Regensburg 2004 – scientific programme
Parts | Days | Selection | Search | Downloads | Help
AKSOE: Physik sozio-ökonomischer Systeme
AKSOE 1: Finanzm
ärkte und Risikomanagement
AKSOE 1.2: Talk
Monday, March 8, 2004, 10:30–11:00, H8
Modeling Correlations in Portfolio Credit Risk — •Bernd Rosenow1, Rafael Weißbach2, and Frank Altrock2 — 1Institut für Theoretische Physik, Universität zu Köln, 50923 Köln, Germany — 2Risk Management Support & Control, WestLB AG, Düsseldorf, Germany
The risk of a credit portfolio depends crucially on correlations between the probability of default (PD) in different economic sectors. Often, PD correlations have to be estimated from relatively short time series of default rates, and the resulting estimation error hinders the detection of a signal. We present statistical evidence that PD correlations are well described by a (one-)factorial model. However, when using the model to generate short time series and calculating their correlation matrix, one typically observes large statistical fluctuations in the correlation structure. Due to these fluctuations, the parameter estimation for a one–factor model is plagued by large uncertainties. When estimating the model parameters in such a way that the empirically observed ones appear as a worst case scenario, the reliability of the estimate is increased in a systematic way, leading to a moderately increased CreditVaR.