Internal Loan Ratings, Supervision, and Procyclical Leverage
Pinar Uysal1, Lewis Gaul2, Jonathan Jones2, Stephen Karolyi2
1Federal Reserve Board; 2Office of the Comptroller of The Currency, United States of America
We build a Markov model of banks' internal loan ratings to illustrate the relationship between ratings inflation and systematic ratings drift. Using administrative data from the Shared National Credit (SNC) Program, we find evidence of systematic downward drift in ratings, consistent with initial ratings inflation. The drift is predictable based on pre-issuance borrower characteristics, suggesting that screening and pricing information is not being fully incorporated into ratings. We use the conditional random assignment of loan examinations to study the causal impact of loan-level supervision on ratings, and find not only that supervision reduces ratings inflation, but also that these effects spill over within a bank's loan portfolio, consistent with learning. We employ our model to investigate various counterfactual capital ratios and provide new insights about the relationship between bank supervision in bank capital cyclicality.