Measure for measure: evidence on the relative performance of regulatory requirements for small and large banks

This paper compares the performance of regulatory thresholds as predictors of distress for large banks with their performance for small banks. Using a data set of capital and liquidity ratios for a sample of UK‑focused banks in 2007, we apply simple threshold-based rules to assess how regulatory thresholds might have identified banks that subsequently became distressed. We compare results for large banks with results for small banks, optimising thresholds separately for the two groups. Our results suggest that the regulatory ratios we use are better aligned with risks which cause distress of large banks than with those which cause distress of small banks. We find that when thresholds are set to correctly identify a high proportion of banks which subsequently became distressed, they generate materially lower false alarm rates for large banks than for small. This result is robust to definitional choices and to resampling. We also test whether supervisors’ judgements about the quality of banks’ governance have predictive power with regard to distress. We find that adding supervisors’ judgements to regulatory ratios improves predictions for small banks but not for large banks.