Skip to content

RESERVE BANK OF AUSTRALIA

Financial Stability Review – September 2005

Collateralised Debt Obligations in Australia1

Introduction

Collateralised debt obligations (CDOs) are securities that are exposed to the credit risk of a number of corporate borrowers. In the simplest form of a CDO, this credit risk exposure is generated in the same way as for any asset-backed security (ABS): the CDO is backed by outright holdings of corporate debt, such as corporate bonds and corporate loans. Increasingly, however, the exposure to corporate credit risk is synthesised through the use of credit derivatives. Unlike other forms of ABS, where the collateral pools usually consist of loans with broadly similar characteristics, CDO reference pools are typically quite heterogeneous, with exposures to a variety of borrower types and credit ratings and across a number of countries. A CDO will usually have exposures to between 50 and 200 bonds or large corporate loans, or up to 2,000 loans to small and medium-sized businesses.

CDOs are important instruments in the financial system since they facilitate the transfer of credit risk between financial market participants.2 CDO issuance has increased significantly in recent years, with more complex structures evolving in response to demands of investors and issuers. While the growth of CDOs has allowed credit risk to be spread across a broader range of financial market participants, the increasing complexity of some deals has at times made it difficult for issuers and investors to properly price risk.

CDO Structures

The simplest forms of CDOs are known as 'cash' or 'vanilla' CDOs, and are similar to other forms of ABS. A special purpose vehicle buys loans and securities from financial institutions and other market participants, and funds these acquisitions by selling securities to investors. The manager of the CDO vehicle will usually deduct fees and expenses from the interest income received from the assets in the collateral pool, with the remainder used to make regular coupon payments to investors. The term to maturity of the loans and bonds in the collateral pool will determine the maturity of the CDO securities sold to investors.

Like other forms of structured finance, the claims issued against the collateral pool are usually sold in tranches with differing degrees of credit support – that is, protection from losses should there be any defaults on the underlying loans or bonds. This is most commonly achieved by subordinating some of the tranches, whereby the most senior tranche has first legal claim on the CDO assets, with the priority of claims decreasing down to the most junior tranche (which is typically an unrated 'equity' tranche that is frequently retained by the issuer). An alternative form of credit support is to assign different priorities amongst investors' claims on the principal repayments received by the CDO over its lifetime. In this arrangement, as the value of the CDO's collateral pool reduces as the underlying debts mature, these debt repayments are used to pay back the senior tranche investors' principal. Only when all of the senior tranche has been fully retired will repayments on less senior tranches begin.

The size of each tranche relative to the value of the collateral pool will determine the degree of protection given to more senior tranche holders. An example of a cash CDO vehicle's balance sheet is shown in Figure 1. In this example, the size of the equity tranche is equivalent to 3 per cent of the CDO vehicle's assets, while the size of the next most junior tranche – the BBB-rated tranche – is equivalent to 5 per cent of assets. This means that the BBB-rated tranche is protected against the first 3 per cent of losses in the asset portfolio, but bears the full risk of the next 5 per cent of losses. If defaults by borrowers amount to 4 per cent of the collateral pool, the equity tranche will be wiped out and holders of the BBB-rated tranche will absorb the remaining 1 per cent of losses – that is, 20 per cent of their investment will be lost.

As one would expect, the credit rating assigned to a particular tranche by rating agencies can be increased (decreased) by increasing (decreasing) the amount of credit support provided by more junior tranches. Underlying the ratings given to tranches, however, is the rating agency's assessment of the likelihood of default of individual securities in the collateral pool. If these securities, on average, have a relatively low credit rating, then this will lower the weighted-average credit rating of the tranches. Credit ratings are also very sensitive to the estimated default correlation on the underlying claims that comprise the collateral pool – higher default correlation will weigh down the overall credit rating. Estimating the extent of this correlation is difficult, and rating agencies use methodologies that rely on assumptions about correlations of defaults within and across industry sectors, as well as information on specific company-to-company exposures. Where a collateral pool has a higher estimated default correlation or lower average credit quality, highly rated tranches can still be issued, but these will require greater levels of credit support than if the collateral pool had a lower default correlation or higher average credit quality.

Comparing the credit rating of a CDO tranche with the rating of a conventional bond is complicated by the fact that the respective investors' loss burdens will not have the same profile. Some rating agencies have dealt with this problem by defining equal ratings to mean the same ability to make full payments of interest and principal – that is, the rating is determined by the likelihood of whether the security will bear any loss, known as the 'probability of first dollar of loss'. But, as noted above, once a tranche incurs its first dollar of loss, it bears the burden of all further losses until its value is wiped out. In this sense, a tranche of a CDO that is rated, say, BBB is more risky than a BBB-rated conventional security.

While cash CDOs comprised the bulk of issuance during the early part of the CDO market's development, issuers are increasingly making use of credit derivatives to create 'synthetic' CDOs, also known as 'credit linked notes'. Rather than directly holding a pool of corporate debt as collateral, an equivalent credit risk exposure is created by entering into credit default swaps (CDS). Typically, the CDO vehicle invests the funds it has raised in bank deposits or highly rated securities. Additional return, and risk, is then generated by the vehicle entering into a series of CDS contracts whereby it receives 'insurance' premia from counterparties in return for agreeing to pay compensation in the event of a default (or some other credit event) by the specified corporate borrowers, or other 'reference entity'. The premia will be larger where the reference entity is considered to have greater credit risk. In the normal course of events, income from the return on the relatively safe investment plus the CDS premia is used to make interest payments to CDO investors. In the event of a corporate default, the CDO vehicle uses part or all of its funds (at the investors' expense) to compensate its swap counterparty.

A key advantage of synthetic structures over cash structures is that it is often faster and cheaper to assemble a portfolio of CDS for a particular reference pool of borrowers than to purchase the equivalent portfolio of bonds or loans. CDS contracts can also be tailored to the desired timing and currency denomination of cashflows. An additional advantage of synthetic structures is that issuers need not sell CDO tranches for the full amount of the underlying credit exposure. Since no outright purchases of assets have been undertaken, funds will only need to be raised to the extent that there is a need to fund provisions against the notional exposure agreed to in the CDS contracts. In contrast, since an issuer of a cash CDO has made outright purchases of bonds and loans, it must raise funds to the full amount of the collateral pool.

More recently, some issuers have offered 'CDO-squareds', which are CDOs that have reference pools consisting of tranches from other CDOs. Default risk on a given tranche of a CDO-squared depends on the seniority of the CDO-squared tranche, and the seniority of the CDO tranches included in the reference pool. It also depends on the level and correlation of defaults within and between the underlying CDOs' reference pools. Estimating these correlations can be even more difficult than for other CDOs, and it is possible that issuers of CDO-squareds may therefore underestimate the risks they continue to bear, with some of their offerings consequently being underpriced. The Appendix gives more information on these and other types of CDOs.

Trends in Issuance

Issuance of CDOs has increased rapidly both globally and in Australia over the past few years. Globally, US$160 billion of CDO tranches were issued in 2004, up from an annual average of less than US$100 billion between 1998 and 2002 (Graph 1).3 This strong growth was driven by increased issuance in the United States and Europe.

The Australian CDO market has been slower to develop than overseas markets. Between 1998 and 2002, total issuance of CDOs was only $2 billion, most of which was in the form of cash CDOs issued by domestic and non-resident banks to help manage their credit risk (Graph 2). Since then issuance of CDOs has been substantially higher, with around $4½ billion of CDOs issued over this period. At end August 2005, outstandings of publicly offered Australian dollar CDOs stood at around $5.7 billion.4

The majority of Australian CDOs have been issued at maturities of between 3 and 7 years. Credit exposures mostly consist of corporate debt, with bonds accounting for 55 per cent of all exposures and loans accounting for 25 per cent. The remaining 20 per cent of exposures consists of other CDOs or ABS. Although domestic banks were instrumental in initiating the Australian CDO market, accounting for 55 per cent of CDO issuance until 2001, more recently their market share has declined; since 2004 they have accounted for only 15 per cent of issuance, with overseas institutions accounting for the remainder.

Synthetic CDOs have accounted for the bulk of new offerings since the end of 2002. Part of this growth has been driven by the increasing recognition of the advantages of synthetic structures discussed above. As well, the relatively rapid growth of the domestic CDO market has made it increasingly difficult to assemble sufficiently distinctive pools from within the Australian corporate debt market. As a result of this, issuers have been using synthetic structures to access offshore credit exposures more readily, with Australian debt having come to comprise a relatively small share of domestic CDO exposures: over the past year and a half, it has accounted for only 10 per cent of newly issued CDOs' credit exposures (Graph 3).

CDO Investors

Purchasers of Australian CDOs include large fund managers, middle-market investors, and retail investors. Market liaison suggests that since 2002 around 20 per cent of new issuance has been taken up by large fund managers, which is a relatively small share in comparison to overseas markets. These managers are divided into two broad groups: high-yield bond funds, which typically buy the equity and lower-rated tranches; and standard bond funds, which buy the higher-rated tranches. These investors tend to view CDOs as just another type of corporate debt security.

Middle-market investors account for the largest share of the market, having purchased around 65 per cent of issues since 2002. This market segment consists of local governments, university and charity endowment funds, and high net worth individuals, as well as smaller boutique fund managers. Liaison with market participants indicates that within this segment local governments are quite prominent – partial data suggest that, in aggregate, CDOs comprise around 10 to 15 per cent of total local government financial assets, with some councils' holdings substantially higher.

Retail investors have also been an important source of demand for CDOs in Australia. Whereas early issues of CDOs were sold only to institutional investors, since 2002 retail investors have purchased around 15 per cent, by value, of newly issued CDOs. The retail share of the market in Australia is high compared to some overseas markets, perhaps reflecting the relative scarcity of high-yield money market funds in Australia compared with overseas managed fund markets. This perhaps also accounts for some of the strength in demand from middle-market investors.

Retail CDOs have been offered to investors through prospectus subscriptions, with most securities being listed on the Australian Stock Exchange (ASX). These prospectuses tend to emphasise that their reference portfolios contain a diversified selection of higher-rated and lower-rated names, the implication perhaps being that the higher-rated borrowers will balance out the lower-rated borrowers. In reality, the range of credit risk is at least as important as its average, since any defaults by lower-rated borrowers will reduce the credit support of each tranche, and potentially result in losses to investors.

Over the past few years, retail offerings have had a somewhat lower average credit rating than wholesale offerings – only one retail CDO offering since 2002 has been rated AAA, while more than half were rated BBB or lower. In contrast, over this period almost a third of the value of CDO offerings sold to middle-market investors and large fund managers were rated AAA, and only a tenth were rated BBB or lower (Graph 4). Recent retail offerings have, however, had higher credit ratings than earlier offerings.

In response to investor concerns following rating downgrades on earlier retail offerings, issuers of some more recent offerings have put a cap on the losses that will be borne by investors due to any one borrower defaulting. As well, some issues have included a degree of capital protection, which has ranged from an explicit capital guarantee to a more general aim of capital stability. An issuer can provide a capital guarantee by purchasing some form of insurance from a highly rated financial institution, although investors will still ultimately be exposed to the credit risk of that institution. Alternatively, capital stability can be generated by structuring the CDO as a 'combo note', where most of the principal is invested in a highly rated tranche, with the remainder invested in a riskier tranche (often the first-loss tranche). The investment in the highly rated tranche accumulates interest over the life of the CDO, so that at maturity the full principal amount should be able to be repaid, while income from the investment in the lower-rated tranche is used to make coupon payments. Combo notes are usually described as having the same rating as that given to their principal tranche, even though the stream of coupon payments is subject to higher risk.

CDO Ratings Performance

Given the increasing international content of domestic CDO reference pools, it is not surprising that the ratings performance of Australian CDOs has largely followed that of the global CDO market (Graph 5). Downgrades outnumbered upgrades in late 2003 and early 2004, but this trend was reversed in late 2004 with falling default rates and generally strong corporate prospects.

These offsetting movements have resulted in little net change in the credit ratings of outstanding Australian CDOs. Ninety per cent of both institutional and retail offerings are currently rated in the same major category as when they were first issued. Of the remaining 10 per cent, roughly equal numbers are currently rated higher or lower than their original ratings.

The increasingly international nature of domestic CDO reference portfolios also means that Australian investors have greater exposures to corporate downgrades and defaults that occur in overseas markets. The default of the Italian company Parmalat in late 2003 contributed to the rating downgrades of many CDOs in Australia, as well as overseas, around this period. More recently, the rating downgrades earlier this year of the US companies General Motors and Ford – two of the largest corporate borrowers in the world – had some ramifications for CDO equity tranches, though other tranches were largely unchanged.

These latter events also highlighted the difficulties of modelling correlation behaviour within and between CDO pools, with incorrect correlation assumptions resulting in a brief period of high volatility in global credit markets in May following the initial rating downgrades of these companies by Standard & Poor's and Fitch (Moody's announced downgrades in August). Prior to these announcements, some market participants had assumed that any downgrades of these companies would be positively correlated with downgrades of other corporate borrowers, thereby affecting the price of both equity tranches and the next most junior tranche. This expectation formed the basis of trading strategies implemented by a number of hedge funds and investment banks. In the event, however, the downgrades of General Motors and Ford were not accompanied by a rash of downgrades across the corporate sector, with the consequence that only equity tranches of CDOs were affected. As a result, market participants that had positioned themselves to profit from expected co-movements instead sustained losses, though many other CDO holders were little affected by events.

CDO Primary and Secondary Market Pricing

The yields at which CDOs have been issued in the Australian market have, for the most part, been higher than for similarly rated securities. Over the past year AA-rated CDO tranches have been issued at spreads of between 100 and 250 basis points above swap rates (Graph 6). In contrast, AA-rated tranches of residential mortgage-backed securities (RMBS) have been issued at around 50 basis points over swap rates in the past year, though these tranches have additional credit enhancement by being backed by lenders mortgage insurance which bears the first losses should household borrowers default.5 For BBB-rated securities, spreads have also been wider for CDO tranches than for other ABS tranches, though the pricing of these is much more variable, due in part to the greater variability of their asset pools. Spreads on conventional corporate bonds have been much narrower than for structured finance tranches for both AA-rated and BBB-rated securities.

Secondary market trading of CDOs is much less developed in Australia than in overseas markets. At present there are very few market-makers for these securities, though a number of institutions are prepared to transact on a best-endeavours basis. There is some price transparency for retail CDOs that are listed on the ASX, although trading volumes are usually quite low. Prices for individual issues have varied over the past year but, in aggregate, CDO spreads have declined in the secondary market, in line with the broader fixed income market.

Some of the pricing differential between CDOs and conventional securities would be expected, given the differences in the structural characteristics of these securities. As discussed earlier, credit ratings tend to indicate the likelihood of the first dollar of loss, rather than the expected total loss on the investment. The wider spreads are also partly explained by the illiquid secondary market for CDOs, with investors requiring a premium to hold securities that could be difficult to sell at a later date.

Conclusions

In general, the growth in CDO issuance in Australia is supportive of financial stability to the extent that it has allowed credit risk to be spread across a range of investors, rather than concentrated on the balance sheets of a small number of domestic financial institutions. With total outstandings of $5.7 billion, the Australian CDO market is currently not large enough to be of systemic importance to the financial sector. However, the available evidence suggests that CDOs constitute a reasonable proportion of some investors' financial assets, and the increased issuance of CDOs does raise a number of issues. Most notably, some investors, in seeking higher returns in a low-interest rate environment, may be underestimating the risks of these securities.

One issue for Australian investors, and the financial system more generally, is that the proportion of Australian debt in the reference pools of domestically issued CDOs has fallen to quite low levels over the past few years. This has meant that the major drivers of the risk characteristics of CDOs held by domestic investors may not be credit events related to the Australian financial system. While it is necessary for non-resident borrowers to be included in domestically issued CDOs to ensure that the reference pools are well diversified, to the extent that overseas CDOs do not incorporate an offsetting amount of domestic corporate debt, the Australian financial system has been a net recipient of global credit risk.

Another issue for Australia is that retail investors have tended to buy lower-rated CDO tranches than have their institutional peers, potentially leaving them more exposed to losses if the global economy were to suffer a period of economic stress. Also, the growing complexity of CDO structures has increased the difficulty of calculating risk based on the characteristics of the collateral pool, with some evidence of difficulties in pricing as a result.


Appendix – Types of CDOs

CLO – Collateralised Loan Obligation

A CDO in which most of the reference pool is comprised of corporate loans.

CBO – Collateralised Bond Obligation

A CDO in which most of the reference pool is comprised of corporate bonds.

CDO of ABS

A CDO in which most of the reference pool is comprised of tranches of asset-backed securities.

CDO-squared

A CDO in which most of the reference pool is comprised of tranches of other CDOs, which are usually synthetic. The figure below shows a CDO-squared transaction that has been divided into six tranches, and is referencing a portfolio of equal holdings of 10 tranches of other CDOs (each of which is exposed to losses from 5 per cent up to 10 per cent of the assets of their respective CDO pools). In this example, none of the CDO-squared's tranches will incur losses even if each of the 10 underlying CDOs incurs losses of 5 per cent. But if each of the underlying CDOs incurs another 1 per cent of losses, this will see one-fifth of the CDO-squared's portfolio lost, which will wipe out the equity tranche and all of the rated tranches except for 20 per cent of the AAA-rated tranche. More generally, the rated tranches of CDO-squared transactions start incurring losses later than the CDOs that comprise their reference portfolio, since the CDO-squared's own equity tranche gives even the lowest-rated tranche some protection. However, further defaults in the underlying pool can wipe out tranches of the CDO-squared much more rapidly than for standard CDOs.


Footnotes
  1. This article was prepared by the Securities Markets Section of Domestic Markets Department.
  2. For more discussion of credit transfer markets, see Hall, K and E Stuart (2003), 'Credit Risk Transfer Markets: An Australian Perspective', Reserve Bank of Australia Bulletin, May.
  3. Global and Australian issuance figures refer to funded CDO tranches, and exclude the unfunded tranches of synthetic CDOs.
  4. These figures exclude private deals, such as bilateral transactions of tailored CDO-like securities that have been undertaken by large institutions for the purposes of diversifying the credit exposures of their balance sheets.
  5. For more information on the influences of lenders mortgage insurance and other factors on the credit quality and pricing of RMBS tranches, see Bailey, K, M Davies and L Dixon Smith (2004), 'Asset Securitisation in Australia', Financial Stability Review, September.