Charge-offs, Defaults and U.S. Business Cycles

Abstract

We uncover a new fact: U.S. banks counter-cyclically vary the ratio of charge-offs to defaulted loans. The variance of this ratio is roughly 15 times larger than that of GDP. Canonical financial accelerator models cannot explain this variance. We show that introducing stochastic default costs into the model helps to resolve the discrepancy with the data. Estimating the augmented model using Bayesian techniques reveals that the estimated default cost shocks not only help account for the variance of the banking data but also help account for a significant fraction of the U.S. business cycle between 1984 and 2015.

Keywords: charge-offs and defaults, default cost shocks, financial accelerator models, business cycles

JEL Classification: E3, E44