Credit Metrics™ was formed in response to the unique challenges posed by the nature of credit within a portfolio context. It had its foundation within the Morgan Guaranty Trust Company in the 1980s. It is now an independent consultancy owned by JP Morgan Chase and co-sponsored by five of the leading banks.
Credit Metrics™ was ‘released’ in the same spirit as Risk Metrics™.
That is to make the underlying methodology freely available. At the time it was perceived as a very novel way of running a venture particu- larly when that business is often concerned with very proprietary approaches. However the strategy is one of conferring leadership in an area which is one of the leading challenges in the financial market- place. After all this is clearly an area that needs leadership.
The philosophy can be viewed as a portfolio approach, as opposed to traditional ‘bottom up’ credit analysis, and as such the heart of credit metrics is the assessment of the changes in value of a portfolio, con- sisting of potentially quite disparate asset classes, due to changes in the credit quality of the issuers. The application, Credit Manager™ which is the technological manifestation of this approach, addresses both credit migration and default within a portfolio context.
The technology presented a systematic quantitative portfolio approach to credit risk management, enabling the user to measure concentra- tion risk within a portfolio. (This can be described as the risk due to exposure to one or more groups of obligors by sector or location.) It is not sufficient just to be able to measure concentration exposure. The next step is to manage the amount of additional risk. This can be achieved through a programme of appropriate diversification; credit metrics will suggest a number of ways to modify the portfolio which reflect existing credit lines and limits.
We wish to refer you to the Section 6.3 in Chapter 6. This goes into detail about the inherent difficulties of portfolio credit risk modelling, however just to recap on the two thorny problems. The first is the distribution of credit returns, which I would hazard a guess and say that every man and his dog knows by now that these are not normal (Figure 1.36).
If you need a brief refresher on the shape of this profile, then in the diagram below we have the credit returns on a portfolio. Notice that they extend much further out than the returns due to market risk which are highlighted as the broken line. This can be put into a slogan that credit returns are characterized by a fairly high chance of making a small profit together with a fairly small chance of losing a large investment.
The other major bugbear when it comes to looking at credit in a portfolio context is the difficulty of assessing correlation. This is com- paratively easy for market risk on highly liquid instruments because there is ample data available. This contrast is painted by most authors in black-and-white with the implication that market risk is done and dusted and credit risk is tough. The reality though is shades of grey.
Correlation within a market context is still historic which must be qualified as a basis for the future. For less liquid instruments even this directly inferred number is questionable. Credit correlation as applicable to a fixed income ‘marked to market’ business, can be measured quite well and adequately captures the risk due to changes in spreads, but default correlation is perhaps the greyest arena.
We press on now with a programme of understanding and applying Credit Metrics. The route map provided by the group is reproduced as shown in Figure 1.37.
Credit rating
Volatility of value due to credit changes on a single exposure Credit spreads Seniority
Volatility of value due to credit changes on a portfolio E
x p o s u r e s
C o r r e l a t i o n s
Migration Recovery Revaluation
Figure 1.37 The credit metrics route map. Source:Credit Metrics™.
Portfolio loss distribution
0 0.05 0.1 0.15 0.2 0.25 0.3
0 2 4 6 8 10 12 14
Losses (%)
Probability
Figure 1.36 The losses of a portfolio.
Start
Possible outcomes
Figure 1.38 The standalone bond and its possible states.
Table 1.18 Migration probabilities. Source:Standard and Poors, 2002.
Rating Outcome
AAA AA A BBB* D
AAA 89.37% 6.04% 0.44% 4.16% 0%
AA 0.57% 87.76% 7.30% 4.36% 0.01%
A 0.05% 2.01% 87.62% 10.26% 0.05%
BBB 0.03% 0.21% 4.15% 95.24% 0.37%
D 0 0 0 0 1
*We have aggregate probabilities between investment grade and default partially for pedagogical purposes and further due to the lack of spread data on Euro denominated speculative grade credit.
We give a brief example on the application of this framework con- sisting of just one bond. It is a AA bond with a maturity of 10 years and a coupon of 5 per cent. We want to know the impact of credit risk on its value in 1 year? The starting point is to consider the states that the bond could migrate into, there are four possibilities. It either gains value, loses (due to spread widening), stays the same or defaults. These outcomes are depicted graphically as shown in Figure 1.38.
We then use the standard data from the agencies to furnish us with the probabilities for evolution. We reproduce just such a Table 1.18.
The next step is to revalue the bond in each of the states. This was achieved using the Bloomberg Fair Market curves on trade date 30/12/2002. We took the starting yield from the 10 year maturity AA Table 1.19 The 6 year bond using credit metrics.
Revaluation for the AA bond
State Probability Relative Accrued Bond Total
(%) spread (bps) (€) value (€) value (€)
AAA 0.57 13 5 102.23 107.23
AA 87.76 13 5 102.23 107.23
A 7.30 38 5 98.59 103.59
BBB 4.36 121 5 93.00 98.00
Default 0.01 Very large 0 0 0
curve and the ending yield at the 9 year tenor from the curve repre- senting the evolved state. This enables us to revalue the bond, by a simple duration calculation. We then add on the coupon to get the total valuation displayed in the final column of Table 1.19.
Finally to determine the expected value of the bond we weight these terminal values by the probability of landing in the state:
You will encounter a more realistic example incorporating the com- bination of different asset classes within Section 2.13 in Chapter 2.