(The is the second installment of a two-part series. The first installment discussed the valuation challenges presented by the current difficult supbrime mortgage environment. This installment focuses on the other valuation complications resulting from troubles in the sector. )

Other Complications: Geography

Home price appreciation in California will probably be especially important. A substantial portion of all subprime mortgage loans originated in the past few years are secured by homes in California. Subprime mortgage ABS routinely include California loan of 35% or more, and concentrations of 45% are not uncommon.

Derivatives and Manipulations

Some market participants have raised concerns that servicers, or entities that control servicers, may attempt to manipulate the performance of subprime mortgage pools by varying the intensity of their servicing efforts or by other means. The concern relates particularly to entities with substantial positions in credit default swaps. An entity with long credit exposure would presumably try to influence credit performance in a positive way. Such an action, however, would be detrimental to the holders of short positions in the affected credits. Conversely, an entity with short credit exposure would presumably try to influence performance in a negative way. This possibility is arguably more troublesome, because it resembles the Black Sox Scandal of 1919 or the problem of a prize-fighter taking a dive. There seems to be an inherent conflict of interest when a servicer (or entity that controls a servicer) takes short credit positions. However, there does not seem to be any rule against it.

In addition, the price volatility of subordinate and mezzanine subprime mortgage ABS and of the ABX indices arguably have been amplified because the notional amount of speculative trading through derivatives greatly exceeds the aggregate volume of the underlying instruments. Also, although the trading in derivatives makes the market appear more active and liquid, it may not really be so, especially when speculators all congregate on the same side of the market. Thus, the market for triple-B subprime mortgage ABS arguably could have remained more orderly had derivatives never entered the scene.

Concentrated Triple-B Exposure in CDOs

Structured finance CDOs had a voracious appetite for the triple-B' layer of risk from subprime mortgage deals. According to some reports, they accumulated about $130 billion of exposure to that layer. Some of the exposure was in the form of actual bonds, but the great majority was through derivatives (credit default swaps). Indeed, I estimate that only around $40 billion of subprime mortgage ABS was issued at the triple-B level (including triple-B-plus and triple-B-minus) through all of 2005 and 2006. The extensive use of derivatives allowed CDOs to speculate on the triple-B tier of subprime mortgage credit risk much more than they could have using only actual bonds. That has made the CDO sector highly vulnerable to the "all or nothing" proposition embodied in many of the tranches. Even the triple-A-rated tranches of a typical "mezzanine" structured finance CDO could default if only 25% to 30% of the underlying triple-B tranches get wiped out.

Alt-A Mortgage Loans

So far, most concern about rising delinquencies has been focused on the subprime mortgage sector. However, the Alt-A mortgage sector is showing similar signs of stress. Some of the stress may come from subprime loans having been disguised as Alt-A loans and included in Alt-A mortgage securitizations. The remainder of the stress likely comes from a spillover effect from the subprime space into true Alt-A production.

Loan Modifications and Policy Initiatives

Federal policymakers are pushing servicers to make greater use of loan modifications to help families avoid losing their homes through foreclosures. So far, the pressure has been informal. However, policymakers may attempt to implement policies that make the heightened use of loan modifications compulsory. Such policies could affect both the timing and level of cumulative losses on subprime mortgage loans. In addition, such policies could affect the application of performance covenants ("trigger tests") in ABS transactions, and thereby might shift credit protection from senior tranches to subordinate tranches. In other words, subordinate tranches might benefit at the expense of senior tranches if loan modifications dampen delinquencies and spread out losses over a longer period of time. [5]

Lender-Borrower Relationship

Some policymakers have proposed fundamentally changing the relationship between mortgage lenders and mortgage borrowers. For example, a new bill introduced into the House of Representatives (H.R. 3081) would establish a fiduciary relationship between a lender and a borrower. Such a change arguably would not apply to loans that already exist, but it might have a spillover effect into the servicing area.

Wholesale Business Model

Various market participants have argued that subprime mortgage loans originated through wholesale channels - brokers and correspondents - are one of the main causes of poor credit performance. I believe that there is significant merit in this view. During good times, the weakness of wholesale-originated loans may be masked by benign conditions. However, stressful conditions can make credit quality differences appear that otherwise might not have been readily visible. Some originators, such as Wells Fargo, have already started to scale back their wholesale origination activities.

The notion that wholesale loan production is weaker than retail production is hardly new. [6] However, most market participants have chosen to downplay or ignore the issue until now. Pricing and credit models for residential mortgage loans generally do not draw significant distinctions based on whether a loan was originated through retail or wholesale channels. The file format for Standard & Poor's' LEVELS(r) model includes the "loan origination source" (retail, broker, or correspondent) as field #71, but the inclusion of the data is optional rather than required. [7] Moody's Investors Service does not include the origination channel at all in the data file format for its residential mortgage credit model. [8] It remains to be seen whether the market evolves to differentiate the pricing (and credit enhancement levels) on mortgage loans based on origination channels.

Identity of Originator - Provenance of Loans

Pricing and credit models do not generally differentiate between loans based on the identity of the originator. Nonetheless, many market participants recognize that differing business practices among originators can produce differences in loan quality that may not be apparent from standard data fields. For example, in announcing its recent downgrades of subprime mortgage ABS, Moody's noted that roughly 60% of the rating actions had been on tranches backed by loans from just four originators: Fremont Investment & Loan, Long Beach Mortgage Co., New Century Financial Corp., and WMC Mortgage Corp. [9] Some experts have argued that differences in loan credit quality can sometimes be detected by measuring the difference between the actual interest rate on a loan and the typical interest rate for loans with similar characteristics (sometimes called the "residual spread at origination" or "SATO residual"). However, even using that technique is an incomplete solution. I believe that market participants can be more effective in estimating the riskiness of loans by placing greater emphasis on how the loans were originated and who originated them. However, here too, it remains to be seen whether the market will embrace such an approach.

The Blame Game

Some market investors have tried to blame the rating agencies for the credit deterioration in subprime mortgage ABS and in CDOs. They have argued that the rating agencies applied flawed methodologies that were insufficiently transparent. I believe that these criticisms are misplaced.

Rating agency methodologies for both CDOs and subprime mortgage ABS rely heavily on quantitative models. The models, in turn, are developed from a limited body of historical data and embody numerous assumptions. For example, CDO models embody assumptions about the correlation of risk among the constituents of a CDO's underlying portfolio. Similarly, mortgage security rating models necessarily embody assumptions about the combined impact of multiple risk attributes in a single loan (i.e., risk layering), even though there may be little actual performance history on loans with those combinations of attributes. Some professionals do not seem to appreciate that any model-based analysis has inherent limitations relating to its underlying assumptions and the range of its development sample. The predictive power of any model can decline when real-world conditions violate the underlying assumptions or stray far beyond the range of conditions reflected in the model's development sample.

The rating agencies have provided the utmost transparency about their CDO and mortgage rating models. They have published reams of research reports describing the models and incremental changes to the models over time. Moreover, they make the actual models available to market participants to use. On the issuer side of the market there are many professionals who become experts in using the rating agency models in order to optimize their transactions to achieve "best execution."

An investor who believes that the rating methodologies were not transparent has probably not bothered to read the reports or to experiment with the models. The resources from which an investor can learn all about a rating methodology are readily available. As the saying goes, you can lead a horse to water but you can't make him drink. Investors who use ratings as tools in making investment decisions have a duty to learn what the ratings mean and how they work. More broadly, investors should refrain from using any type of decision tools unless they understand them and can use them properly. [10]

The argument that the rating methodologies were flawed is equally spurious. It reflects the naive view that the rating is a simple product created through a simple process. In fact, a structured rating is anything but that. It is a complex product created through a model-based process with embedded assumptions and inherent limitations. Investors who master the nuances of the rating methodologies can understand the assumptions and limitations and can determine their own tolerance for "model risk" in an informed way. Rating agency methodologies may represent only one of several reasonable approaches for estimating the risk of certain types of securities. By understanding the methodologies, investors can make informed choices about measuring and pricing risk. In that light, the recent credit deterioration of structured finance CDOs seems unsurprising because of the vulnerability of the deals to high correlation among the underlying mezzanine ABS tranches. The rating models assumed somewhat lower correlation, but investors accepted that assumption.

In the end, it boils down to this: neither ratings nor models buy bonds - but professionals do. Professionals use ratings and models as tools to help them make real-world business decisions. They want their tools to be reliable predictors. However, they need to understand the assumptions and limitations of their tools so that they have reasonable expectations about when the predictions become less reliable.


Performance uncertainty will likely rule in the short run for many subprime mortgage ABS deals backed by loans from the late 2005, 2006, and early 2007 vintages. That uncertainty may persist until after the loans reach their reset dates, when actual performance - rather than projections and estimates - becomes visible. Accurately predicting the outcome of the resets now is impossible because it will depend on the level of home prices, the level of interest rates, and the availability of credit. Each of those factors is hard enough to predict by itself, but together they represent an intractable Gordian knot. The uncertainty thwarts efforts to price subordinate and mezzanine tranches through fundamental analysis because small changes in the projected level of losses translate into large changes in indicated price. Accordingly, market sentiment and the forces of supply and demand appear to be the dominant forces driving prices.

Thus, the market has become hyper-sensitized to each new piece of "information" that might clear some of the uncertainty. That partly explains the strong market reaction to the recent rating actions and the high volatility of prices on the ABX series of indices. However, the real truth of the matter is that the subprime mortgage sector has a pulse that beats only once a month, on the 25th (or the next succeeding business day), when trustees transmit the monthly remittance reports for the deals. That is when real information appears from which market participants may have a basis to change their fundamental outlook.


Mark Adelson is an independent consultant on securitizations. He previously worked at Nomura Securities as head of structured finance research, where he reported to David Jacob. Jacob provided essential insights and comments during the preparation of this article.

[5] Loan modifications can cause further cash flow distortions that can particularly affect holders of senior tranches. For example, modifying a loan to make it have a fixed interest rate can alter the duration and convexity of floating-rate senior tranches (because of the available funds cap).

[6] See e.g. Teicher, D., et al., Theme and Variation - Understanding Why Credit Enhancement Levels Vary Among Jumbo MBS Issuers, Moody's special report (18 Sep 1998).

[7] Standard & Poor's, Residential Mortgage Input File Format for Standard & Poor's LEVELS(r) Version 5.7 (1 Mar 2007).

[8] Riggi, M. and K. Ramallo, Moody's Revised U.S. Mortgage Loan-by-Loan Data Fields (3 Apr 2007).

[9] Weill, N. and A. Tobey, Moody's Downgrades Subprime First-Lien RMBS, Moody's press release (10 Jul 2007). Subsequently, Moody's published a report stating that it will adjust its loss estimates for pools from different originators so that its "loss expectations for loan pools from originators that have performed poorly will now be as much as 20% higher than [for] pools with similar credit attributes from originators that have performed well." Shrivastava, A. and K. Bains, U.S. Subprime Mortgage Market Update: July 2007, Moody's special report (25 Jul 2007).

[10] Retail investors cannot reasonably be expected to develop the same degree of technical knowledge and expertise as investment professionals. Thus, if a retail investor buys a CDO tranche on the basis of its rating, the investor may legitimate cause to complain if the security behaves differently than he expected based on his experience with corporate and municipal bonds. In such a case, the dealer who sold the tranche to the retail investor arguably may have violated its duty to determine the tranche was a "suitable" investment (e.g. NASD Rule 2310 and NYSE Rule 405).

(c) 2007 Asset Securitization Report and SourceMedia, Inc. All Rights Reserved.

http://www.asreport.com http://www.sourcemedia.com

Subscribe Now

Access to a full range of industry content, analysis and expert commentary.

30-Day Free Trial

No credit card required. Access coverage of the securitization marketplace, including breaking news updated throughout the day.