Systemic Risk Contributions: A Credit Portfolio Approach

Abstract

We put forward a framework for measuring systemic risk and attributing it to individual banks. Systemic risk is measured as the expected loss to depositors and investors when a low-probability systemic event occurs. The risk contributions are calculated based on derivatives of the systemic risk measure and, thus, ensure a full risk allocation among institutions. We apply our approach to a panel of 54 to 86 of the world's major commercial banks, using 13 years of monthly data. This empirical exercise shows that, whereas the median systemic risk in the sample is about $3tr, it peaks at $20tr at the beginning of 2009. Thereby some 4 to 10 of the biggest contributors account for more than 50% of the system-wide risk, contributing considerably more than their relative size suggests. Based on data for banks' liabilities, default probabilities and asset correlations, we can match very closely the list of G-SIBs revealed by FSB. The individual risk contributions may not only be used for identification of systemically important banks. Being an estimate of negative externalities, they can also be used to compute bank-specific capital surcharges, as we describe in the paper. In addition to this cross-sectional dimension, we also address the time dimension of systemic risk and suggest a method for smoothing the cyclicality of the underlying risk measure. The analysis of risk drivers confirms that the main focus of macroprudential banking supervision should be on a solid capital base throughout the cycle and de-correlation of banks' asset values.