Skip to content
All library documents

Modeling Random Credit Recovery with Copulas and Beta Distributions

Article Quant Q&A · Author: quant_dev
Read the original

Summary

The document addresses how to represent uncertain recovery rates after a credit default when pricing credit derivatives. It identifies random recovery as an extension to Gaussian copula credit models, alongside random factor loadings. The cited discussion connects stochastic recovery to calibration difficulties encountered with standard one-factor base-correlation models during the 2007–2008 period.

A second approach mentioned is CreditMetrics, which uses Monte Carlo simulation with a beta distribution fitted to historical recovery rates. These are brief pointers to references rather than a full specification: the document gives no calibration recipe, data requirements, comparison of model performance, or treatment of dependence between default and recovery. The beta distribution is presented as one modeling choice, while the copula extension is offered as a source for further study; neither is shown here to be universally appropriate.

Key ideas

  • Recovery rates can be modeled as random variables when valuing credit derivatives.
  • Gaussian copula extensions include random recovery and random factor loadings.
  • Stochastic recovery was discussed in connection with base-correlation calibration problems in 2007–2008.
  • CreditMetrics is described as using Monte Carlo simulation with a beta distribution fitted to historical recoveries.
  • The document gives references but no detailed implementation or comparative evidence.

Tags

Fixed incomeDerivatives pricingStatisticsRisk management
Full text
# Stochastic recovery rates


# Stochastic recovery rates












How do I model the randomness of recovery rate given default when pricing credit derivatives?

## Answer by ldnquant (score 5, accepted)

https://quant.stackexchange.com/a/488

The standard reference is Anderson and Sidenius Extensions to the Gaussian Copula: Random recovery and random factor loadings. Random recovery proved necessary in 2007/2008 when you couldn't calibration standard one factor base correlation models. This paper discusses this, and might be an easier starting point than the Anderson and Sidenius paper.

## Answer by Owe Jessen (score 2)

https://quant.stackexchange.com/a/472

CreditMetrics uses monte carlo simulation assuming a beta-distribution fitted to historical recovery rates.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.