Crash Testing German Banks
by Klaus Düllmann of Deutsche Bundesbank, and
Abstract: In this paper we stress-test credit portfolios of twenty-eight German banks based on a Merton-type multifactor credit-risk model. The stress scenario is an economic downturn in the automobile sector. Although the percentage of loans in the automobile sector is relatively low for all banks in the sample, the expected loss conditional on the stress event increases substantially by 70-80 percent for the total portfolio. This result confirms the need to account for hidden sectoral concentration risk because the increase in expected loss is driven mainly by correlation effects with related industry sectors. Therefore, credit-risk dependencies between sectors have to be adequately captured even if the trigger event is confined to a single sector. Finally, we calculate the impact on banks' own-funds ratios, which decrease on average from 12 percent to 11.4 percent due to the stress event, which indicates that banks overall remain well capitalized. These main results are robust against various robustness checks, namely those concerning the granularity of the credit portfolio, the level of intersector asset correlations, and a cross-sectional variation of intrasector asset correlations.
Published in: International Journal of Central Banking, Vol. 5, No. 3, (September 2009), pp. 139-175.