Interbank Exposures: Quantifying the Risk of Contagion

This paper quantifies contagion risk present in the U.S. banking system. Unlike previous studies that infer contagion indirectly by identifying common characteristics of banks that are affected by some event (e.g., third-world debt crisis, large bank failure), this study estimates contagion directly by examining data containing the complete universe of federal funds transactions across banks. Using such data allows for straightforward simulation exercises that demonstrate the degree of contagion that might arise from these exposures. The cost of this direct approach to measuring contagion is clear. The data analyzed only incorporate federal funds transactions. Because of severe data limitations, other exposures among banks cannot be examined on a bilateral basis. As a result of this, an obvious criticism of the results that follow is that other exposures may actually be much higher or may be distributed in a particularly contagion-enhancing way. While it will be argued that the federal funds exposures used in this paper make up a substantial fraction of unsecured interbank credit exposures, one must realize that the conclusions reached are conditional on the set of exposures being examined. That is, the estimates of contagion reported here are accurate, yet potentially conservative. Despite this caveat, the approach employed in this paper to measure contagion has at least three important advantages. First and foremost, the data measure exposures bilaterally. That is, each bank’s exposure to every other bank is known. This