Collateralized debt obligations (CDOs): identity crisis.
The term CDO is relatively new, coined by the rating agencies and others in late 1997 to describe asset-backed securities (ABS) backed by a broad variety of collateral mixes with a variety of structures introduced into the market. A collateralized loan obligation (CLO) is a CDO backed only by bank loans. A collateralized bond obligation (CBO) is a CDO backed only by bonds. Just to make things interesting, the term CDO can also refer specifically to a CDO subclass that has a mixture of loan and bond collateral or a CDO subclass with other types of fixed income collateral, such as mortgages. The typical structure of a CDO is shown in Exhibit 1.
A CDO is an ABS backed by various types of fixed income securities, bank loan collateral or a mixture of both. Varying senior/subordinated classes of passthrough debt (or certificates) and a modest amount of equity are issued by the special-purpose vehicle (SPV) that purchases the collateral from the seller. The collateral is held in trust and pledged to secure the repayment of the CDO. The SPV is a bankruptcy-remote entity and does not engage in any other business than managing the collateral. The SPV will engage an asset manager to manage the debt and this is usually an affiliate of the seller. To place a particular CDO in its proper category, the following items must be known:
1. Issuer motivation (balance sheet or arbitrage).
2. Underwriting approach (cash flow, synthetic or market value).
3. Whether the ratings of the CDO are linked to the seller or de-linked.
4. Credit quality of the underlying collateral.
5. Mix of collateral by type (bank loan, corporate bonds or percentage combination).
6. Geographic mix of collateral (U.S. domestic, non-U.S., emerging market or percentage combination).
This information distinguishes one CDO from another and, together with other details such as average life, rating and the experience of the asset manager, is needed to properly price a CDO and to determine its relative value.
II. COMPOSITION AND BRIEF HISTORY OF THE MARKET
The asset-backed market has grown dramatically over the past four years. This growth has been presented in Exhibit 2 in a slightly different fashion that more appropriately depicts the sources of growth and how the market has changed during this period. Most investors have focused on U.S. public ABS market volumes to gauge growth and the potential for increased liquidity in the market.
It is easy to see that the increased growth in the repackaged market (which includes CBOs and mortgage CBOs) and the international market more than offset the slowing of growth in the U.S. public and U.S. private markets. Some argue that repackaged ABS do not represent new issuance but merely refinance existing product. In some respects this may be true, however, the repackaged market has for the most part taken securities from other nonasset-backed markets (e.g., high yield bonds, loans and the mortgage markets) and offered new product to the asset-backed market, thereby providing liquidity to the market where the assets were originated. Also, from a truistic perspective, all asset-backed issuance represents the repackaging of financial assets.
In this regard, total growth in the asset-backed market is more accurately portrayed by including international and repackaged product, which shows total asset-backed issuance increasing 17% to $372 billion in 1998 from $319 billion in 1997 (Exhibit 2). If growth was measured by issuance in the U.S. public and private markets, the annual percentage increase would be only 11% to $267 billion in 1998 from $240 billion in 1997. Perhaps the most worrisome aspect of using just the U.S. numbers as a barometer for growth is the sheer size of the international and repackaged market, which reached $107 billion of new issuance in 1998 or 28.6% of the total market. This amount is up from only $12 billion in 1995, which accounted for less than 10% of the market.
III. ISSUER MOTIVATION (BALANCE SHEET OR ARBITRAGE)
There are two basic motivations for CDO issuance. The first is to transfer the collateral off the balance sheet of the issuer to achieve regulatory capital relief or manage the credit risk profile of the issuer's balance sheet (balance sheet motivated). Balance-sheet-motivated CDOs have been done mostly by banks. The second is a leveraged arbitrage of the high-yielding collateral by insurance companies and asset-management companies (arbitrage motivated) to generate asset-management fees and/or share in the attractive returns generated by the equity ownership.
IV. UNDERWRITING APPROACH (CASH FLOW, SYNTHETIC OR MARKET VALUE)
Cash flow, synthetic structures (credit-linked notes and default swaps) and market value are the three fundamental approaches used to structure CDOs. Cash flow and synthetic structures in the CDO market are similar in that the analysis of the collateral depends heavily on the expected loss. Meaningful analysis of the expected loss of leveraged loans and corporate bonds was made possible when Moody's Investors Service, Inc., and Standard and Poor's Corp. began publishing detailed histories of defaults on their respective rating universes. Moody's study, for example, includes all bonds ever rated by the agency and dates from the 1920s. Because the universe of rated issuers was so large and could easily be categorized by industry, it was statistically relevant to use these average default levels by rating category to predict what losses might be expected by a similar portfolio of bonds over a defined time horizon.
The key to this analysis is that the portfolios be "similar," and the rating agencies therefore required initial and ongoing industry and issuer diversification, so that loss experience would be "similar." Because the future expectation of losses is based on historical analysis, active management of a portfolio would invalidate the underwriting assumptions. We have labeled this approach as "passive portfolio management," or "buy and hold," and the rating agencies for this reason also limit discretionary reinvestment, or turnover, in the CDO collateral. Selling of investments is, however, allowed when the credit quality of an issuer or the average rating of the portfolio as a whole deteriorates outside original underwriting parameters. Proceeds from any sale of collateral are reinvested or used to retire the senior-most class.
In a cash flow structure, the cash flow net of losses is sized around the liability structure so sufficient cash flow is available to each class of security in the CDO to achieve the desired rating. The structure is then stressed in many different ways to ensure that the default expectation of the newly created CDO classes is consistent with the rating assigned by the rating agencies. The stress scenarios are patterned after historical worst-case economic situations. Some scenarios are based on historical experience and others are fictional but intended to attack the weakest point of the cash flows. For example, after running the cash flow stress scenarios, expected losses to the AAA tranche would be nil, which is consistent with the loss expectation of an investor in a AAA security. A typical structure may have a large AAA tranche, followed by subordinated tranches rated A, BBB and BB. A cash flow structure will pass through the principal and interest payments of the bond and loan collateral sequentially to the CDO certificate holders, less expenses (such as asset management fees and net losses due to the payment default of loans and bonds).
Synthetic structures have become an effective and useful tool for managing balance sheet risk and isolating credit risk. As with cash flow structures, performance of the issue depends on the default-and-loss experience of the collateral, which in this case is a "reference" portfolio. The synthetic CDO issuance is much smaller than the reference collateral, and the payment of principal and interest on the "new money" does not depend on cash flow from the reference pool.
There have been two approaches so far to synthetic execution--default swaps and credit-linked notes. Both obviously exploit the technology developed in the burgeoning credit derivatives market. All this sounds complex but, in fact, it is straightforward. The amount of capital raised by the CDO issuance covers the "expected loss" of the reference pool. In the case of default swaps, an ongoing fee is paid to the SPV for the portfolio default swap contract by the "seller," which in this case is the beneficiary of the default swap. The swap contract stipulates that at the end of a defined period (equal to the bullet maturity of the CDO issue), the SPV or trust is obligated to pay any net loss (par amount of the reference pool less any recoveries) that occurs less a first-loss deductible. The first-loss position is comparable to the equity provided in a traditional cash flow CLO. Issues to date have had 3- and 5-year maturities. The cash raised is invested in Treasuries or other high-quality current coupon (AAA) s ecurities with maturities prior to the term of the issue. The current income from the Treasuries, together with the default swap fees, is used to pay the current interest when due on the various classes of the CDO issuance. The classes are tranched in the same manner they would be for a CLO.
To date, the reference pool has exceeded the related CDO issuance from two to nine times. At maturity, the cash available from the Treasuries would be used to pay any amounts due on the default swap (no payment is due until maturity under the swap agreement). The remaining cash could then be used to pay the principal of the CDO classes in sequential order (senior to junior), so that the most subordinated tranche would be the first to incur a loss if losses exceeded the first-loss deductible. The risk transfer achieved by the synthetic CLO of the reference pool and associated regulatory capital savings could also be achieved by a traditional CLO. For a large bank, however, on-balance-sheet funding will be 2O bps-30 bps cheaper (all-in cost) than a traditional CLO. A synthetic CLO would not be the desirable execution for those wishing to achieve generally accepted accounting practices (GAAP) sale treatment. From a pure economic or financing execution perspective, a synthetic CLO would be the best choice for a b ank.
Although it is unlikely that any bank would sell or securitize 50% of its assets, the example does demonstrate, and exaggerate, the effect a synthetic or traditional CLO would have on regulatory and GAAP returns. A credit-linked note synthetic CLO would use credit-linked notes rather than default swaps. The credit linked notes would be obligations of the bank buying the default protection and would require payment of principal and interest on the note by the bank only if the reference credit did not pay. Because these notes are obligations of the bank, the highest rating that could be obtained on the synthetic would be the rating of the bank. In this case, the rating of the CLO would be linked to the sponsor/seller. Otherwise, a CLO collateralized by credit-linked notes would have to meet all the requirements of a traditional cash flow CLO.
The third fundamental approach to structuring a CDO is a market value transaction. In this approach, the majority of the collateral is marked to market frequently (at least weekly) and the tranching of CDO classes are sized around an advance rate (the inverse of overcollateralization), which is based on historical studies of market price volatility of the collateral by type and rating. If the historical analysis shows one collateral class to be more liquid and less volatile than another, then the advance rate for that collateral will be higher than for the others.
Recently, the rating agencies, particularly Moody's, have added the dimension of portfolio theory. The new approach is described as primarily historical, but it also imposes volatility assumptions on asset classes and correlation assumptions within and across asset classes. Higher volatilities were imposed when Moody's felt there was not sufficient data to draw reasonable conclusions about the behavior of the asset. Simulations were then run using the historical parameters and the ones added to generate advance rates needed to achieve certain rating levels in a portfolio context. This approach therefore considers interest changes, investor preferences, rating, duration and convexity based on certain portfolio assumptions to arrive at advance rates. Higher advance rates are generated in this case than if the historical volatility of the asset was measured on a stand-alone basis.
Avoiding the math, common sense dictates that some asset classes, and even different securities within asset classes, can have different convexities. Discovering that mortgages can be negatively convex and 30-year passthroughs can have six-month durations is one of those life events after which you always remember where you were and what you were doing when it happened. Presumably, you were reading a book and not managing your clients' money. Exhibit 8 presents a sampling of the advance rates for a portfolio of 20 issuers, five industries and 100% investment in one asset type with a 5-year maturity.
If investment in more than one asset type is contemplated, all of the advance rates in Exhibit 8 would change to a different target. A new set of simulations would be run based on desired portfolio characteristics. Market value CDOs have been the minority of issuance in the market, even with this refined technology. Some investment managers are drawn to the structure due to the flexibility of investment into varied asset classes and the freedom to trade securities as long as portfolio constraints, including the advance rate, are met. In a CDO, an advance rate would be set for each tranche, with collateral to be liquidated and used to pay down principal until the program meets the advance rate tests for each class. Tranches will usually be soft bullets occurring on the same day with paydown occurring from an orderly liquidation of the portfolio over a defined period that would not exceed one year. Most structures provide for a one-year extension if adverse market conditions persist.
Market value CDO collateral managers are total-return managers. The choice of assets and the trading flexibility afforded by these structures allow the manager to take a "view" of the market. The manager is also concerned with liquidity and whether any credit event will affect the market value of the holdings.
In contrast, the collateral manager of a traditional cash flow CDO will take a credit view of the portfolio without being concerned about the current market value of the holdings. The strategy for cash flow CDOs is much more buy and hold and seeks to avoid credit deterioration of the portfolio.
V. STRUCTURAL AND LEGAL CONSIDERATIONS
A. True Sale
The issue of true sale is less problematic for bond collateral, where securities are acquired in the market, than for the transfer of loan collateral, which may take many forms. The transfer of loan collateral from a financial institution can take the form of an assignment, a secured participation or a participation or that of credit-linked notes or default swaps. A true sale can only be accomplished with assignments or secured participations. In the event of the bankruptcy of the seller/transferor of assets to the SPV, a creditor of the seller may attempt to recharacterize the transfer as a secured financing instead of a true sale. If successful, the assets would be considered as assets of the transferor and the automatic stay provisions of the bankruptcy code would apply. For a true sale to occur, the transferor must convey the future economic risks and benefits of ownership of the assets to the issuer.
B. Transfer of Loan Collateral
With regard to an assignment of a bank loan, the transferee (SPV or trust) becomes the lender of record. In this case, a true sale has clearly occurred and the seller is no longer involved with the loan. Provisions that could lead the Federal Deposit Insurance Corp. (FDIC) or a court to conclude that the participation was a loan to the trust and not a true sale are repurchase provisions that are subordinated to the seller's share to other participants, a shorter term or a less-than-identical term to the original loan or special payment provisions not solely dependent on the payment from the borrower. A perfected security interest in all the assets of the issuer maintained by the trustee for the benefit of the holder of any secured participation overcomes these concerns for the purpose of determining the participation to be a true sale. Credit-linked notes and default swaps are conditional contracts and do not involve the concept of true sale. These contracts can transfer risk and either be linked or not linke d to the issuer of the contract and do not involve the transfer of real property.
C. Cash Flow Triggers
Cash flow triggers, usually a multiple of required interest payments, and a multiple of overcollateralization are used to redirect the collateral cash flow from subordinate tranches in a CDO to the more senior tranches and are set at levels to maintain the rating of the senior tranches receiving the redirected cash flow. The redirected cash flow is used to pay down principal, reducing the amount of interest paid, until the required coverage test is met, at which time cash flow again becomes sequential. This priority claim is an important feature of CDOs and must be modeled correctly for any cash flow analysis of the transaction.
D. Market Value Triggers
Market value triggers are used in market value CDOs to maintain the rating of each class in the structure. In most cases, the deficiency must be remedied quickly within a required time frame and would result in the sale of assets to reduce the principal of the class where the trigger was breached, bringing the transaction back into line with the market value test. If not cured within the required time frame, an early liquidation of the portfolio would ensue and the entire issue would face a special early redemption.
E. Credit Quality, Diversification and Trading Limitations
These tests are used during the reinvestment period or during the ramp-up period of a prefunded or delayed draw CDO. The tests are designed to ensure that the original credit quality and diversification of the collateral pool, the basis on which it was originally rated, do not change over the life of the transaction. In cash flow transactions, trading is generally limited to 10%, or up to 20% for some collateral types, or to selling assets that have deteriorated significantly in credit quality or appreciated significantly in price. The rationale for this requirement is to ensure that the original portfolio characteristics are maintained for the life of the transaction.
VI. COLLATERAL DISCUSSION
As with all ABS, ultimate repayment depends on the collateral and the certainty with which cash flow from that collateral can be expected. If one collateral pool is more volatile than another, a higher level of over collateralization will be needed in that pool to achieve the same level of certainty as the other.
The collateral is diversified broadly into a large number of industries as required by the rating agencies and has varied credit quality based on the objectives of the issuer. The issuers have been banks, insurance companies and asset-management companies. Most of the collateral has been higher-quality (BB/B) high yield bonds or senior secured bank loans. Recently, investment-grade collateral has been used in bank balance sheet structures where the motivation has been regulatory capital savings. The freeing of regulatory capital for other risk-taking activities more than offsets the lower excess spread of investment-grade collateral over the cost of the asset-backed issuance.
A. Leveraged Loan Market
Historically, banks have syndicated loans of both investment- and speculative-grade quality as a risk-management tool (similar to reinsurance in the insurance industry) to redistribute single-transaction risk to many participants in the banking industry. This has allowed the banks to originate loans that exceed their legal lending limits. The market is also becoming increasingly similar to the bond market, with the same documentation, same daily trading conventions and similar trading confirmations. Basic motivations for issuance of highly leveraged loans include acquisition, general corporate purposes, debtor in possession financing, exit financing, expansion or capital expenditures, leveraged buyout, merger, project financing, recapitalization, refinancing and spin-offs. Over the past year, 38% of total issuance has been for acquisition, 29% for leveraged buyouts and 15% for refinancing. CDOs, and the continued development of the swap market, have provided liquidity and technology to the market and enabled banks to compete with securities firms for traditional high yield market product.
Because of the remarkable growth and commoditization of the leveraged loan market, many investors are becoming increasingly aware of the market. Consolidation in the banking industry has allowed larger syndications to fewer participants. European banks are still active participants in the market, and the volume lost by the exit of the Pacific Rim banks from the leveraged loan market has been offset by the advent of institutional investors taking a more active role in the new issue market. Participating in this sector directly or through the CDO market represents an opportunity to achieve superior returns, captured as understanding of the sector increases, the efficiency of the market for loan product and for CDOs increases and spreads narrow over time. Since 1987, the average debt multiple of the leveraged loan product has improved dramatically (Exhibit 12).
A leveraged loan is defined by the market as any loan priced at LIBOR + 225 bps or greater when originated. Prior to 1996, the hurdle was LIBOR + 250 bps. The average debt multiple of highly leveraged loans originated in the first quarter of 1999 was 4.5x EBITDA, which is substantially below the 8.8x multiple for highly leveraged loans originated in 1987. Since 1990, the leveraged loan market has been much more sensibly leveraged with multiples remaining in the relatively narrow range of 4.5x-5.5x. Primarily as a result of the credit crisis in the fourth quarter of 1998, the fourth quarter of 1998 and the first quarter of 1999 saw multiples contract to around 4.5x from the 1998 average of around 5.5x.
The leveraged loan market is a subclass of the $872 billion bank syndicated loan market. The leveraged loan component grew to $273 billion in 1998 and consists primarily of senior secured bank term loans. Most arbitrage CLOs will invest in the term loan segment, referred to as institutional loans, which are normally sold to nonbanks. Institutional term loans typically have final maturities that range from six to 10 years with required amortization beginning at the midpoint (three to five years) that fully amortize by the final maturity.
The reinvestment period (typically three or five years) and the tranche structure of the CLO are designed to closely match the amortization of the loan collateral beginning in Year 3 or Year 5. If a prepayment of a loan occurs during this period, the investment manager must find another investment that meets the original industry and trading requirements of the CLO and has a maturity prior to the final legal maturity. After the end of the reinvestment period, the collateral begins to amortize and the CLO tranches begin to pay down. The loss experience of the leveraged loan market has been excellent compared with that of the high yield bond market (Exhibit 15).
B. High Yield Bonds
The high yield bond market emerged in the 1980s and grew dramatically as Michael Millken's Drexel Burnham Lambert offered term fixed-rate funding through the public markets to high-growth companies and others to which the banks would only lend floating-rate funding. The high yield market allowed capital structures to be created for growth companies that were appropriate for their industry, albeit highly leveraged. MCI WorldCom, Inc., has become the often-cited example. Just 15 years ago, Michael Millken predicted the virtual bank, which would become a conduit for managing risk originated by the bank and sold to investors. The banking community seems well on its way there.
The high yield market experienced record growth for the third consecutive year as new deals increased to 19.7% over the prior year. Total dollar-denominated new issuance reached $151 billion in 1998, although $106 billion occurred during the first half. The prolonged U.S. economic expansion, coupled with low inflation, led to record 144A new issuance in 1997 and 1998. The media and telecommunications sector continued to represent the bulk of new issuance in 1998, accounting for $60.2 billion. Exhibit 16 shows the volume of high yield new issuance.
Outstandings increased to approximately $570 billion at year-end 1998, rising 26% over 1997. The domestic default rate remained below the historical rate but increased to 3.84% in 1998 from 2.14% on a per issuer basis according to Moody's default study. The U.S.-only speculative-grade default of 3.84% in 1998 was higher than the worldwide speculative-grade default rate of 3.31%, which has generally been the case since 1989. This should change in 1999 as the number of emerging market defaults is increasing in the first half of 1999 and as a result of the ongoing significant growth of the European high yield market. At year-end 1998, the market was demanding a 400 bps spread over the 3.84% default rate, which is significantly above the 125 bps average spread required in 1997. This credit risk premium on high yield bonds has receded to more normalized levels in 1999 but remains above the levels experienced prior to the fourth-quarter 1998 credit crisis. Exhibit 17 compares the annual default rates of the total w orldwide speculative-grade market rated by Moody's against the U.S. speculative-grade market.
VII. RELATIVE VALUE
The relatively young CDO market is far from being efficient. Spreads on similarly rated tranches will depend on a number of factors. We believe that as the CDO market grows and becomes more efficient investors will place more emphasis on the credit quality of the collateral. In the future, spreads over Treasuries of similarly rated tranches of CDOs should parallel the volatility of the expected losses of the collateral. As should be clear at this point, the volatility of credit losses depends on the type of collateral, the average rating of the collateral and its geographical mix. Most CDO new issuance can be grouped into the structures shown in Exhibit 18.
The CDO market has been a buy-and-hold market until recently. Given the recent growth in CDO volume and the term structure of the product, CDO outstandings now rival credit cards and home equities for the top spot. Increased volume and greater investor understanding are creating greater liquidity for the product in the higher-rated classes and, as a result, secondary market activity is increasing for these levels. Once the market becomes more efficient, as has been the case with the development of markets in all asset classes, spreads should vary with the volatility of the underlying collateral, with the least volatile commanding the tightest spread to Treasuries. Currently, other factors, such as call protection (usually three to five years), the length of the reinvestment period, the strength of the manager, the diversity score and the combination of rating agencies will affect the trading value of CDOs to a greater degree than expected collateral performance and cause CDOs to trade in wide ranges, often wi thout regard for the collateral groupings above. Our approach is to start with the collateral groupings above and make adjustments to the idealized spread for the other factors. New issue spreads drive secondary market pricing. Key new issue pricing for AAA floating-rate CDO issues is plotted in Exhibit 19.
The disparity in spreads in the fourth quarter of 1998 can be explained in large part by the credit crisis and the volatility of the underlying collateral. The lower-rated classes of a CDO will demand a higher new issue spread for the increased risk. As a rule, these tranches are not actively traded. These spreads are best examined in the context of the comparable transactions listed in Exhibit 24 on page 36.
In summary, a solid understanding of the volatility of the underlying credit quality and diversity of the collateral as well as the key structural features mentioned above and an awareness of CDO new issue spreads are needed to fairly price a CDO.
VIII. RATING AGENCY APPROACH
A. General Approach
Rating agencies assign a AAA to a CDO transaction so that an investor, from a credit perspective, will regard that security as having the same creditworthiness as any other AAA at that moment. The rating agencies are looking to achieve a certain rating consistency across all fixed-income sectors at all rating grades. Moody's and Standard and Poor's were the first to rate structured transactions. In a CDO transaction, the rating of the structured bond is based on the sufficiency of the cash flow generated by an underlying portfolio of corporate bonds or bank loans. The transaction reappportions the risk and the return of the collateral pool according to the preferences of the senior and subordinate investors.
B. Moody's Investors Service, Inc.
Moody's rating methodology for cash flow corporate debt-backed securities compares the credit risk inherent in the underlying portfolio with the credit protection offered by the structure.
Today, Moody's analyzes most CDOs by using the binomial method, which is a derivative of the expected value concept. This method essentially reduces, for modeling purposes, the actual pool of collateral assets (typically a pool of heterogeneous assets with correlated default behavior) to a homogeneous pool of uncorrelated assets via the diversity score. The diversity score represents the number of independent, identical assets that have the same loss distribution as the initial pool. Given that the diversity score for any collateral portfolio becomes the number of independent, identical assets in the portfolio to be modeled, the probability of default (same as weighted average of original, taken from Moody's annual default study) and recovery can be determined and (same as original portfolio, also taken from Moody's annual default study) the collateral cash flow projections can be stressed easily. Using a simple formula, a probability for each possible default path (from 0 to the total number of identical ass ets in the collateral pool) and the loss for each default path can be calculated. The expected loss of the stress scenario then becomes the sum of the expected loss of each of these independent default observations. The expected loss is then compared to the yield change limits described below for each rating class above to determine the rating.
A Moody's rating is its opinion about the likelihood of full and timely payment on a rated security. The rating does not address any of the liquidity, market or basis risk of the security but is used to isolate the credit risk. Put a different way, the rating addresses the joint effect of the frequency and the severity of the expected future defaults on that security. Technically, Moody's rating quantifies the amount by which the internal rate of return (IRR) on a diversified portfolio of similarly rated securities would be reduced as a result of defaults on the securities, assuming that all of the securities are held to maturity regardless of any changes in the ratings. For example, a diversified portfolio of securities rated Aaa by Moody's at the time of purchase and held to maturity regardless of any changes in the ratings is expected to suffer a reduction in realized yield of 0.0006 percentage points (i.e., six one-hundredths of a basis point) as a result of defaults. Moody's expects that two diversified portfolios of structured finance securities with the same rating should have roughly the same expected change in IRR caused by credit losses. In addition, those changes in IRR from credit losses should be roughly comparable to the changes in IRR from defaults on portfolios of traditional corporate bonds (Exhibit 23).
These yield change limits are not arbitrary, but are based on the expected value concept such that any expected loss for a Aaa rated security is at least 3 standard deviations away from the mean of any worst-case scenario, or in other words, you have greater than a 99% chance of principal and interest being paid when due. (Aa is 2.5 standard deviations, A is 2.0 and Baa is 1.5). The worst-case scenarios were originally modeled after the Great Depression and events that occurred in the early 1980s when the Federal Reserve Bank switched to controlling the money supply. Some of the scenarios used today are intended to attack a structure at its weakest point. For CDOs, this is usually the first year where Moody's might require 50% of the total expected loss to be realized.
An accurate cash flow model is a critical assumption in Moody's analysis. Diverted cash flows can be used to fund reserve accounts or pay senior principal directly when certain credit quality triggers are violated, either of which would affect the expected loss on other tranches of the transaction. If the transaction was modeled correctly, the probabilities of such events would be reflected in the rating of each of the affected tranches. The documents are reviewed in detail by the analyst rating the transaction to ensure the cash flows have been modeled correctly.
Moody's, like the other rating agencies, looks carefully at certain legal issues discussed earlier in this report, including bankruptcy remoteness, and certain issues surrounding the legal form of the issuer (corporate, limited partnerships), asset transfer, the risk of substantive consolidation and whether subordinated tranches interfere in any way with the enforceability of remedies available to the more senior tranches of multiple tranche transactions.
C. Standard and Poor's Corp.
Standard and Poor's (S&P) approach to rating CBOs and CLOs consists of three main areas--the credit analysis of the asset pool, structural risks and a legal analysis.
S&Ps evaluation of the credit risk of the portfolio includes the credit quality of each obligor, obligor concentration, industry concentration and the adequacy of the credit enhancement. S&P uses its ratings as a measure of default probability. These probabilities have been derived from S&P's default study, which has measured the average defaults of all securities rated by S&P over different time periods. In the absence of an obligor rating, S &P may use the internal scoring models of the seller if there is a reasonable correlation of the seller's ratings to that of S&P. S&P and Fitch IBCA have been more flexible on this issue than Moody's. S&P defines loss severity as the expected loss on default after factoring in or netting any recovery proceeds on the defaulted asset. Recoveries have also been measured by ranking of the debt (senior secured, senior, senior unsecured, subordinated and junior subordinated) as part of the annual default study. The analysis may vary due to the particular risks of each pool of collateral in that certain industries or types of obligors may have better or worse recoveries than others in the same ranking. If the transaction involves swaps or other hedging or support arrangements, the credit ratings of the support provider or counterparty generally must be of credit quality equal to that of the highest rating assigned to any of the issue's tranches.
The model recognizes multiple sources of credit protection that may include overcollateralization/ subordination, cash collateral/reserve accounts, excess spread, financial guarantee insurance and other forms of support. The cash flow analysis is aimed at evaluating the availability of funds for full and timely payment of interest and principal in accordance with the terms of the rated securities. This is done by analyzing the payment structure and the amount of debt to be supported to ensure that the level of credit support afforded each tranche of the transaction is consistent with the desired rating.
S&P does not, however, set specific obligor concentration guidelines. A pool of 40 obligors or 2.5% concentration limit per obligor is likely to experience lower pool losses than a pool of 20 such obligors or 5% concentration limit per obligor. For analytical purposes, small pools consist of bond or loan obligations of fewer than 10 obligors. In these collateral pools, S&P may conduct an obligor-by-obligor default analysis.
Industry diversification limits exposure to any particular industry and thus limits the potential loss exposure in a given economic environment. S&P realizes there is a diminishing marginal benefit to diversifying a portfolio by industry. In cash flow transactions, S&P research indicates that a pool of bonds or loans representing 13 industries, each accounting for no more than 8% of assets per industry, is fairly diversified. S&P requires five different sets of stress scenario assumptions for the timing of defaults to determine the level of credit enhancement needed for each tranche of the transaction to be rated.
The most important element of Fitch's approach to rating cash flow CDOs is the default probability of the collateral assets as determined by stressing points along the Fitch default curve, which plots ratings against the percentage probability of default. As the rating decreases, the default probability increases. Fitch requires a bankruptcy-remote entity and diversification of assets. Fitch 's criteria are designed to evaluate the credit enhancement of debt for various rating levels. The analytical framework is based on an expected loss analysis equal to the product of the default frequency of an asset in the pool and the expected loss severity on the default of that asset. The expected loss calculation is performed using a cash flow model that factors in the timing of defaults, recoveries and interest rate movements. The output of the model is a key determinant of the amount of credit enhancement required to support a given rating. Fitch's default curve is a cumulative 10-year default probability taken from previously published studies by others and serves as a baseline for measuring the default probabilities of collateral pools to be securitized. The baseline curve is stressed by multiples to derive a range of default probabilities for each rating category. The range for the four investment-grade categories (BBB to AAA) encompasses points equal to 1.0, 2.0, 3.0 and 5.0 standard deviations from the mean observed defaults for each collateral rating. Additional factors include the final maturity of the rated debt, the experience of the asset manager and the relative concentrations (by issuer, industry group and geographic location) within the proposed portfolio. Fitch also requires a weighted average rating guideline for actively traded portfolios. The timing of defaults is spread over a five-year period with 33% percent of the assets defaulting at the end of the first year, which is a less severe test than Moody's 50%. A ~vell-diversified pool is defined by Fitch to limit issuer concentration to 4% of the total pool and industry concentrations to 10%. In summary, Fitch's cash flow approach to rating CDOs is similar to that of S&P.
IX. INVESTOR CONSIDERATIONS
The credit ratings of CDOs have been remarkably stable since their inception in the late 1980s. In 1998, Moody's downgraded tranches within 20 transactions, more than any other year, primarily due to the sheer number of cumulative transactions rated, downgrades of Japanese bank CLO sponsors and the emerging market crisis. Eight of the 20 were linked to their sponsor and were downgraded when the bank was downgraded. Twelve of the 20 were linked to the deterioration of the emerging market component of collateral pool backing the issues. In the first quarter of 1999, Moody's downgraded tranches within seven CDOs. Five of the seven were due to bank downgrades, one was due to a deterioration in a collateral pool consisting of U.S. and Canadian credits and the other was due to continued fallout from the emerging market crisis. The performance downgrades occurred primarily in the subordinated tranches. Of the approximately 300 CDOs rated by Moody's, there have been no defaults in the 13-year history of the market, d espite the most widespread emerging market crisis since the 1930s in 1998. Only emerging market transactions with 25% or more exposure to emerging markets were placed on review for downgrade or were downgraded by Moody's. Those with less exposure did not suffer credit deterioration outside of expectations and were not downgraded. The trend of no defaults may not continue, but rating migration to lower ratings and the default experience of CDOs will continue to be much less than that of the corporate market.
The benefit of a diversified portfolio (50-70 issuers, 18-25 industries) can be obtained with a small investment.
Favorable risk/return characteristics
Leveraged loans and high yield bonds, as investment classes, have outperformed most other fixed-income markets on a risk-adjusted basis.
Investment strategy that cannot be directly recreated by investors
The research staff and operational infrastructure needed to properly manage a high yield or leveraged loan portfolio require a significant investment. If a CDO tranche is purchased, a highly experienced asset manager has already completed the collateral credit analysis, and there is no need for the investor to develop a large internal staff to make investment decisions. Investment in many of the risk classes of a CDO offer exposure to the market at a minimal transaction cost and may provide the experience needed to enter the market or a subsector of the market at some point in the future.
Free of event risk
As with other asset-backed classes, the bankruptcy-remote nature of the issuer, the sequential pay and passthrough structure, diversified collateral pool and the comfort of actuarial analysis of collateral performance render CDOs largely free from event risk. This is not to be confused with headline risk or illiquidity during a credit crisis, where spreads will temporarily widen in concert with the swap and spread markets.
Secondary market liquidity improving
As the volume of outstandings in this asset-backed class continues to grow, secondary market activity will increase as more investors become familiar with the asset class and understand its behavior. In general, the illiquidity premium demanded by the investor should narrow over the intermediate term. Secondary market activity should be driven in the near future by a fine-tuning of exposure to the asset class rather than an exit altogether.
Exhibit 5 1997 CDO Insurance by Seller Type Bank/Thrift 60% Investment Company 5% Securities Company 21% Other 14% Note: Table made from pie chart Exhibit 6 1998 CDO Insurance by Seller Type Bank/Thrift 43% Investment Company 30% Securities Company 20% Other 7% Note: Table made from pie chart Exhibit 7 Comparison of a Traditional CLO and a Synthetic CLO $5 Billion Basic Bank Basic Bank Nonaccruing Traditional CLO Balance Sheet Assets $10 billon $5.075 billion Total Regulatory Risk Assets $10 billon $5.075 billion Tier 2 Regulatory Capital $800 million $475 million GAAP Equity $600 million $305 million (after stock buyback) Asset Yield LIBOR + 60 LIBOR + 60 Bank Cost of Funds LIBOR - 8 LIBOR - 8 CLO/Synthetic Cost of Funds NA LIBOR + 30 Pretax Regulatory Return on Capital 13.34% 14.39% Pretax GAAP Return on Equity 17.80% 22.41% Securitized/Reference Pool NA $5 billion CLO Issuance NA $5 billion AAA Rated Tranche NA $4.727 billion A Rated Tranche NA $99 million BB Rated Tranche NA $99 million Equity/Deductible NA $75 million Number of Loans/Reference Entities NA 145 Weighted Average Credit Quality NA BBB Minimum Credit Quality NA BB- Diversity Score NA 90 Average Size NA $30 million $5 Billion Basic Bank Synthetic CLO Balance Sheet Assets $10 billion Total Regulatory Risk Assets $5.075 billion Tier 2 Regulatory Capital $475 million GAAP Equity $600 million Asset Yield LIBOR + 60 Bank Cost of Funds LIBOR - 1 (including 14 bps cost-of-default swap) CLO/Synthetic Cost of Funds LIBOR + 30 Pretax Regulatory Return on Capital 17.47% Pretax GAAP Return on Equity 13.80% Securitized/Reference Pool $5 billion CLO Issuance $672 million (13.4% of reference pool) AAA Rated Tranche $474 million A Rated Tranche $99 million BB Rated Tranche $99 million Equity/Deductible $75 million Number of Loans/Reference Entities 145 Weighted Average Credit Quality BBB Minimum Credit Quality BB- Diversity Score 90 Average Size $30 million CLO: Collateralized loan obligation; GAAP: Generally accepted accounting practices. Source: First Union Capital Markets Corp. Exhibit 8 A Sampling of Advance Rates Asset Type Aaa Target A2 Target Baa2 Target Performing Bank Loans Valued $.90 87.0% 91.0% 94.0% and Above Distressed Bank Loans Valued $.85 76.0% 81.5% 87.0% and Above Performing High Yield Bonds 76.0% 84.0% 88.0% Rated Ba Performing High Yield Bonds Rated B 72.0% 79.0% 83.0% Distressed Bank Loans Valued 58.0% 68.0% 74.0% Below $.85 Performing High Yield Bonds 45.0% 58.0% 67.0% Rated Caa Distressed Bonds 35.0% 48.0% 57.0% Reorganized Equities 31.0% 46.0% 52.0% Source: Moody's Investors Service, Inc. Exhibit 13 Primary Market for Highly Leveraged Loans by Investor Type in Q1 1999 Domestic Banks 30.8% European Banks 20.0% Finance Companies 4.4% Insurance Companies 2.3% Loan and Hybrid Funds 29.5% Securities Firms 1.6% Asian Banks ** Canadian Banks 7.3% Note: Excludes hybrids as well as all left and right agent commitments, including administrative, syndication and documentation agent as well as arranger. Source: PMD Investment Co. Note: Table made from pie chart *[Unreadable in original source] Exhibit 14 Primary Market for Highly Leveraged Loans by Investor Type in 1998 Domestic Banks 27.3% European Banks 21.0% Finance Companies 4.3% Insurance Companies 4.8% Leasing Companies 0.2% Loan and Hybrid Funds 25.8% Securities Firms 1.8% Thrifts 0.5% Asian Banks 7.1% Canadian Banks 7.3% Note: Excludes hybrids as well as all left and right agent commitments, including administrative, syndicator and documentation agent as well as arranger. Source: PMD Investment Co. Note: Table made from pie chart Exhibit 18 Collateral Volatility Characteristics by Type of CDO % Loan % Bond % U.S. % Emerging Type Acronym Collateral Collateral Collateral Market I CLO 100% 0% 100% 0% II CLO 100% 0% 75%-85% 15%-25% III CDO 75% 25% 100% 0% IV CDO 75% 25% 75%-85% 15%-25% V CBO 0% 100% 100% 0% VI CBO 0% 100% 25%-50% 50%-75% Collateral Type Volatility Ranking I Low II Low/Moderate III Low/Moderate IV Moderate V Moderate VI High CBO: Collateralized bond obligation CDO: Collateralized debt obligation CLO: Collateralized loan obligation Source: First Union Capital Markets Corp. Exhibits 20 CDO Ratting Statistics by Agency Combination (1995-1998) Moody's/S&P 58% Moody's/Fitch 21% Moody's/D&P 9% S&P/D&P 8% S&P/Fitch 4% Note: Table made from pie chart Source: Asset-Backed Alert and First Union Capital Markets Corp. Agency No. % Moody's/S&P 86 57.7% Moody's/Fitch 32 21.5% Moody's/D&P 13 8.7% S&P/D&P 12 8.1% S&P/Fitch 6 4.0% Other Combinations (14) NA Total 135 100.0% Not Rated 32 7.6% Single Rating 255 60.4% Multiple Ratings 135 32.0% Total 422 100.0% Source: First Union Capital Markets Corp. Exhibits 21 CDO Rating Statistics by Category (1995-1998) Single Rating 60% Multiple Ratings 32% Not Rated 8% Note: Table made from pie chart Exhibits 22 CDO Ratting Statistics by Agency (1995-1998) Moody's 50% S&P 28% Fitch 12% D&P 8% Other 2% Note: Table made from pie chart Agency No. % Moody's 280 51.5% S&P 151 27.8% Fitch 63 11.6% D&P 41 7.5% Other 9 1.7% Combinations (154) NA Total 390 100.0% Source: Assert-Backed Alert and First Union Capital Markets Corp. Exhibit 23 Moody's Limits on IRR Change Due to Credit Losses by Rating Category Yield Change Rating (bps) Aaa 0.06 Aa2 1.30 A2 9.00 Baa2 27.00 Ba2 106.00 B2 231.00 Source: First Union Capital Markets Corp. Exhibit 24 Comparison of Recent CDO Transactions Athena CDO Ltd. Blank Diamond Size of Offering $251 million $425 million Rating Coupon % WAL Rating Coupon % WAL Structure Aaa L + 75 66.93% AAA L + 60 60.00% 5.7 Baa3 7.83% 21.12% A- L + 150 16.25% 7.8 Equity 11.95% BBB- L + 225 6.25% 9.1 Equity 7.25% Percentage of Capital Structure Below Investmnet Grade 11.95% 7.25% Collateral Minimum 75% Minimum 82% Senior Senior Secured Secured Bank Loan Bank Loan Maximum 18% High Maximum Yield Bonds 25% High 0% Emerging Yield Bonds Markets Minimum Diversity 30 35 Minimum Average Rating 2675 (B2) 2720 (B2) Percentage of Collateral Beyond Deal Maturity 0% 0% Percentage [less than or equal to] CCC+ 5% 5% Percentage Paying Zero Interest 0% 5% Minimum Average Coupon 9.00% at 10.05% at 100.0% Minimum Average Spread 100.0% 250 pps 275 bps at 100.0% Final Maturity at 100.0% 12 years Closing Date 12 years Dec. 23, 1998 Dec. 16, 1998 Mountain Capital Summit CBO I, Ltd. ELC CDO 1999-1 Size of Offering $425 million $350 million $500 million Rating Coupon % WAL Rating Coupon % WAL Structure AAA L + 65 77.66% 5.7 AAA L + 65 66.23% 8.50 A- L + 150 6.88% 7.8 AA- L + 100 10.80% 12.00 BBB- L + 250 8.45% 9.1 BBB T + 350 10.00% 12.00 Equity 7.00% 9.1 BB- 5.00% 12.00 Equity 7.97% Rating Coupon % WAL Structure AAA L + 57 40.00% 6.90 AAA L + 57 37.00% 6.90 A- T + 130 9.50% 9.70 BBB- T + 260 4.75% 10.70 BB- 2.00% 11.40 Percentage of Capital Structure Below Investment Grade 7.00% 12.97% 9.93% Collateral 100% Bank Minimum Minimum Loans 100% High 95% Senior 0% Yield Secured Bank Emerging Bonds Loan Maximum Markets 5% Senior Unsecured Bank Loan Minimum Diversity 35 44 NA Minimum Average Rating 2720 (B2) 2600 (B2) B rating Percentage of Collateral Beyond Deal Maturity 0% 3% Percentage [less than or equal to] CCC+ 5% 5% 5% Percentage Paying Zero Interest 0% 0% 0% Minimum Average Coupon 9.25% at Minimum Average Spread 250 bps 100.0% 265 bps at Final Maturity at 100.0% NA 100.0% Closing Date 12 years 12 years 12 years Not Settled May 10, 1999 WAL: Weighted average life. Source: First Union Capital Markets Corp.
R. RUSSELL HURST joined First Union in the Asset-backed Research Department in February 1999. As a Senior Analyst and asset-backed generalist, he covers all asset classes with primary responsibility for the Collateralized Debt Obligation (CDO, CLO and CBO), Home Equity, and Credit Card securitization markets. His research experience includes asset-backed, worldwide financial institutions, distressed, and industrials, both investment grade and high yield. Functional experience includes a unique blend of buy-side, origination and rating agency experience. Prior to joining First Union, he held positions at Chase Asset Management, which included Head of Structured Investments and Global Head of Credit Research. Prior to joining Chase, he was Managing Director and head of Research for Financial Security Assurance (FSA), a mono-line financial guaranty firm, and a member of the executive underwriting committee for taxable and tax-exempt insured transactions. His prior experience also includes positions as a Senior A nalyst at Moody's Investor's Service and as a lending officer at the Irving Trust company. Rusty is a member of the Fixed Income Analysts Society. He holds a BA in Political Theory form Tulane University as well as an MBA in finance from the Tulane Graduate School of Business.
|Printer friendly Cite/link Email Feedback|
|Author:||Hurst, R. Russell|
|Publication:||The Securitization Conduit|
|Date:||Sep 22, 2000|
|Previous Article:||Evaluating credit enhancement floors in equipment ABS.|
|Next Article:||Aircraft securitization cleared for take-off.|
|CREMAC liquidates $180m of its bonds.|
|The CDO product.|
|Marubeni Announces Establishment of New Asset Management Business.|