Title: Modeling exposure at default under varying systematic conditions
Authors: Daniel Roesch - University of Regensburg (Germany)
Jennifer Betz - University of Regensburg (Germany)
Maximilian Nagl - University of Regensburg (Germany) [presenting]
Abstract: In the advanced internal ratings based approach, banks are allowed to use own estimates of exposure of default to determine their regulatory capital. For volatile segments, downturn estimates i.e., estimates which reflect economic downturn conditions are demanded. Furthermore, banks are obliged to base their models on credit conversion factors for credit lines. This regulatory setting might be challenging. First, the distribution of credit conversion factors is highly bimodal and reminds of loss rate distributions. Second, downturn estimates require an adequate consideration of systematic effects which might be non trivial. A unique data set of defaulted credit lines from the U.S. and Europe is used, and a quantile regression approach is applied to model conditional distributions of credit conversion factors. If macroeconomic variables are incapable of revealing the true systematic patterns, the model is enhanced by random effects to ensure adequate downturn estimates. To the best of the authors' knowledge, this is the first attempt to provide a sound framework to model the full conditional distribution of credit conversion factors which ensures adequate downturn estimates considering observable and unobservable systematic variables.