Math Gone Mad: Systemic Dangers of the Federal Reserve’s Stress Tests

Featuring George Selgin and Kevin Dowd

The Federal Reserve’s “stress tests” were intended to make the financial system safer. Using risk modeling, the tests subject banks to various stress scenarios in the economy to see how well they would perform and determine the capital “buffer” banks need to remain solvent and safely weather poor economic conditions. But with the Fed relying on risk modeling to determine regulatory capital requirements for banks, what happens if the models are wrong? Risk models can suffer from major weaknesses, ranging from poor assumptions to inadequate data, and can be particularly blind to tail-end risks—for example, the subprime crisis. Rather than making us safer, could the reliance on risk modeling by regulators actually be paving the way for the next systemwide financial crisis?