Our low-tech, overcollateralization-based approach has the immediate advantage of being able to be performed now and without great expense. But such older methods and measures of CDO analysis have the disadvantage of focusing on descriptive details rather than financial conclusions. For example, a CDO portfolio’s WARF and diversity score
450 OTHER CDO TOPICS
do not suggest the return a CDO investor will receive. And regretfully, neither does excess overcollateralization delta.
Newer Monte Carlo methods have the advantage of making assump- tions about credit risk at the lowest level of detail: default probability, default correlation, default recovery; and then putting these assumptions into an objective process to obtain not only a mean result (a tranche’s expected internal rate of return (IRR), for example), but also a probabi- listic distribution of that result (such as the likelihood a tranche’s IRR will be between X% and Y%). If the estimation of underlying model parameters are haphazard at this point in the evolution of finance, the sensitivity of results to different assumptions can be tested (e.g., what happens to expected tranche yield and the distribution of tranche yield if default probability is higher than assumed).
Which method will win out? The newer Monte Carlo method will, because some investors will understand how to exploit the additional information it provides and because many other investors will just think it is sexy. But in interpreting and explaining the results of the new meth- odology, people will still fall back upon the old-fashioned, more transpar- ent, intuitive concepts: overcollateralization, weighted-average rating factor, diversity score, and, perhaps, excess overcollateralization delta.
CONCLUSION
CLO portfolios, even from CLOs issued in different years, tend to have a lot of underlying borrower names in common, especially among CLOs managed by the same manager. The degree of collateral overlap among the CLOs we studied ranges from 25% to 71% of par.
For SF CDOs, the single name of interest is the originator of the ABS, CMBS, and RMBS assets in the portfolio. Many defaults in structured finance have been originator driven, related to overly lenient underwriting standards or even to fraud by the originator. We showed that the degree of originator overlap among these portfolios is as high as 77%. However, when investment grade assets are excluded, the highest example of origi- nator overlap is only 6% and no single originator of speculative-grade collateral appeared in more than two of the SF CDOs we sampled.
We compared two approaches to assessing single name risk throughout a portfolio of CDO tranches. Our offering is excess OC delta. Software merchants are currently building or marketing simulation solutions to this and other CDO analysis problems.
The advantage of excess OC is that it is easy to calculate and relates to observable factors: the amount of excess OC a tranche has and a sin-
c23-QuantSingleName Page 450 Monday, March 6, 2006 11:28 AM
gle name’s share of that excess OC. The problem with excess OC delta is that it is an incomplete measure and does not incorporate, for example, the credit quality of the single name or the size and maturity of the tranches with exposure to the single name. While some fixes are possible for these deficiencies, excess OC delta still will not come to a conclusion in dollar-and-cent terms.
The advantage of Monte Carlo simulation is the completeness of its answer in dollar-and-cent terms. The problem with Monte Carlo simu- lation is that the model’s answer depends on inputs whose values are guesswork: default probability, default recovery, and default correla- tion. While extremely specific in terms of dollars, averages, and stan- dard deviations, it is opaque in terms of the factors leading to the result.
Of the two approaches, we think Monte Carlo methods will win out. First, because some investors will learn how to exploit the informa- tion the model can provide. Second, because the high-tech method will be perceived to be a better and more scientific way to risk manage a CDO portfolio. But we think investors will continue to fall back upon the observable, intuitive measures to check simulation results. Measures such as weighted average rating factor, diversity score, overcollateraliza- tion, and perhaps, excess OC delta.
c23-QuantSingleName Page 452 Monday, March 6, 2006 11:28 AM
CHAPTER 24
453
CDO Rating Experience
n this final chapter, we look at how CDOs have performed by rating level and types of collateral. In particular, we look at Moody’s CDO rating actions on 1,049 CDOs and 3,014 CDO tranches across 22 types of CDOs and eight years of issuance. Our unique contribution to rating transition studies is a vintage-by-vintage comparison of the frequency and severity of cumulative CDO rating downgrades across different types of CDOs. This includes an analysis of the most severe down- grades, also by both vintage and CDO type.
We are conscious that there have been criticisms of the timeliness and subjectivity of ratings actions. Even so, no other CDO performance statistic has the potential to encompass all the quantitative and qualita- tive factors that comprise a CDO’s credit quality.
We will see that most of the CDOs that experienced high downgrade frequency and severity suffered from one of two problems. For many arbi- trage CDOs, poor performance was the result of an imbalance between col- lateral spreads on the one hand and CDO debt spreads and targeted equity returns on the other. The difficulty of satisfying all CDO constituents tempted CDO managers into low-quality, high-yielding, and misrated assets. For many balance sheet CDOs, poor performance was the result of using the CDO structure to offload credit risk upon unsuspecting CDO investors. But both of these specific manifestations arose out of a more fun- damental CDO problem: poor management of the conflict of interest between CDO classes. This issue was addressed in Chapter 21.
Most of the CDOs that experienced low downgrade frequency and severity had one of two advantages. For loan and low diversity structured finance-backed CDOs, good performance resulted from timely and con- servative collateral ratings. When collateral credit quality is underesti- mated, CDO debt investors benefit from relative overprotectiveness in the
I
454 OTHER CDO TOPICS
CDO structure. For market value CDOs, good debt rating performance naturally arises from the forced liquidation of collateral assets in the event of a violation of market value coverage tests. But an exception exists when collateral liquidation is not automatic. It is our understand- ing that in all the cases of market value CDO downgrades, senior debt holders exercised their veto over collateral liquidation.