 |
 |
|
News & Events |
|
|
|
|
|
|
 |
|
Media Coverage |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Media Relations |
|
|
|
We are always interested in speaking to members of the media on news and events affecting the world of corporate credit risk. We look forward to hearing from you.
Contact Media Relations Representative |
|
|
|
|
|
|
|
|
Barra joins crowded market for Merton models
Source: Risk Magazine
Date: November 2003
Author: John Ferry
Investors say Merton-type models for estimating credit default probabilities are now so common they are driving pricing in the credit market. And there are more models on the way.
The market for predicting bond defaults is getting increasingly crowded. It's easy to see why. Some big investors in the US say anyone wishing to play the bond markets needs to take into account the outputs of these models if they want to stay on top of the game. And that creates more sales for model suppliers such as Moody's KMV and RiskMetrics. The latest to enter the field is Berkeley, California-based Barra, which plans to release a competitor product next month.
Jean-Martin Aussant, director of fixed-income strategy at Barra in London, says his company's system will extend the Merton framework, which infers the probability of default of an individual company from its equity market valuation, in a new way that will make it more reliable as a default predictor.
"We think this is one step beyond the Merton-type models," he says. Barra's model, called Barra Credit, will try to combine the Merton framework with information from cash and derivatives markets to come up with a default probability. "The product will combine all the possible sources to analyse an issuer's creditworthiness - information from the equity market through default probabilities, information from the bond market through the market implied ratings, and information from the derivatives market through credit-default swaps (CDS) data."
Aussant says Barra's model also relaxes an assumption inherent in Merton models that will make it more accurate. In the models offered by his competitors, the level of debt that will trigger a default is fixed. "We don't think this is very realistic in light of what investors can see in the market - the exact level of liabilities, for instance, that will put a firm into default," says Aussant.
Barra's model allows for the varying of this trigger level, which Aussant says is crucial, because public information that is available with regard to a company's liabilities may not be correct. "If you look at the recent cases of Tyco or WorldCom, the liabilities disclosed were not the right ones. So you have some doubt on the numbers, but you can actually adjust the model for that uncertainty."
Probability distribution
Rather than having a fixed level of debt, Barra's model uses historic and current levels of company debt ratios to come up with a probability distribution for the trigger level. Further, the user decides how wide this distribution should be. If a user believes there is a lot of uncertainty surrounding the publicly available information on a company's liabilities, then the range of numbers that will trigger a default can be increased, and vice versa if they believe there to be relatively little uncertainty surrounding a company's financials.
"The fact that you can vary the trigger range will bring the model much more in line with what you see in the market in terms of credit spreads," says Aussant.
Barra is entering the market at a time when leading investors say Merton models are now so frequently used that they are actually driving the credit market. Take Barclays Global Investors (BGI) in San Francisco, with $780 billion under management. BGI has been using Merton models - an in-house model as well as those available on the market - for a number of years, but its head of US fixed-income, Peter Wilson, says the model is now becoming so prevalent with other investors that it is starting to dictate which way credit markets will move.
"As people start to use the same types of tools, pricing in the market starts to reflect those tools, and they actually affect the market they are being used in," says Wilson, adding that Moody's KMV has "almost become a brand name".
One of those to jump on the bandwagon more recently is the Columbus, Ohio-based financial services company Nationwide, which implemented Moody's KMV last year. "It's now a disadvantage if you don't have it," says Bill Chang, a Nationwide quantitative analyst.
So if Merton models are now driving the credit market, does this mean investors can discard traditional fundamental or other qualitative analysis in favour of the purely quantitative? BGI's Wilson says it is still important to use a combination of techniques. "I would put Merton models more in the category of understanding risk, then I would put other analysis in as helping us predict value," he says. "What matters in terms of managing returns is what's going to happen, and the Merton model itself isn't going to tell you that."
The Merton model is thoroughly entrenched at BGI, but it has a formal system in place to take account of all forms of analysis. It has one team engaged in looking at what Wilson calls credit signals - using forward-looking data to predict how a company and its corporate debt will perform. Another team uses traditional fundamental credit analysis, while another is engaged in what Wilson terms credit analytics, which involves understanding the mechanics of how corporate bonds will perform, for example, under different asset volatility scenarios. It is this team that uses Merton models. "We put them all together and make an investment decision within the portfolio," says Wilson.
Ethan Berman, New York-based chief executive of RiskMetrics, says his company always highlights the importance of fundamental analysis to complement the Merton model. He says he would be the first to tell investors that they should use CreditGrades along with their own qualitative assessments.
Berman says Merton models will either give forward-looking indicators or misleading indicators depending on market conditions. "If you are looking at both qualitative and quantitative measures and they are telling you the same thing, then you should have a higher level of confidence. If one of them is telling you something different from the other, then you need to do more analysis."
Sensible system
BGI's system seems sensible, especially when Merton models do not always give off the 'right' signals on credit, even though market participants often feel that the models themselves drive the spreads. For example, in Europe, during March, immediately before the war in Iraq, equity prices plummeted, and according to Merton models credit and CDS spreads should have widened. But they did not.
JP Morgan Chase and Morgan Stanley's CDS index for Europe, Trac-x Europe, narrowed from 140 basis points in October 2002 to 105bp during the war, while the Dow Jones Eurostoxx 50, which was at 2,275 in early October 2002, reached a low of 1,928 in March. The apparent discrepancy was largely due to the fact that a lot of European companies were reducing their debt levels at this time, according to Viktor Hjort, a fixed-income analyst with Morgan Stanley in London. The credit market took the view that corporate debt reduction was a stronger indicator of company performance than the negative macroeconomic indicators that were pummelling stock levels.
Meanwhile, in the US, falling equity volatilities mean Merton models are not giving out strong signals on default probabilities. "KMV-type models raise a red flag when equity prices are falling and volatility is rising rapidly, both of which have not been happening in the US," says Sivan Mahadevan, a fixed-income analyst at Morgan Stanley in New York. The table below shows this. Morgan Stanley analysts measured the effect of a nine-point drop in equity implied volatility, and a 13% rise in stock prices for this year against corresponding Moody's KMV default probabilities - trademarked as expected default frequency (EDF) measures. They found that the EDF measures fell by more than 40 basis points for the group, while credit spreads rallied 74bp. The smaller EDF numbers mean the signals produced by Moody's KMV measures are not as pronounced as they were last year.
But Mahadevan adds that, in the past year, Merton models have become "part of the mainstream culture" for investors. And indeed it is on the buy side where Barra believes there is a gap in the market. "All the portfolio managers we've met have been interested in this product," says Aussant.
But Tim Kasta, Moody's KMV managing director, says some of the biggest growth in sales for his company in the past two years has been from the buy side, especially hedge funds, proprietary trading desks and funds of funds.
"The challenge of getting good default probabilities is how extensive your default database is, and we're unaware of any competitor that has the breadth and depth of ours," he adds.
Nationwide's Chang says he has been in talks with Barra about its new model, but he is sceptical about the chances of success for a new entrant. "KMV will not stand still," he says. "I'm confident they have enough research power to keep up with the market."
Merton vs the market
| |
Current |
Dec. 31, 2002 |
Sep. 30, 2002 |
YTD change |
One-year change |
| CDS premium (bp) |
78.0 |
152.0 |
211.0 |
-74.0 |
-134.0 |
| Mkt imp def rate (%) |
1.3 |
2.5 |
3.2 |
-1.2 |
-1.9 |
| Imp equity volatility (%) |
32.6 |
41.4 |
54.9 |
-8.8 |
-22.3 |
| Equity return (%) |
|
|
|
13.1 |
18.7 |
| Moody's KMV EDF (%) |
0.41 |
0.83 |
0.93 |
-0.42 |
-0.52 |
| Mkt imp def rate/EDF |
3.1 |
3.1 |
3.5 |
0.1 |
-0.3 |
Note: universe of 120 equally weighted issuers
Source: Morgan Stanley, Moody's KMV
Back to Media Coverage Archive |
|
|
|
