QuarcHome
Model catalogueBack to the model index

CreditRisk+

Models default counts and portfolio losses with sector-dependent intensities and generating functions.

How it works

CreditRisk+ builds a default-loss distribution from exposures, default frequencies and sector dependence. Small individual default probabilities motivate a Poisson count approximation. Random sector intensities allow counts to become more dispersed than for independent issuers at fixed rates. A generating function or recursion combines counts and loss sizes after discretisation.

Worked example

As a baseline before adding random intensity, take 100 equal borrowers with 1% annual probability each and approximate the default count by Poisson with mean 1. With 60 units lost per default, mean loss is 60. The probability of at least two defaults is 1 − 2/e, about 26.4%. A model with uncertain sector intensity can have the same mean and a different tail.

Validation and limitations

Check exposure-band rounding and benchmark with exact Bernoulli counts for small books. The original default-loss setup leaves migration repricing outside its scope. Compare the unconditional and stressed count distributions and consider exposure and recovery dependence when the portfolio is concentrated.

Research sources