Synthetic multi-sector CBOS.
In this article, we evaluate structured product collateral as a candidate for synthetic multi-sector CBOs.
There are myriad collateral combinations in multi-sector CBOs. However, the two principal structure types are cash and synthetic. The traditional cash structure is the most visible in the market, because cash deals are almost always publicly rated and widely advertised. The synthetic variety falls into the category of tranched portfolio default swaps or CLNs, which allow investors to leverage a well-diversified portfolio of credits without exposure to interest rate, price and collateral sourcing risks normally associated with cash CBOs. Many portfolio default transactions are unfunded, with tranched participations swapped out to counterparties off balance sheet. Further, since many tranches are unrated, most of the issuance has fallen under the radar screen, making this type of structure substantially less visible than the cash format.
The two principal motivations for issuing multi-sector CBOs are arbitrage and capital relief. Whereas the former is structured in either the cash or synthetic format, the latter principally relies upon synthetic structuring. It is the high funding cost in the cash alternative that requires managers to target a weighted average rating factor (WARF) in the BBB area. Therefore, CBO managers specialized in buying subordinated structured product tranches are ideally suited for running a traditional multi-sector CBO fund. However, with a substantially lower funding cost in the synthetic format, it is possible to achieve an adequate arbitrage with an AAA/AA+ WARF. Of course, higher WARF multi-sector CBOs can have substantial leverage (e.g., less equity, or first loss protection), which further reduces the funding cost. Lastly, many synthetic transactions are more passively managed and therefore involve significantly lower, if any, management fees, which again enhances the equity IRR.
The ability to invest the majority of the collateral in AAA and AA structured product greatly reduces sourcing risk in synthetic multi-sector CBOs. With over $5 trillion of senior structured product notes outstanding (see appendix for supply figures) and financial institutions seeking regulatory and economic capital relief for this asset class, we foresee robust issuance of synthetic multi-sector CBOs in the years to come. To the extent the recent terrorist attacks dampen CDO issuance, we believe any decline will be minimal for a few reasons. First, among all collateral going into CBOs, structured product is the most stable and free of event risk. Second, synthetic multi-sector CBOs focus their buying on the highest rated tranches of structured product. Lastly, we believe the need for insurance companies to obtain alternative sources of funding off balance sheet in the wake of large P&C losses will result in additional securitization of structured product via multi-sector CBOs.
II. MULTISECTOR CBO ISSUANCE TRENDS
Between January 2000 and August 2001, $25 billion of cash multi-sector CBOs had been issued in 60 transactions. The first publicly rated multi-sector cash flow deal--the $300 million DASH--was issued in late 1999 when the cash arbitrage spread for subordinated structured product was near today's level of approximately 140 bps. Despite a tightening arbitrage spread in February 2000 as Y2K fears started subsiding, issuance took off and has held relatively steady ever since. Exhibit 1 shows monthly issuance volume in dollars (bars) and by number of deals (lines).
Over this period, an average of three deals have been issued per month, totaling about $1.25 billion per month. In fact, the only month to have no issuance was July 2000. Year to date through August 2001, $11.5 billion of cash multi-sector CBOs have been issued, an annual run rate of $17.2 billion--29% above last year's $13.4 billion of issuance. We expect multi-sector CBO issuance to continue to account for at least 30-40% of total CDO issuance, primarily due to the high defaults being experienced in both high grade and high yield corporate collateral, the competing asset class. The major threat to future cash multi-sector CBO issuance is sourcing risk due to the dependence on subordinated structured product to earn a high arbitrage spread. However, managers have recently somewhat circumvented this supply constraint by diversifying into corporate collateral. Although increasing default risk, adding more corporate debt to the mix also increases the diversity scores in these deals.
Although the limitations of deal disclosure force us to estimate, we believe synthetic multi-sector CBO issuance to have already exceeded cash issuance of $25 billion on a notional basis. On a rated note basis, we estimate there is over $7 billion in synthetics outstanding. Because many of the synthetic deals are privately rated and unfunded, or only partially funded, little information is available. Further, to a large extent the Street prefers to maintain confidentiality for these deals because the structures tend to be proprietary. Nevertheless, Exhibit 2 details 14 publicly rated synthetic multi-sector deals totaling almost $4 billion in notes that have closed since May 2000. The fact that they were publicly rated meant that they were intended for wide distribution, necessitating dissemination of the deals' structure and pricing.
Although the rating agencies got comfortable with rating cash multi-sector deals last year, they are still climbing the learning curve for rating synthetic structures. Consequently, among the publicly rated deals, it has been typical that only one agency rated a particular deal. Of the 14 deals in Exhibit 2, 11 were rated by Fitch only and two by Moody's only, while the remaining deal was rated by all three rating agencies. For the eight deals disclosing the full capital structure, on average, 84% (with a range between 75% and 92%) is rated AAA and/or super senior. Although most of these publicly rated deals are denominated in USD, many others are in Euros. With little data on the overall synthetic multi-sector CBO market publicly available, it hardly makes sense to draw conclusions on structures for past deals. That said, in our Structure and Arbitrage section (see page 7), we present a template for a generic structure going forward.
III. ISSUANCE MOTIVATIONS
The primary motivation for issuing traditional cash multi-sector CBOs is arbitrage. Hedge funds and other asset managers specializing in structured product have relied greatly on CBOs to increase assets under management and create a stable source of fee income. On the synthetic side, the primary motivation has historically been regulatory capital arbitrage, internal economic capital relief, and off-balance sheet funding. However, arbitrage is increasingly driving synthetic issuance as well.
As regulatory capital rules still require 8% equity (100% risk weight x 8% minimum capital requirement) applied against all structured product except for some residential mortgage classes, financial institutions are strongly motivated to free up regulatory capital by securitizing these assets via balance sheet CBOs. Exhibit 3 shows risk weights for all structured product sectors. Only home equity ABS and RMBS pass-throughs have risk weights under 100%. Further, since the sponsor typically retains only the first loss exposure, securitization allows for a reduction in the amount of economic capital allotted to the securitized portfolio. Lastly, banks and insurers use multi-sector CBOs as a means of off-balance sheet financing. Particularly in times of distress, financial institutions can obtain funding synthetically without alarming the markets publicly and raising their cost of funds.
Financial institutions hold structured product on balance sheet in one of four ways:
* As an unsold portion of an underwriting.
* As an unsold portion of a sponsored transaction.
* As a securities portfolio investment.
* As a loan portfolio investment (whole loan).
Financial institutions have typically used synthetic securitization to securitize the highly rated collateral on their balance sheets. At least until 2008, regulatory capital charges for structured product will remain onerous. Further, most market participants believe that even the new capital rules will be out of sync with true economic capital requirements and remain overly conservative. By retaining only a sliver of equity to support the low probability of default in these highly rated structured assets, the bank is able to reduce capital requirements significantly. Generally, if certain conditions are met, a bank must hold dollar-for-dollar capital against the first loss position and an additional 1.6% against the senior loss position, which is calculated as the reference portfolio notional minus the notes, minus the retained first loss position.
Although regulatory capital has been the primary motivation behind synthetic issuance, increasingly arbitrage and funding have become the motivating factors. Whereas the collateral for most balance sheet CBOs is fairly passively managed with limited substitution rights, arbitrage transactions involve more active management. Although the arbitrage spread is a secondary consideration for balance sheet deals, it is a for more important driver for doing arbitrage deals. In arbitrage deals, relatively more equity must be sold requiring more excess spread in the deal to support the equity IRR. In the next section, we will show how the arbitrage spread is determined in a synthetic multi-sector CBO.
IV. STRUCTURE AND ARBITRAGE
Synthetic multi-sector CBOs can be either funded, unfunded or a combination of the two. In a funded deal, the issuer or sponsor buys protection on each credit via a portfolio default swap and pays a premium to the trust, which issues rated or unrated CLNs with various degrees of exposure to losses incurred by credit events in the portfolio. The first loss, or a portion thereof, is typically retained by the issuer or sponsor. Then, the second and third loss exposures may be taken by respective BBB and AAA rated CLNs, for example. The highest part of the capital structure is referred to as the super senior tranche and typically represents greater than AAA protection. In a partially funded deal, some of these exposures--usually the super senior tranche--may be swapped out to a third-party entity through a credit default swap. Finally, in an unfunded transaction, all the exposures are swapped out, requiring no CLNs.
Whether the transaction is a fully rated balance sheet deal or a privately rated, unfunded tranched portfolio default swap, or something in between these two extremes, synthetic structures require less equity and note placement than traditional cash structures. Consequently, deal execution is more efficient. Often much, if not all, of the first-loss equity piece is retained by the sponsor, avoiding the time-consuming and expensive equity placement process.
In a synthetic transaction, the dealer or sponsor may simply retain the senior loss position, as it represents a de minimis risk. If the super senior loss position is swapped out on an unfunded basis, the payment is usually in a range of 6-20 bps, depending on the structure and collateral. Therefore, the arbitrage can be significantly more attractive than that seen in cash CBOs, whose AAA tranches require at least L+45 bps due to the complexity and liquidity premium demanded by traditional cash investors.
While cash multi-sector CBOs usually require a maximum WARF of BBB+ to achieve an attractive arbitrage, synthetic structures allow for a WARF as high as AAA/AA+, although some deals may have collateral WARFs as low as those seen in the traditional cash structure. A driver behind the higher WARFs in synthetic deals is regulatory capital arbitrage, which is greatest in AAA structured product held on balance sheet. That said, the synthetic format creates an arbitrage in AAA collateral.
In Exhibit 4, we present a generic synthetic structure with seven equally weighted asset sectors having a WARF of Aaa/Aa1- funded by a mix of Baa1 (3%), Aa2 (5%) and super senior liabilities (92%). The structured product sectors in the asset mix include Aa and Aaa rated CMBS, RMBS, home equity (HEQ) ABS and CDOs. The liability mix is that of the North Street 2001-3 synthetic multi-sector CBO issued in early 2001. This deal was among the first to not require an equity tranche per se, other than the Baa1 first loss tranche.
The arbitrage spread for this generic structure is 55 bps (82 bps asset yield minus 27 bps funding cost based on actual North Street pricing for rated tranches). Whereas we estimate the typical cash multi-sector CBO with a WARF of BBB has had an arbitrage spread in the 100-150 bps range over the last two years, the synthetic structure having a WARF of AAA/AA+ may require an arbitrage spread of only 50-60 bps. It is principally the relatively high leverage in the synthetic structure and the ability to bypass traditional cash investors for the higher rated funding that allows for such a low arbitrage spread. The major advantage of the higher WARFs in synthetic deals is that collateral sourcing risk is negligible because outstanding AAA structured product paper is plentiful. We estimate that AAA paper accounts for 85-90% of all structured product. On the other hand, there is a limited supply of subordinated paper, which traditional cash deals require to achieve a suitable arbitrage (for this reason, the cash CBO bid has driven subordinated paper to artificially tight levels in the past few years).
For investors, the advantages of the synthetic structure are equally compelling. Investors have exposure only to credit events as defined by specific language written into the credit default swap or CLN agreement. Unlike cash CBOs, which have strict collateral performance tests affecting the waterfall (e.g., priority of payments to rated tranches), synthetic investors are not affected by collateral trading losses and spread widening unrelated to a credit event. Further, investors do not bear other risks associated with cash deals, such as ramp-up risk, interest rate risk, reinvestment risk, currency risk, hedging risk or prepayment risk on the assets in the reference pool.
V. COLLATERAL PERFORMANCE
A. Spread Performance
Spreads on structured product have maintained relatively high stability through all cycles. The reasons spreads have been stable are several: credit enhancement features (particularly subordination levels for all but the most junior tranche), low event risk, very few defaults on all outstanding deals to date, relatively few downgrades and, as of yet, an untested environment for a downturn in credit for the collateral supporting the deals. However, regarding the last point, in the wake of the terrorist attacks on September 11, it is possible to imagine a spread contagion period similar to that experienced in the Fall of 1998 when spreads widened despite steady collateral performance. Rather than being a deterrent to future issuance, though, we think such spread widening would provide a window of opportunity for CBO managers to lock in wide spreads before they return to more normal levels once the crisis dissipates.
In Exhibit 5, we show how all investment grade tranches for CMBS have performed between December 1997 and August 2001. We chose CMBS as our sample asset class because this sector has the best price discovery over a full range of ratings due to the large number of outstandings for tranches rated below AAA. Through August 1998, 10-year AAA CMBS traded at an average spread of 29 bps, which is still below the 48 bp average for 2001 to date. During the LTCM crisis, AAAs widened to 106 bps on October 16 before returning to the 40- to 50-bp area. Since the 1998 spread contagion period, AAA CMBS have held in this tight range except for the month of October 2001, when they traded briefly in a range of 50-55 bps.
For tranches rated below AAA, we see a similar, although exaggerated, trading pattern. It is important to note that the spread widening during the Fall 1998 contagion period only lasted a few months. The CMBS spread widening was an indicator of temporary spread contagion and a permanent rise in the liquidity premium rather than the market's perception of potential losses. So if we consider how the average spread levels were changed by the Fall 1998 contagion period once volatility subsided in January 2000, we have some idea of how investor sentiment regarding liquidity changed permanently. In this regard, spreads were, on average, 33-44% wider from January 1999-August 2001 than they were in the first seven months of 1998, before the crisis.
There have been only two spread widening periods in CMBS since the Fall of 1998. The pre-2000 Y2K-related spread widening was only one-third of the size of the 1998 market contagion period. Subsequently, in early 2001, CMBS spreads widened temporarily due to a sector rotation out of CMBS into corporates, whose spreads had risen substantially by the end of 2000 before tightening again in January 2001.
Wider spreads in contagion periods provide an opportune time for multi-sector CBOs to ramp up. If structured product spreads widen meaningfully in response to the September 11 terrorist attacks as a result of spread contagion, we think it will be an opportunity for more CBO issuance to come to market. The wider arbitrage spreads should provide a more than adequate cushion needed to withstand possible defaults emanating from the events of September 11.
In Exhibit 6, we show new issue spreads for most structured product sectors in early September 2001, right before the terrorist attacks. We acknowledge spreads have widened since that period for the most part. But because levels are changing so much day to day, we believe the early September spreads better represent the levels where multi-sector CBOs have been headed over the last few months and where they will end up once certainty returns to the market. For each sector, we show the spread to LIBOR and spread to same-rated corporates as measured by our HG Broad Market Index and HY Large Cap Index. Much of the pickup in structured product derives from a liquidity and complexity premium. However, since synthetic CBOs are largely buy and hold investments with little trading, liquidity is not a significant risk. Further, complexity is not an issue because the CBO sponsor or manager presumably has expertise in structured product.
In ABS, we see the largest spreads to LIBOR and same-rated corporates in HEQ paper. AAA and BBB HEQ paper are 103 bps and 55 bps, respectively, wider than same-rated corporates. This significant pickup is mostly attributable to the negative convexity (discussed in a later section) in HEQ paper. Whereas non-mortgage ABS--which has relatively high average life stability--has traded wide to same-rated corporates in years past, the CBO bid and general corporate spread widening has resulted in convergence between spreads of subordinated non-mortgage ABS and same-rated corporates.
In CMBS, the major pickup to corporates is in AAA and AA tranches--the focus for synthetic multi-sector CBOs. Subordinated CMBS paper trades slightly through same-rated corporates. Interest only (IO) CMBS tranches, which are AAA rated, trade extremely wide to same-rated corporates due to the credit-related prepayment sensitivity in these structures. IO CMBS issues are very sensitive to prepayments induced by credit losses in the underlying collateral. Nevertheless, small allocations to IO CMBS are often used to boost the arbitrage spread in multi-sector CBOs.
In RMBS, there is a substantial arbitrage to same-rated corporates across subsectors and ratings due to additional spread to account for negative convexity. Prime home mortgage borrowers are more likely to refinance their mortgages with minimal rate declines. In AAA subsectors, the arbitrage to corporates ranges from 81-133 bps. In subordinated private CMOs, there are pickups to same-rated corporates of 56-162 bps. However, unfortunately for CBO managers, there is relatively little outstanding supply of subordinated private CMOs.
In new issue high yield bond CDOs, we also see substantial gains to same-rated corporates: AAAs have a pickup of 53 bps, while BBs have a pickup of a full 312 bps. Secondary CDOs on credit watch trade at significantly wider spreads than these levels, and the pickup is even more attractive to same-rated corporates on credit watch. The idea of CBOs buying CDOs as collateral seems complex. However, in our opinion, if the manager is an experienced CDO buyer and picks healthier vintages (not 1997, 1998 or 1999) with the best managers, the strategy makes sense.
In a traditional cash multi-sector CBO, spread gapping in the absence of defaults can have significant negative ramifications, particularly for subordinated tranches. As spreads in the underlying collateral gap, the first impact is normally felt on the WARF tests and the rating bucket tests (e.g., the <Bal bucket must be greater than x). Then, as it becomes clearer these tests may result in a downgrade to the rated tranches, the manager may begin trading out of losses. It is at this point that O/C and I/C tests can be breached. The stringent cash CDO tests direct the waterfall--i.e., allow principal to be paid to senior tranches at the exclusion of junior tranches--and impact the rating migration of the rated notes. By contrast, synthetic multi-sector CBOs normally have far less stringent waterfall rules and expose the investor only to defaults. Therefore, synthetic deals are not as vulnerable as cash deals to temporary spread gapping as occurred during the 1998 spread contagion period.
B. Default and Recovery Performance
The merits of structured product (e.g., ABS, CMBS, RMBS and CBOs, which, combined, represent most of the collateral in multi-sector CBOs) from a historical default perspective are compelling. According to Fitch data, average annual defaults since 1989 for all structured product are less than 0.01% and have a cumulative default rate of 0.05% by original principal balance. When viewed in contrast to same-rated corporate securities, whose default rate averages 0.23%, the advantages for structured product are clear.
Fitch's January 8, 2001 report, Structured Finance Default Study, covers default performance through June 30, 2000. The data thus excludes the recent high-profile ABS defaults such as Heilig-Meyers and Hollywood Funding, which, although considered isolated events by the rating agencies, both went from AAA to near-default ratings in one day. According to Fitch research, there is no significant difference between the three agencies' default experience. Further, with an overall annual default Exhibit below 0.01%, any differences are statistically insignificant. In addition, an S&P study through June 2001, which is detailed below, reports that the total number of defaulted structured product bonds is 116 in the period 1978-2001, whereas Fitch counts 89 defaults in the period 1989-2000. Nonetheless, it is dangerous to compare rating agency default statistics, as they define default differently. For instance, for CMBS and RMBS, whereas Fitch defines default as a loss of principal, S&P and Moody's also count impairment of interest as default. Fitch's cumulative default rate of 0.05% (by original principal balance) for all structured product means that 99.95% of all rated structured finance securities have not defaulted. According to Fitch, "This high rate of success supports the concept that isolating a pool of assets from an originator or seller significantly reduces default risk." Despite the less-than-favorable economic environment we are entering, there are several factors that protect structured product from defaults:
* Securitization of a diversified pool of assets avoids event risk associated with corporate bonds. By isolating assets in a bankruptcy-remote SPV, default risk is strictly a function of the underlying assets.
* Structured product withstands multiple levels of due diligence and outside review by underwriters, auditors, rating agencies, counsel and investors before issuance.
* In many instances, the originator has stepped in to support a transaction to maintain access to the securitization market. Correction techniques have included trapping residual interest to build up reserves, purchasing underperforming collateral from the trust, subordinating servicing fees and simply adding more credit enhancement.
According to its report, "Since these factors are naturally inherent in structured finance securities, Fitch believes that default rates for structured finance securities will remain below those of similarly rated corporate bonds." Further, we note that in sequential pay structures, when collateral quality triggers are met, principal is accelerated to senior tranches at the exclusion of subordinated tranches until the collateral test requirements are corrected.
Of the 89 defaulted bonds, RMBS accounted for 52 from 34 deals; CMBS, only two from one deal; and ABS, 38 from 21 deals. Although it is difficult to draw concrete conclusions from such a dearth of default experience, a few generalities are clear. According to Fitch, with no exceptions, poor collateral quality was the reason for RMBS defaults. However, the 0.04% cumulative default rate for RMBS is still extremely low, principally due to the strong economy and stable housing market during the study period. The vast majority of defaulted RMBS started out as noninvestment grade and were originated in 1993-95, when competition between originators was fiercest.
The two classes of CMBS that defaulted were rated B- and BB- by Fitch originally in 1993. In a separate study up to December 1999, Fitch found that among all rated US CMBS, only 12 classes within six deals have defaulted, resulting in a cumulative default rate of 0.21% by original principal balance.
As opposed to RMBS and CMBS, the ABS market has traditionally structured a large majority of classes to the AAA and A rating category, with only 2-3% of new issuance in the noninvestment grade categories. Thus, defaults are more heavily weighted to classes rated in the 'A' and 'BBB' categories. It is clear from Fitch's statistics that when one originator gets into trouble, several of their deals are impacted: of the 35 defaulted classes, 22 came from three originators: Aegis, AMN, and NAL. On a sector basis, 71% of the 35 defaulted classes were in subprime auto deals originated in 1995 and 1996. The root causes for the subprime auto losses were looseunderwriting standards and poor servicing amid a highly competitive and saturated marketplace. The other three sectors suffering defaults were HEQ, rental car fleet and franchise ABS. Fitch expects the sectors most exposed to defaults in the coming years to be franchise, equipment, HEQ and manufactured housing.
The low occurrence of structured product defaults is particularly impressive when we compare them to same-rated corporate bonds. An average annual default rate of 0.01% for structured product is substantially lower than both IG corporate bonds (0.08%) and HY corporate bonds (3.07%). Even if we weight the ratio of HG to HY corporate bonds to correspond with the ratio for structured product, the comparison remains remarkable (Exhibit 8). Taking into consideration the actual 77/23% split between HG and HY corporate bonds outstanding, the average annual default rate for all corporate issuance is 0.77%. However, when we weight the corporate bond default statistics to correspond with Fitch's estimate of a 95/5% split between HG and HY rated bonds in structured product, the adjusted average annual corporate bond default rate becomes 0.23%.
S&P's September 4, 2001 report, Life After Death: Recoveries of Defaulted US Structured Finance Securities, indicates in the last 23 years up to June 30, 2001 there have only been 116 defaults out of 13,538 rated classes. Accordingly, S&P's cumulative and average annual default rates for structured product are 0.86% and .04%, respectively, on a per bond basis. Of the 116 defaulted bonds, RMBS accounted for 83 of those defaults; CMBS, 14; and ABS, 19. According to the S&P report, "Not only do [structured securities] rarely experience default, but they also recover a major portion of their original principal even after default." Defaulted structured product has generally managed to recover more than 50% of principal. Of the 116 defaults, recoveries for RMBS (in contrast to Fitch's definition, this includes HEQ ABS and other subprime mortgages), CMBS and ABS averaged 61%, 66% and 29%, respectively. Unlike RMBS, there was no clear relationship between recovery and original credit ratings among CMBS classes. Charged-off credit cards and franchise loans accounted for 17 of the 19 total ABS defaults.
C. Downgrade Performance
From a rating migration standpoint, structured product is superior to same-rated corporates. According to Moody's data, ABS and CDOs have a 7.8% average one-year downgrade rate for all ratings between B3 and Aaa since 1986, whereas the comparable downgrade statistic for corporate bonds is 13.6%--a difference of 5.8 percentage points. Over the same period, according to S&P data, the average rate of annual downgrades for CMBS and RMBS rated between B and AAA is 1.8% and 3.2%, respectively, versus 7.5% for corporates.
Exhibit 9 summarizes Moody's one-year average rating migration data for ABS (mortgage and non-mortgage) and CDOs between 1986 and 2000. On average, ABS ratings overall have been quite stable, with over 90% of ratings staying the same over one-year periods for most rating categories. Conversely, less than 80% of corporate bond ratings have stayed the same over a year. Further, in most cases, any rating changes in ABS were due to changes to entities related to the transactions (servicer or corporate guarantor) rather than collateral deterioration.
From a downgrade standpoint, the best performance in ABS and CDOs relative to corporates is in the A3 and above rating categories. On average, one-year average downgrades for ABS and CDOs is 10.1 percentage points less than that for corporates. Interestingly, Baa2 rated ABS and CDO paper has performed substantially better than Baal or Baa3. Noninvestment grade paper is difficult to compare, as the number dABS and CDO issues for these ratings is limited.
Whereas Moody's has the most recent comprehensive data on ABS and CDO rating migration, we have to look to S&P for migration data on CMBS and RMBS versus corporates. In contrast to Moody's data, which evaluates each rating notch, the S&P data evaluates groups of three notches. For instance, whereas Moody's provides data for A1, A2 and A3 collateral, S&P groups all the data into an A bucket.
Exhibit 10 summarizes S&P's average one-year rating migration data for CMBS and corporates between 1985 and 2000. CMBS ratings are extremely stable, with slightly more upgrades than downgrades on average. Reflecting the high subordination in each rated tranche, only 1.8% of CMBS tranches are downgraded each year on average versus 7.5% for corporate bonds. The CMBS tranches most sensitive to downgrade are the BBB and B tranches.
The rating migration statistics for RMBS (defined by S&P to include all mortgage ABS as well as prime MBS) are also remarkable. Exhibit 11 summarizes S&P's average one-year rating migration data for RMBS and corporates between 1980 and 2000. As similar to the performance of CMBS, there have been slightly more upgrades than downgrades of RMBS on average. The BBB and noninvestment grade tranches of RMBS have been the most susceptible to downgrades. Nevertheless, whereas average one-year rating downgrades for corporates are 7.8%, they are only 3.2% for RMBS.
D. Average Life Variability
Unlike traditional cash deals, in the typical synthetic multi-sector CBO, the investor is not exposed to market risks, including prepayment risk. Instead, the manager or sponsor bears this risk. All structured product has some average life uncertainty--the question is the degree of uncertainty. Generally, prime RMBS has the most average life uncertainty while CMBS has the least. After RMBS is subprime mortgage ABS. Next comes non-mortgage ABS, which varies by sector.
In general, for sequential structures, senior classes are more sensitive to collateral prepayment than junior classes because prepayments are allocated first to senior classes. Since synthetic multi-sector CDOs tend to reference predominately senior tranches, average life variability is more of a concern for synthetic CDO managers vis-a-vis cash CDO managers. Therefore, it is important to have ample diversification across asset classes to mitigate average life variability.
In prime RMBS, borrowers have a free option to prepay their mortgages when rates fall. In today's environment of widely disseminated data, even the most unsophisticated mortgage borrower will likely prepay if mortgage rates fall 75 bps or more, which more than covers the cost to refinance. In a falling rate environment when RMBS mature faster than expected, the manager must reinvest in a lower rate environment. Conversely, in a rising rate environment, the manager faces extension risk. Fewer RMBS than expected mature when the manager would like to reinvest in the higher prevailing rates of the market. This so-called "negative convexity," or option cost, is no two models produce the same estimate of the prepayment option's cost.
In subprime RMBS (mortgage ABS such as HEQ, manufactured housing and high loan-to-value (HLTV) ABS), borrowers are typically less astute and have fewer options to prepay. In fact, most HEQ loans have stiff" prepayment penalties. Further, there are fewer lenders to subprime borrowers than to prime borrowers, particularly in recent years after many lenders have exited the business. As in prime RMBS, this prepayment option is priced into the bond and can be quantified by a model. However, this option cost is not easily found from a commercially available source such as Bloomberg. For the most part, Wall Street firms have the best models for quantifying the option cost. Therefore, the manager must be particularly specialized in these sectors to determine value.
For non-mortgage ABS (e.g., credit cards, autos and equipment), there is substantially less average life uncertainty. In fact, many credit card ABS are issued as bullet or soft bullet structures. With trustee reports coming up shy on the necessary data, it is difficult to quantify even historical--much less future--prepayment patterns for auto and equipment ABS. Due to this lack of information and the negligible average life uncertainty in non-mortgage ABS, there are no standard prepayment models for these sectors.
Apart from the IO strips, all tranches of CMBS are well protected from prepayments due to substantial call protection in the form of lockouts, defeasance, yield maintenance and prepayment penalty points. The only non-IO tranche to have ever suffered spread widening due to rapidly falling prepayment rates has been the AAA rated five-year tranche because it is the first tranche in the waterfall. However, this tranche will only prepay if there are credit losses in the underlying collateral, which would start prepaying this first tranche in the waterfall at the exclusion of the most junior tranches. For this reason, spreads on AAA rated five-year CMBS tranches have widened out to those of the AAA rated 10-year tranches in recent months. As discussed previously, the only tranche to be significantly impacted by falling interest rates is the CMBS IO, whose principal on which the interest payments are based quickly disappears on credit losses in the underlying collateral. In this regard, CMBS IOs are as sensitive to rising credit losses as RMBS IOs and mortgage servicing portfolios are to falling interest rates.
APPENDIX: STRUCTURED PRODUCT VOLUME STATISTICS
This last section provides a wealth of figures addressing the supply of structured product in the market. Based on Bond Market Association data as of June 30, 2001, there is over $5 trillion in supply to source for synthetic multi-sector CBOs in US collateral alone (Exhibit 12). This is an astounding amount of eligible collateral, considering structured product represents 37% of the total US taxable bond market with the remainder being in treasuries, agencies and corporates.
The remaining exhibits in the appendix provide detail by sector, domicile and time for structured product issuance including ABS, CMBS, cash CDOs and RMBS.
LANG GIBSON is responsible for the research and strategy group for Structured Credit Products (SCP), whose product areas include credit derivatives, synthetic CDOs and cash CDOs. He writes and publishes the only extensive credit derivative/CDO weekly in the industry as well as numerous topical and primer reports covering SCP's three product areas. Prior to Bane of America Securities, Lang was the structured product strategist at First Union Securities. Prior to joining First Union in January, 1999, Lang brought eight additional years of structured product research and risk management advisory experience from Goldman Sachs, J.P. Morgan and Ferrell Capital Management. In addition to firm research, Lang has published numerous articles and chapters in well-known trade journals and Frank Fabozzi publications. Lang holds an M.B.A. in Finance from the NYU Stern School of Business and a B.A. from the University of Virginia.
Exhibit 2. Publicly Rated Synthetic Multi-sector CBOs Date Offered Deal name Manager Collateral 5/8/00 Equinox Funding Rabobank Diversified pool of ABS 5/30/00 North Street Ref. UBS Brinson ABS Linked Notes 8/31/00 PARIS CDO Natexis Banque 60% CC, 40% CBO Populaires 10/2/00 Stuyvesant CDO I, Ltd. Rabobank 100% CDOs 10/16/00 Natexis Natexis Banques ABS ABS-Backed Populaires 10/27/00 North Street Ref. UBS 60% Structured Linked Notes 2000-2 40% Corp. bond 12/7/00 CORVUS Barclay's COS, ABS 12/22/00 Hector Funding Ltd. Barclays CDOs, ABS, Series I CMBS, RMBS 1/9/01 Structured Finance Barclays 25% CDOs, ABS Asset Ltd - I CMBS, RMBS 3/12/01 Hector Funding II Ltd. Barclays CDOs, ABS, Series I CMBS, RMBS 4/17/01 NorthStreet CDO-III UBS Principal AAA/AA ABS Finance 5/16/01 Savannah - II Barclays 60% Struc, 20% CDO Ltd. LM, 20% other 6/30/00 Nerva Ltd. Barclays CDOs, ABS, CMBS, RMBS 8/6/01 Taunton CDO Ltd. Barclays CDOs, ABS, CMBS, RMBS Date Offered Deal name Class Size 5/8/00 Equinox Funding A $37.50 B 22.50 C 16.88 D 30.00 Equity 15.00 121.88 5/30/00 North Street Ref. A 36.00 Linked Notes B 40.00 C 31.00 D-1 14.00 D-2 20.00 Equity 43.00 184.00 Lev. Amt 1,344.00 8/31/00 PARIS CDO A 30.00 B 40.00 70.00 10/2/00 Stuyvesant CDO I, Ltd. 25.00 10/16/00 Natexis Snr. 336.90 ABS-Backed A 25.00 B 40.00 Equity 10.00 411.90 10/27/00 North Street Ref. A 60.80 Linked Notes 2000-2 B 32.60 C 29.00 D 7.50 E 36.10 F 43.00 209.00 12/7/00 CORVUS A-1 550.00 A-2 200.00 B 65.00 C 60.00 D 40.00 E 25.00 F 10.00 950.00 12/22/00 Hector Funding Ltd. A 50.00 Series I 1/9/01 Structured Finance A JPY 15000 Asset Ltd - I B 430 C 260 D 690 E 350 Equity 520 JPY 17250 3/12/01 Hector Funding II Ltd. A Euro 50.00 Series I 4/17/01 NorthStreet CDO-III Credit Swap 1,840.00 A 100.00 B 60.00 2,000.00 5/16/01 Savannah - II A 300.00 CDO Ltd. B 18.45 C 22.50 D 6.75 E 5.25 Equity 22.05 375.00 6/30/00 Nerva Ltd. A 528.00 B 12.00 C 30.00 D 12.00 E 18.00 600.00 8/6/01 Taunton CDO Ltd. A-1 220.00 A-2 80.00 B 26.00 C 24.00 D 16.00 E 10.00 F 4.00 G 20.00 400.00 Date Offered Deal name Senior Moody's S&P 5/8/00 Equinox Funding Aaa Aa2 A2 Ba2 NR 5/30/00 North Street Ref. 86% Linked Notes 8/31/00 PARIS CDO 10/2/00 Stuyvesant CDO I, Ltd. 10/16/00 Natexis 82% Aaa AAA ABS-Backed NR NR NR NR NR NR 10/27/00 North Street Ref. Linked Notes 2000-2 12/7/00 CORVUS 79% 12/22/00 Hector Funding Ltd. Series I 1/9/01 Structured Finance 87% Asset Ltd - I ($144.10) 3/12/01 Hector Funding II Ltd. Series I 4/17/01 NorthStreet CDO-III 92% Aa2 Baa1 5/16/01 Savannah - II 80% CDO Ltd. 6/30/00 Nerva Ltd. 88% 8/6/01 Taunton CDO Ltd. 75% Date Offered Deal name Fitch Pricing 5/8/00 Equinox Funding 6mL + 47 6mL + 80 6mL + 130 6mL + 690 5/30/00 North Street Ref. AAA 3mL + 70 Linked Notes AA 3mL + 105 A 3mL + 175 BBB 3mL + 260 BBB BB- 8/31/00 PARIS CDO A+ BBB 10/2/00 Stuyvesant CDO I, Ltd. AA 10/16/00 Natexis AAA ABS-Backed A+ 3mL + 125 BBB 3mL + 240 NR 10/27/00 North Street Ref. AAA 3mL + 70 Linked Notes 2000-2 AA 3mL + 105 A 3mL + 175 A- 3mL + 260 BBB BB- 12/7/00 CORVUS AAA 6mL + 65 AAA AA 6mL + 90 A 6mL + 130 BBB 6mL + 250 BB 6mL + 625 B 12/22/00 Hector Funding Ltd. AA 3mL + 90 Series I 1/9/01 Structured Finance AAA 6mL + 35 Asset Ltd - I AA 6mL + 55 A 6mL + 90 BBB 6mL + 155 BB 6mL + 355 NR Residual 3/12/01 Hector Funding II Ltd. AA 6mEL + 135 Series I 4/17/01 NorthStreet CDO-III 3mL + 100 3mL + 375 5/16/01 Savannah - II AAA CDO Ltd. AA BBB BB B NR 6/30/00 Nerva Ltd. AAA 3mL + 75 AA 3mL + 95 BBB 3mL + 275 BB 3mL + 475 NR Residual 8/6/01 Taunton CDO Ltd. AAA 6mL + 75 AAA 8.15% AA 6mL + 110 A 6mL + 150 BBB 6mL + 300 BB 6mL + 750 B 16% NR Residual Exhibit 3. Risk Weightings for US Banks ABS: Private HEQ 50% Other ABS 100% CMBS 100% RMBS: GNMA Passthroughs 0% FNMA/FHLMC Passthroughs 20% Private Passthroughs 50% CMOs 100% CDOs 100% Source: Federal Reserve, BIS. Exhibit 4. Arbitrage Spread--Synthetic Multi-Sector CBO Sept. 3, 2001 Libor Spread Weight Assets: AAA-rated 10-Yr HEQ 95 14% AAA 10-Year CMBS 44 14% 30-year Passthrus 82 14% 10-Yr Agcy CMOs 103 14% AA-rated CMBS 58 14% AA-rated Private CMOs 120 14% AA-rated CDOs 74 14% Weighted Avg. Yield 82 Liabilities: Super Senior 12 92% Aa2 100 5% Baa1 375 3% Weighted Avg. Cost 27 Arbitrage Spread 55 Source: Banc of America Securities LLC. Exhibit 6. Structure Product Spreads to LIBOR and Corporates Pick-up to Sept. 3, 2001 same-rated Sector Subsector Libor Spd Corporates ABS AAA-rated 10-Yr HEQ 95 103 AAA-rated 10-Yr Cards 24 32 A-rated 3-Yr Prime Autos 34 (50) A-rated 3-Yr Equipment 70 (14) BBB-rated Top-Tier Cards 105 (45) BBB-rated Equipment 135 (15) BBB-rated HEQ 205 55 CMBS AAA 44 52 AA 58 11 A 74 (10) BBB 133 (18) BBB- 178 (23) AAA-rated 10 400 408 RMBS AAA-rated: 15-year 73 81 30-year 82 90 ARMS 74 82 3x1 Hybrids 109 117 10-Yr Agcy CMOs 103 111 10-Yr Priv CMOs 125 133 Subordinate Privates CMOs: AA 120 73 A 140 56 BBB 210 60 BB 515 162 HY Bond CDOs AAA 45 53 AA 74 27 A 130 46 BBB 218 68 BB 665 312 Corporates AAA (8) AA 47 A 84 BBB 150 BB 353 B 817 Source: Banc of Securities LLC. Exhibit 7. Structured Product Annual Default Rates, 1989-2000 (a) ABS RMBS CMBS Defaults ($MM) 668 116 69 Total Issuance ($MM) 892,352 497,874 172,338 Cumulative Default Rate (%) 0.07 0.02 0.04 Annual Default Rate (%) 0.01 < 0.01 < 0.01 All Structured Product Defaults ($MM) 853 Total Issuance ($MM) 1,562,565 Cumulative Default Rate (%) 0.05 Annual Default Rate (%) 0.01 (a) Includes public, 144A and privates; includes international as well as US Source: Fitch and Banc of America Securities LLC. Exhibit 8. Structured Product vs. Corporate Bond Defaults, 1989-2000 Average Annual IG/HY Defaults (%) Mix Structured Products 0.01 95%/5% IG Corporate Bonds 0.08 77% HY Corporate Bonds 3.07 23% All Corporate Bonds 0.77 77%/23% Adjusted Corporate Weights 0.23 95%/5% (a) Includes public, 144A and privates; includes international as well as US Source: Fitch and Banc of America Securities LLC. Exhibit 9. Average One-Year Rating Upgrades/Downgrades, 1986-2000 ABS & CDOs Corporates Upgrades Downgrades Upgrades Downgrades Aaa 0.0% 0.3% 0.0% 9.5% Aa1 3.6% 10.8% 3.0% 16.5% Aa2 2.5% 1.6% 3.2% 15.0% Aa3 2.7% 3.6% 3.0% 14.4% A1 2.8% 1.0% 5.5% 12.0% A2 4.7% 0.5% 6.2% 11.8% A3 3.1% 3.6% 10.0% 12.8% Baa1 2.2% 10.3% 10.8% 12.7% Baa2 1.6% 2.7% 11.6% 10.8% Baa3 1.7% 12.0% 14.7% 12.2% Ba1 50.0% 12.9% 14.1% 12.0% Ba2 1.6% 8.1% 13.9% 12.7% Ba3 0.0% 12.7% 10.7% 15.5% B1 0.0% 17.9% 10.4% 14.1% B2 0.0% 7.7% 11.9% 16.7% B3 0.0% 19.0% 11.9% 18.4% Average 4.8% 7.8% 8.8% 13.6% Difference Upgrades Downgrades Aaa 0.0% (9.2)% Aa1 0.6% (5.7)% Aa2 (0.7)% (13.4)% Aa3 (0.4)% (10.8)% A1 (2.7)% (11.0)% A2 (1.5)% (11.4)% A3 (6.9)% (9.2)% Baa1 (8.6)% (2.4)% Baa2 (10.1)% (8.1)% Baa3 (13.0)% (0.2)% Ba1 35.9% 0.9% Ba2 (12.3)% (4.6)% Ba3 (10.7)% (2.8)% B1 (10.4)% 3.80% B2 (11.9)% (9.0)% B3 (11.9)% 0.6% Average (4.0)% (5.8)% Source: Moody's and Banc of America Securities LLC. Exhibit 10. Average One-Year Rating Upgrades/Downgrades, 1985-2000 CMBS Corporates Upgrades Downgrades Upgrades Downgrades AAA 0.0% 0.9% 0.0% 6.3% AA 1.9% 1.3% 0.6% 7.6% A 2.9% 1.6% 2.3% 5.9% BBB 3.0% 3.3% 5.1% 5.6% BB 2.4% 1.1% 7.2% 9.6% B 2.6% 2.8% 6.6% 9.8% Average 2.1% 1.8% 3.6% 7.5% Difference Upgrades Downgrades AAA 0.0% (5.5)% AA 1.3% (6.3)% A 0.6% (4.4)% BBB (2.2)% (2.4)% BB (4.8)% (8.4)% B (4.0)% (7.0)% Average (1.5)% (5.7)% Source: Standard & Poor's and Banc of America Securities LLC. Exhibit 11. Average One-Year Rating Upgrades/Downgrades, 1980-2000 RMBS Corporates Upgrades Downgrades Upgrades Downgrades AAA 0.0% 0.2% 0.0% 6.3% AA 4.7% 2.4% 0.6% 7.6% A 5.6% 2.4% 2.3% 5.9% BBB 5.7% 4.0% 5.1% 5.6% BB 4.2% 5.4% 7.2% 9.6% B 3.4% 4.6% 6.6% 9.8% Average 3.9% 3.2% 3.6% 7.5% Difference Upgrades Downgrades AAA 0.0% (6.1)% AA 4.1% (5.3)% A 3.2% (3.6)% BBB 0.6% (1.7)% BB (3.0)% (4.2)% B (3.2)% (5.2)% Average 0.30% (4.3)% Source: S&P and Banc of America Securities LLC. Exhibit 12. Outstanding US Taxable Bond Market Debt (Dollars in Billions) ABS & CDOs 1.2 Treasury & Agcy 4.8 Mortgage-Related 3.8 Corporate 3.6 Source: Banc of America Securities & Bond Market Association Note: Table made from pie chart. Exhibit 14. Year-to-Date Volume Composition (Dollars in Billions) 2001 2001 (%) 2000 2000 (%) US Public 170 62% 132 62% US 144A 28 10% 24 11% US Private 2 1% 1 0% Non-US 73 27% 55 26% Total 273 212 Source: Banc of America Securities & ABS Alert. Exhibit 15. US ABS Sector Composition Home Eq. 10% Other 27% Auto 22% Cards 25% Subprime Res. 16% Source: Banc of America Securities & ABS Alert. Note: Table made from pie chart. Exhibit 16. Non-US ABS Domicile UK 36% Italy 14% Australia 10% Japan 8% Netherlands 8% Other 24% Source: Banc of America Securities & ABS Alert. Note: Table made from pie chart.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||collateralized bond obligations|
|Publication:||The Securitization Conduit|
|Date:||Mar 22, 2001|
|Previous Article:||Introducing the BAS CDO ROE barometers.|
|Next Article:||US MBS market.|