how banks allocate their risk budget to credit and market risk. Evidence is generated challenging the adequacy of the tested SRM as they appear to be overly sensitive to market risk. In line with this, their underwhelming performance can be explained by credit risk being the more prevalent risk type. Thus, this paper raises the question whether the current SRM are generic measures of systemic risk.
Keywords: Bank Failure, Bank Default, Bankruptcy, Systemic Risk Measure
JEL Classification: G01, G21, G33
Suggested Citation: Suggested Citation