The two faces of interbank correlation

Correlations of stock returns across banks are an essential input into systemic risk measures. We demonstrate that such correlations can be decomposed into two parts: a systematic component arising from diversification activities, and a systemic component specific to banks. We find that at U.S. Banking Holding Companies correlations are to a large extent driven by the systematic component. However, applying the decomposition to the Marginal Expected Shortfall (MES), we show that it is the systemic component that predicts bank failure and risk during the Global Financial Crisis. The results suggest that it is important to distinguish between the two sources of correlations when measuring systemic risk at banks.