Standard & Poor’s changes to its collateralized debt obligation rating methodology, primarily in the form of lower assumed default rates for investment grade assets rated A and above, are significant and will have a large impact on the CDO market. The main implication of these changes is that the subordination required for a tranche to achieve a rating has dropped, and thus tranche spreads for a given rating have increased. In absolute terms, the spreads of AAA and AA tranches of 10-year portfolios benefit the most from these changes. In relative terms, the spreads of five year tranches benefit the most. We expect more CDOs will be backed by high-quality portfolios, and fewer CDOs will contain mostly BBB credits. With investment-grade CDO demand standing at USD485billion last year, these changes are likely to affect relative pricing by rating and industry.
We discuss these changes, as well as the impact they are likely to have on the types of tranches that come to market and the makeup of the underlying collateral pools.
Changes In Default-Probability Assumptions
Assumed default rates have been reduced for collateral rated A and above, but remain the same for collateral rated BBB through B. Collateral rated A experienced the largest change--the five-year default probabilities dropped by 63 basis points, from 114 bps to 51 bps. Similarly AAA and AA cumulative default probabilities are 41 bps lower, going from an average of 57 bps to 16 bps. In high yield, the only significant change was to CCC collateral, which is assumed to have a significantly higher five-year default probability--it has been increased from 40% to close to 70%.
At a simple level, and without accounting for portfolio/correlation effects, a measure of the attractiveness of a reference entity is the ratio of its spread to its default probability. That is, we divide the average five-year credit-default swap spread within a given rating by the annualized default probability of that rating category. While the ratio was in the 0.8-1.0 range using the old methodology, under the new methodology, that range has increased to about 2.7 for AA reference entities, but remains unchanged for BBB names. Higher-quality underlying portfolios are relatively more attractive.
Other Model Changes
In general, the new rating methodology assumes investment grade defaults are more back-loaded, i.e., a larger proportion of the losses are assumed to occur close to maturity.
S&P raised its inter-industry correlation measure to 5% from 0%, and decreased intra-industry correlation to 15% from 30%. This has a significant impact on the required subordination of CDO-squared transactions. These trades have become much less common. In this enviroment investors are more focussed on combining vanilla CDOs and good credit picking.
Finally, S&P has moved to stochastic recovery rates. This change is important for deals backed by high-yield collateral, which has higher underlying loss rates.
Impact On Spreads, Subordination Levels
Reduced default probabilities allow tranches to achieve the same rating with less subordination, and thus pay higher spreads. Importantly, this only applies to highly rated portfolios. For example, seven-year AAA tranches now require 1.8% less subordination. The required subordination for a 10-year A tranche has decreased from 7.1% to 4.9%. Figure 1 contains subordination levels for rated tranches across tenors of an optimized high-quality bespoke CDO under the new and old rating methodologies. The underlying collateral is a portfolio of mostly A credits with an average five-year spread of 35 bps.
As a result tranche spreads increased, see figure 2. On an absolute basis, the changes are largest for higher-rated tranches of longer-dated CDOs. Five-year spreads benefit most on a relative basis, with AAA spreads at 50bps (+294%).
We expect the new methodology to lead to increased issuance across tenors, with a greater emphasis on higher-quality portfolios.
Tighter mezzanine spreads drastically reduced five-year CDO activity after May 2005. The new S&P methodology results in significantly higher five-year tranche spreads, a feature that could increase five-year issuance, especially if spreads widen from their current levels.
| Figure 1: Required subordination levels for an optimized portfolio | ||||||
| 5 year | 7 year | 10 year | ||||
| Methodology | Old | New | Old | New | Old | New |
| AAA | 4.70% | 3.20% | 5.70% | 3.90% | 7.10% | 4.90% |
| AA | 3.80% | 2.80% | 4.60% | 3.40% | 5.80% | 4.40% |
| A | 3.10% | 2.50% | 3.90% | 3.00% | 5.00% | 3.90% |
| BBB | 2.50% | 1.80% | 3.10% | 2.30% | 3.90% | 3.10% |
| BB | 1.30% | 1.10% | 1.80% | 1.60% | 2.60% | 2.20% |
| Spreads as of 3/6/06 | ||||||
| Source: Barclays Capital | ||||||
New CDOs: Fewer Industry Sectors, More Global
The new correlation assumptions are likely to mean the use of fewer industry sectors within CDO portfolios, but greater diversification within industries. We also believe that the portfolios will be more global in nature. In other words, CDO portfolios are likely to more closely resemble corporate bond indices.
The changes in default probabilities under the new methodology should also lead to a change in the mix of industries and credits that are included in CDO portfolios. We expect a substitution away from BBB-focused industries, such as utilities, industrial goods, consumer goods and technology, media and telecommunications, into industries with a larger share of issuance rated AA and A, such as financials and insurance.
Impact On Standardized Tranche Market
The dynamics of the 3-7%/3-6% tranche of the 10-year index have potentially changed. This tranche has been used as an equity equivalent: according to the base correlation model it is long correlation. Bespoke mezzanine tranches mapped to the 6-9%/7-10% tranche of the 10-year index, requiring dealers to buy protection on the 10-year 3-6%/3-7% tranche. With less subordination now required of the rated bespoke tranches, lower-rated 10-year tranches are likely to require dealers to sell protection on this tranche.
The response of this tranche to mezzanine prints may have changed: the equivalent index tranches for a typical high-quality bespoke portfolio with average credit quality of A sits either between the 3-7% and 7-10% tranche or fully inside of the 3-7% tranche.
Despite this change, we remain bearish on the 3-7%/3-6% 10-year tranche. Firstly, leveraged super-senior trades put upward pressure on the 10-year 3-7%. Secondly, AAA tranches still map primarily to the 7-10%. To the extent that demand in the 10-year sector is focussed in AAA tranches, the effect we discuss will not materialize. Finally, the cap on 10-year equity spreads means that spread widening could shift risk into the 10-year junior mezzanine tranches. Increased interest in AA and A tranches, which map into junior mezzanine tranches could change these dynamics.
| Figure 2: Spread of tranches of an optimized portfolio | |||||||||
| 5 year | 7 year | 10 year | |||||||
| Old | New | Chg | Old | New | Chg | Old | New | Chg | |
| AAA | 17 | 50 | 294% | 59 | 87 | 147% | 53 | 194 | 366% |
| AA | 38 | 88 | 232% | 77 | 137 | 178% | 116 | 249 | 215% |
| A | 62 | 124 | 200% | 118 | 183 | 155% | 186 | 309 | 166% |
| BBB | 117 | 258 | 221% | 207 | 353 | 171% | 306 | 448 | 147 |
| Spreads as of 3/6/06 | |||||||||
| Source: Barclays Capital | |||||||||
Rating Tiering
These changes also further increase the divergence between the CDO rating approaches used by Moody’s Investors Service and S&P, such as expected loss versus probability of loss and binomial expansion versus Monte Carlo simulation. Given that Moody’s is looking to change the recovery assumptions for high-yield debt, we expect these differences to increase. As a consequence we expect to see investors increasingly seeking ratings from different rating agencies.
This week’s Learning Curve was written by Lorenzo Islaand Jeff Meli, structured credit strategists at Barclays Capital in London.