Conditional Value at Risk Optimization of a Credit Bond Portfolio, a Practical Analysis

38 Pages Posted: 20 Jan 2004

See all articles by Albert Mentink

Albert Mentink

AEGON Group - AEGON Asset Management

Date Written: January 2004


Recently, research has been published in which optimal portfolios of credit risky bonds are determined that are less risky, while having at least the same expected return, using the CreditMetrics model. In this paper, we investigate whether the "optimal" bond portfolios are really an improvement by analyzing the characteristics of the individual bonds in the optimal portfolio. We find that a portfolio manager should be careful in following blindly the trades as suggested by the optimal portfolio because optimal portfolios are dominated by only one or two bonds. Moreover, the composition of such an optimal portfolio is very sensitive to small changes in the expected forward price of its main constituents. However, the portfolio optimization can be used in combination with some common sense restrictions to produce portfolios that both have a lower risk and higher return than a fully diversified portfolio. We also improve on the portfolio by replacing the dominant bond in the optimal portfolio by similar bonds. As a risk measure we use the Conditional Value at Risk, which at a given percentile equals the expected value of the losses that exceed the Value at Risk at that percentile. Conditional Value at Risk also provides information about the losses larger than the Value at Risk. Furthermore, the Conditional Value at Risk can be optimized using linear programming.

Keywords: Conditional value at risk, optimization, bond portfolio

JEL Classification: C61, G21

Suggested Citation

Mentink, Albert, Conditional Value at Risk Optimization of a Credit Bond Portfolio, a Practical Analysis (January 2004). Available at SSRN: or

Albert Mentink (Contact Author)

AEGON Group - AEGON Asset Management ( email )

P.O. Box 202
2501 CE The Hague

Register to save articles to
your library


Paper statistics

Abstract Views
PlumX Metrics