Is the regulatory downturn LGD adequate? Performance analysis and alternative methods

This paper addresses the drawbacks of macroeconomic approach to estimate downturn loss given defaults (LGD) that is currently used in EU banking regulation. Using a large dataset more than 186.000 defaulted credit exposures covering time span 18 years we demonstrate by Monte Carlo simulations proposed European Banking Authority (EBA) does not reach confidence level 99.9%. As an alternative develop LGD estimation based on latent variable approach. Our this outperforms two dimensions: It induces capital requirements are sufficient cover potential losses at 99.9% and compared all regulatory eligible approaches provide coverage, our requires least amount capital. The approach, developed applicable for market LGDs as well workout LGDs, i.e., with long period. We argue regulation banks’ risk management can be improved if addition traditional estimating LGDs.