This article looks at correlation, default rates and recovery assumptions in structured finance CDOs. Last week's examined the basic features and why they are attractive to investors. The analysis of default correlation for structured products suffers from a lack of historical data. Structured finance securities are backed by diversified asset pools, which by definition are less exposed to idiosyncratic events. The lower the idiosyncratic events in the portfolio, the higher the correlation of these asset classes. For this reason, correlation between ABS usually is considered higher compared to corporate credits.
The first step in calculating the diversification for structured finance CDOs is to categorize the collateral pool according to asset types that are expected to be driven by the same market and underlying economic factors.
The second stage is to look at the level of correlation between these groups.
A well known measure of diversification, usually introduced as a collateral test in numerous CDO transactions, is the Moody's Investors Service Alternative Diversity Score.
For structured finance CDO, Moody's has developed an alternative approach from the traditional (i.e. corporate) calculation of Diversity Score. Such an approach takes into account the default probability of a given asset as well as its correlation to other assets.
Moody's default correlation is associated with the credit quality of the underlying, with investment-grade securities assigned lower levels of correlation than sub-investment-grade ABS. Hence, other things being equal, the rating of the underlying portfolio has an impact on the calculation of the diversity score for a structured finance CDO. In other words, a pool of AAA RMBS should have a higher diversity score than a pool of BBB RMBS. That is not true for corporate credits where the diversity score is independent from the rating.
In addition, the link between performance of the defined groups of structured securities is taken into account. For example, auto and credit card ABS are assumed to be closely linked to consumer behavior and therefore highly correlated.
Moody's assumes default correlation between ABS securities can go as high as 40% for pairs of non-investment grade securities in narrowly defined sectors.
This approach makes it challenging for a structured finance CDO to achieve the same diversity score as corporate credit CDOs.
Fitch Ratings and Standard & Poor's utilize multi period Monte Carlo simulation models to mimic the behavior of the underlying portfolio based on assumptions of correlation between individual asset classes in a structured finance CDO. These simulation-based default-models assume that correlation can be expressed in terms of a pair-wise sector correlation table. Correlation assumptions typically are higher for ABS than for corporate credits.
In order to show the impact of correlation in the default analysis of a structured finance portfolio, we present a fairly basic case study within the S&P rating framework. We compared a portfolio of 65 BBB rated structured finance securities with a portfolio of 65 BBB rated corporate credits. We have assumed a weighted average maturity of 4.25 years in order to obtain the same expected default rate for the two pools: mean ~ 2.5%. The structured finance securities have been taken from the 13 available S&P structured categories in order to guarantee maximum diversification, while the corporate credits have been chosen from 13 different sectors (five local, four regional and four global). All the assets are equally distributed among five AAA rated countries (U.K., France, Germany, Canada and the U.S.) within Western Europe (30%) and North America (20%). As can be seen, the default density function for ABS assets is characterized by higher probability of default for extreme values. This leads to a higher level of expected default rates for senior rating scenarios. Other things being equal, credit enhancement requirements for high ratings will be greater for structured finance CDOs than for corporate credits.
Default Rates
The limited amount of data available makes the job of the analyst particularly challenging. The default experience, so far, of structured finance securities compare favorably to that of corporate credits and is probably not as sensitive to final maturity. A study published by Fitch in 2001 shows, that between 1989-2000 the average annual default rate for approximately USD1,500 billion of structured products (including investment and sub-investment grade ratings) was in the region of 0.01%. This is favorable when compared with investment-grade corporate credits, which saw an annual average default rate of 0.08% over the same period and sub-investment grade corporate credits with an annual default rate of 3.07%. The outcome of this research seems to be confirmed by the publication of transition matrices this year by the three major rating agencies that highlight lower volatility for asset-backed securities.
It is reasonable to assume that structural mitigants for asset-backed securities have played and will continue to play an important role in limiting the default rates compared to the corporate credit market. In particular, the issuer of virtually every ABS is a bankruptcy remote vehicle, thus allowing for the isolation of the pool of securitized assets from the originator/ seller and therefore from potential negative changes in the borrowers' operating environment and potential "event risk". This contributes to the lower degree of credit volatility as highlighted above.
Moreover, each individual security is structured to withstand a level of performance stressed to be consistent with the assigned rating (for example, typically BBB securities are structured to cover a level of defaults between two to three times the base case on the underlying pool). Finally most transactions include self-correcting mechanisms that preserve credit enhancement when performance deteriorates, such as early amortization events, delinquency and excess-spead trapping triggers.
Regardless of the above and in an attempt to keep consistency between the ratings of different CDO structures, Moody's and Fitch have not developed a default curve for ABS products and generally tend to use corporate default rates in rating structured finance CDOs. The outcome being in our opinion that ratings of SF CDOs, in many instances, are more conservative with respect to similar corporate credit CDOs. S&P, instead, employs specific ABS default rates, which reflect S&P's own investigation into default frequencies.
Recovery Assumptions
Recovery rates for ABS obligations depend on a security's priority within the capital structure, the credit rating of the respective tranche and the tranche within its own capital. All rating agencies assume to a different degree that structured finance securities have shown a lower level of losses compared to similarly rated corporate securities and adjust their analysis to take this into account. For example, Moody's will correct the idealized net cumulative losses (based on corporate defaults and a recovery of 45%) to allow for a different recovery rate, in order to calculate the gross cumulative default rate to be applied in the cashflow modeling of structured finance CDOs and derive the required capital structure.
Fitch and S&P will adjust the results of their Monte Carlo models for the appropriate levels of recovery.
Prepayment Sensitivities
Structured finance CDOs are more likely, than other CDOs, to receive pre-payments when the underlying borrowers redeem part or all of their obligations to take advantage of more convenient funding rates. When a bond prepays faster than expected, the notional value of the principal is reduced, thus, decreasing the expected future cash flow stream. When a bond prepays more slowly, cash flows are received later than expected. Since excess spread is often used to cover losses, prepayments have an important influence on the capacity of the structure to cover any default and must be considered carefully in modeling the underlying cash flows.
It is essential therefore to analyze prepayment scenarios, especially for structured finance categories in which payments are sensitive to funding cost movements (eg RMBS). All rating agencies conduct a stress analysis.
Concluding Remarks
CDOs of ABS or structured finance CDOs have seen a remarkable growth recently. In our view, this was driven by advantages both for investors and for collateral managers as originators of the transactions. The recent rally in corporate spreads, together with increased supply in the ABS market has made credit arbitrage in structured finance CDOs even more efficient. While SF CDOs are based on established CDO technology, they offer an attractive alternative and diversification to the corporate exposure in a traditional CDO. The returns of a structured finance CDO are driven by the performance of a pool of diversified assets, thus benefiting from two layers of diversification. This eliminates corporate headline or event risk almost completely and greatly limits the correlation of the SF CDO investment to specific industry risks. In times of turbulent credit markets these can be used as a defensive investment with attractive returns when compared to other asset classes of similar ratings.
This week's Learning Curve was written by Francesco Cuccovillo, manager in structured credit trading, and Markus Herrmann, director and head of asset-backed securities and structured credit research, at HSBCin London.
