Rating agencies have always had different views of asset correlation assumptions, so when it comes to rating synthetic CDOs, methodologies vary so widely investors have felt less than concrete with the guidance given.
Default correlation, a measure of how credits in a portfolio perform together, appears to be a sticky point. As a result of varying correlation assessments, Fitch Ratings, Moody's Investor's Service and Standard & Poor's all have different attachment points when rating synthetic CDOs. Despite the various opinions, fueled by the tight spread environment, synthetic CDOs are rising in popularity; it seems each quarter, investors face an onslaught of new CDO-of-CDOs, for example.
The CDO research team at Wachovia Securities delved into this issue last week, issuing research titled The Young and the Restless: Correlation Drama at the Big Three Rating Agencies.
"The report is very timely because synthetic CDOs are becoming more and more popular and there is some inconsistency in the way the rating agencies are looking at correlation," said Wachovia Managing Director Arturo Cifuentes. "They have conflicting views."
That said, the agencies are quick to point out they haven't been asleep at the wheel.
"We are currently in the midst of extensive research on a variety of assumptions to our model," reports S&P spokesman Adam Tempkin.
"We have done a great deal of research and we are now comfortable with the correlation assumptions we have and the methodology and analysis we have used to achieve them," added Moody's Managing Director Yuri Yoshizawa.
The current practice is to use equity return correlation as a proxy for industry-level asset correlation. These numbers are plugged into popular Monte Carlo simulations but Wachovia researchers found one exception to the standard view that high correlation is good for junior tranches and bad for senior tranches.
In the report, penned by Analyst Natasha Chen, Wachovia looks at correlation from a loss perspective. But when the synthetic CDO is viewed by a performance measure, such as expected internal rate of return, research shows that the expected IRR of a senior tranche is generally unaffected by high correlation, whereas the expected IRR of a junior tranche is dramatically reduced by high correlation.
"The results for the junior tranche were the most surprising," Chen explained.
Basically, the lower the tranche, the greater the difference in IRR means. Senior investors, CDO researchers found, are barely affected by correlation and junior investors "should actually prefer low correlation," said the report. Wachovia recommends that investors seeking a more complete view of performance should also analyze Monte Carlo-based IRR results.
Copyright 2005 Thomson Media Inc. All Rights Reserved.