Are through-the-cycle credit risk models a beneficial macro-prudential policy tool?

Abstract

Credit risk models are validated to check that they produce unbiased, "high-quality" estimates of credit risk. Credit risk models follow different rating philosophies, ranging from point-in-time (PIT) models that reflect all currently available information to through-the-cycle (TTC) models whose credit risk estimates are independent of cyclical changes in macroeconomic conditions. TTC models have been favoured in particular from a macro-prudential perspective because they produce more stable capital requirements for banks over the cycle, thus avoiding pro-cyclicality. This paper investigates different ways to validate TTC credit rating systems, including possibilities to separate the validation of a TTC system into the validation of its PIT component and the validation of its adjustment for the cycle. We conclude that the validation of TTC models is significantly more difficult than the validation of PIT models, which may make the regulatory promotion of TTC models questionable. We argue that the regulatory requirement of PIT models combined with a more extensive use of the counter-cyclical capital buffer as a macro-prudential policy tool could be a potentially superior alternative to address pro-cyclicality.