More Proof of Math Gone Mad

Last September Kevin Dowd authored a dandy Policy Analysis called “Math Gone Mad: Regulatory Risk Modeling by the Federal Reserve.” In it Kevin pointed to the dangers inherent in the Federal Reserve’s “stress tests,” and the mathematical risk models on which those tests are often based, as devices for determining whether banks are holding enough capital or not.

Recently my Cato colleague Jeff Miron, who edits Cato’s Research Briefs in Economic Policy, alerted us to a new working paper, entitled “The Limits of Model-Based Regulation,” that independently reaches conclusions very similar to Kevin’s. The study, by Markus Behn, Rainer Haselmann, and Vikrant Vig, is summarized in this month’s Research Brief.

The authors conclude that, instead of limiting credit risk by linking bank capital more tightly to the riskiness of banks’ asset holdings, model-based regulation has actually increased credit risk. At the same time, because the model-based approach is relatively costly, large banks are much more likely to resort to it then smaller ones. Consequently, those banks have been able to expand their lending–and their risky lending especially–at the expense of their smaller rivals. In short, big banks gain, small banks lose, and we all are somewhat less safe than we might be otherwise.

Here is a link to the full working paper.

[Cross-posted from]