|
Das Dokument ist frei verfügbar |
|
| Nachweis | Kein Nachweis verfügbar |
|
The use of probability of default estimates to assess the risks of a credit portfolio should not ignore estimation uncertainty. The latter can be quantified by confidence intervals. But assumptions about dependencies of these intervals are inconsistent with assumptions of conventional credit portfolio models. Based on simulation studies this paper shows that a model which include estimation uncertainty but ignore default correlation might estimate the real credit risk more correctly than a model that implicates default correlation but ignore estimation uncertainty. The latter is a trait of conventional credit portfolio models. In this paper quantifying of estimation uncertainty based on the idea of confidence intervals and the underlying probability distributions of these intervals. -- probability of default ; estimation uncertainty ; risk assessment |
|
|