In its series of Consultation Papers on Level 2 Implementing Measures for Solvency II, CEIOPS drafted a new proposal for the calculation of counterparty credit risk. While Consultation Paper 28 (March 2009) gives a general overview of the proposal, the more recent Consultation Paper 51 (July 2009) provides insight into the details of the model.
Counterparty default risk, commonly called credit risk, is one of the core risk components of the Solvency Capital Requirement (SCR). Contrary to the insurance or market risk modules, which have not suffered dramatic modifications from the various quantitative impact studies (QIS) and appear quite well accepted by the QIS participants, the counterparty default risk module has been changed continually in search of a sufficient modeling solution throughout the industry.
Although considered an improvement compared to QIS3, the QIS4 approach was not deemed adequate in all circumstances. The main criticism was that the Vasicek-Herfindahl approach used in QIS4 produced counterintuitive results: the inclusion of additional reinsurers in the portfolio was penalized with a higher capital requirement. Participants also criticized the high default probability assigned to unrated counterparties and argued that the calculation of the loss given default is burdensome. Therefore, CEIOPS proposes a new approach with two substantial methodological changes and refinements in respect of the calculation of the loss given default and the probability of default.
Classification of Unrated Counterparty
The first main change relates to the origin of the counterparty. To better reflect the details associated with different types of unrated exposures, CEIOPS proposes to differentiate between Type 1 exposures (e.g., reinsurers, derivatives, and securitizations) and Type 2 exposures (e.g., intermediary receivables and policyholder debtors). Type 1 exposures will be treated just like rated exposures, and Type 2 exposures will be evaluated by means of a simplified factor-based formula.
Redesign of Credit Risk Approach
The second change is a complete redesign of the credit risk approach. The Vasicek-Herfindahl concept did not produce consistent results, but the newly proposed model is deemed to be free of the issues identified in QIS4 and addresses the limited number of counterparties a (re)insurer generally has.
Whereas the QIS4 model implied the calculation of a capital charge for each counterparty individually, the new model evaluates the variance of the total portfolio’s loss distribution. The capital requirement is then the variance of the total portfolio multiplied by a fixed factor in order to estimate the 200-year default event.
This new model has two key aspects. The first aspect is a random shock that affects all counterparties simultaneously; for instance, a large catastrophe, subprime crisis. The random shock leads to an implicit correlation between the default probabilities of the counterparties. The second aspect concerns the vulnerability of the counterparty to this random shock, making the default probability of the counterparty directly affected by the shock.
Loss Given Default
CEIOPS proposes a few adjustments to the calculation of the loss given default, which was considered burdensome in QIS4. One straightforward adjustment is that the recovery rate should be lowered to 40 percent (from 50 percent in QIS4) due to the lessons learned from the financial crisis and in absence of better evidence.
In addition, simplifications are necessary to assess the risk-mitigating effect of reinsurance treaties and for the calculation of the loss given default when the number of counterparties is large.
With respect to the risk-mitigating effect, it may be calculated jointly for all counterparties and the share of a single counterparty may be assessed by a simple proportional approximation.
A simplification when the number of counterparties is large allows for homogenous grouping of counterparties to decrease the number of iterations. These groupings, though, will produce more conservative results than the long calculation process.
Probability of Default
CEIOPS considers that, despite some deficiencies, there is currently no alternative to having credit rating agencies assign a default probability. Therefore, ratings assigned to counterparties by rating agencies will remain the key measure to assess the probability of default of a counterparty.
For Type 1 counterparties without ratings, two approaches to evaluating the probability of default are suggested. The first one relies on the publicly available SCR and own funds to derive the default probability. This approach should be applicable whenever the counterparty is subject to Solvency II supervision. It is to be applied with caution, though, since the SCR does not provide a long-term average probability of default, as is the case with ratings assigned by rating agencies, but is rather a snapshot at a given point in time. Until the full implementation of Solvency II, it is suitable to use a fixed default probability corresponding to a BBB rating.
The second approach assigns a fixed default probability of 10 percent to counterparties that are not subject to Solvency II supervision or do not meet the requirement of the solvency ratio rating.
- Susan Witcraft, Managing Director, Minneapolis
- Frank Achtert, Managing Director, Munich
- Iain Boyer, Managing Director, Norwalk
- Michelle Harnick, Managing Director, New York
- Dave Lightfoot, Managing Director, Seattle
- Scott Lohman, Managing Director, Seattle
- Don Mango, Managing Director, Morristown
- Eddy Vanbeneden, Managing Director, Brussels
- Jeff Bellmont, Senior Vice President, Minneapolis
- Gina Carlson, Senior Vice President, Minneapolis
- Debbie Griffin, Senior Vice President, New York
- David Flandro, Senior Vice President, London
- Benoît Butel, Vice President
- Sebastien Portmann, Vice President, Zurich