The Credit Risk Sling
For a commercial bank, the risks inherent in its core business are the traditional ones of credit risk, market risk and funding risk. All three of these risks are contained within its loan book.
In the normal course of business, a bank will seek to manage the risk exposures inherent in its loan book through a combination of risk management techniques. This article considers how credit risk can be stripped out of the combined group of risks and managed in its own right. This is achieved through the use of credit derivatives, which enable credit risk to be traded as an asset class in its own right. By combining the use of credit derivatives with securitisation techniques, a bank can manage its credit risk, as well as its regulatory capital costs. This explains the rise in popularity of the static synthetic balance sheet collateralised debt obligation (CDO), which is the heart of this article.
The reasons that banks originate CDOs are two-fold:
A synthetic securitisation structure is engineered so that the credit risk of a pool of assets held on the originator’s own balance sheet is transferred from itself to investors by means of credit derivative instruments. The originator is in effect buying credit protection from investors who are the credit protection sellers. This credit risk transfer may be undertaken either directly or via an SPV. Using this approach, underlying or reference assets are not necessarily moved off the originator’s balance sheet. This makes the vehicle an ideal means by which to manage credit risk. Because the synthetic structure enables removal of credit exposure without asset transfer, commercial banks can use it for risk management and regulatory capital relief purposes. For banking institutions, it also enables loan risk to be transferred without selling the loans themselves, thereby allowing customer relationships to remain unaffected.
Now I would like to discuss the value of the static synthetic balance sheet CDO. To start, a synthetic CDO can be seen as being constructed out of the following:
The economic advantage of issuing a synthetic versus a cashflow CDO can be significant. Put simply, the net benefit to the originator is the gain in regulatory capital cost minus the cost of paying for credit protection on the credit default swap side. In a partially funded structure, a sponsoring bank will obtain full capital relief when note proceeds are invested in zero per cent risk weighted collateral, such as Treasuries or gilts. The super senior swap portion will carry a 20 per cent risk weighting.
In fact, a moment’s thought should make clear to us that a synthetic deal would be cheaper: where credit default swaps are used, the sponsor pays a basis point fee, which for AAA security might be in the range 10-30 basis points, depending on the stage of the credit cycle.
In a cash structure where bonds are issued, the cost to the sponsor would be the benchmark yield plus the credit spread, which would be considerably higher compared to the default swap premium. This is illustrated in the example shown in Figure 1, where we assume certain spreads and premiums in comparing a partially funded synthetic deal with a cash deal. The assumptions are that the:
Figure 1: CDO cost structure – synthetic versus cash flow
A generic synthetic CDO structure is illustrated more thoroughly in Figure 2. In this generic structure, the credit risk of the reference assets is transferred to the issuer SPV and ultimately the investors, by means of the credit default swap and an issue of credit-linked notes. In the default swap arrangement, the risk transfer is undertaken in return for the swap premium, which is then paid to investors by the issuer. The note issue is invested in risk-free collateral rather than passed on to the originator in order to de-link the credit ratings of the notes from the credit rating of the originator. And if the collateral pool was not established, then a downgrade of the sponsor could result in a downgrade of the issued notes.
Investors in the notes expose themselves to the credit risk of the reference assets, and if there are no credit events they will earn returns at least the equal of the collateral assets and the default swap premium. If the notes are credit-linked, they will also earn excess returns based on the performance of the reference portfolio. If there are credit events, the issuer will deliver the assets to the swap counterparty and pay the nominal value of the assets to the originator out of the collateral pool. Credit default swaps are unfunded credit derivatives, while credit-linked notes (CLNs) are funded credit derivatives where the protection seller (the investors) fund the value of the reference assets up-front, and receive a reduced return on occurrence of a credit event.
As the super-senior piece in a synthetic CDO does not need to be funded, this provides the key advantage of the synthetic mechanism compared to a cashflow arbitrage CDO. During the first half of 2002, the yield spread for the AAA note piece averaged 45-50 basis points over Libor1, while the cost of the super-senior swap was around 10-12 basis points. This means that the CDO manager can reinvest in the collateral pool risk-free assets at Libor minus 5 basis points it is able to gain from a saving of 28-35 basis points on each nominal $100 of the structure that is not funded.
1 Averaged from the yield spread on 7 synthetic deals closed during Jan-Jun 2002, yield spread at issue, rates data from Bloomberg.
This is a considerable gain. If we assume that a synthetic CDO is 95 per cent unfunded and five per cent funded, this is equivalent to the reference assets trading at approximately 26-33 basis points cheaper in the market. There is also an improvement to the return on capital measure for the CDO manager. Since typically the manager retains the equity piece, if this is two per cent of the structure and the gain is 33 basis points, then the return on equity will be improved by (.36/.02) or 16.5 per cent.
Another benefit of structuring CDOs as synthetic deals is their potentially greater attraction for investors (protection sellers). Often, selling credit default swap protection on a particular reference credit generates a higher return than going long of the underlying cash bond. In general this is because the credit default swap price is greater than the asset swap price for the same name, for a number of reasons (Choudhry 2001). For instance, during 2001 the average spread of the synthetic price over the cash price was 15 basis points in the five-year maturity area for BBB-rated credits.2
2 Source: UBS Warburg, CDO Insight, 29 March 2002
The two main reasons why default swap spreads tend to be above cash spreads are:
Note, however, that the existence of ongoing counterparty risk for the seller of a default swap is a factor that suggests its price should be below the cash price.
Figure 2: Synthetic CDO structure
The key structural differences between a synthetic and conventional securitisation are the absence of a true sale of assets and the use of credit derivatives. Investors, therefore, must focus on different aspects of risk that the former instrument represents. Although it might be said that each securitisation – irrespective of it being cash or synthetic – is a unique transaction with its own characteristics, synthetic deals are very transaction-specific because they can be tailor made to meet very specific requirements. Such requirements can be with regard to reference asset type, currency, underlying cash flows, credit derivative instrument and so on.
Investor risk in a synthetic deal centres on the credit risk inherent in reference assets and the legal issues associated with definition of credit events. The first risk is closely associated with securitisation in general, but synthetic securitisation in particular. Remember that the essence of the transaction is credit risk transfer, and investors (protection sellers) desire exposure to the credit performance of reference assets. Thus, investors are taking on the credit risk of these assets, be they conventional bonds, ABS securities, loans or other assets. The primary measure of this risk is the credit rating of the assets, taken together with any credit enhancements, as well as their historical ratings performance.
The second risk is more problematic, and open to translation issues. In a number of deals, the sponsor of the transaction is also tasked with determining when a credit event has taken place; as the sponsor is also buying protection there is scope for conflict of interest here. The more critical concern, and one which has given rise to litigation in past cases, is what exactly constitutes a credit event. A lack of clear legal definition can lead to conflict when the protection buyer believes that a particular occurrence is indeed a credit event and therefore the trigger for a protection payout, but the protection seller disputes this. Generally, the broader the definition of ‘credit event’, the greater the risk there is of dispute. Thus, trigger events should be defined in the governing legal documentation as closely as possible.
Indeed, this is critical. Most descriptions of events defined as trigger events include those listed in the 1999 ISDA Credit Derivatives Definitions, but circumstances that fall short of a general default so that payouts can be enforced when the reference asset obligor is not in default. This means that the risk taken on by investors in synthetic deals is higher than that taken on in a conventional cash deal (Choudhry 2002). First off, it is important for investors to be aware that credit ratings for a bond issue will not reflect all the credit events defined by ISDA. In the end, this also means that the probability of loss for a synthetic note of a specific rating may be higher than for a conventional note of the same reference name.
What are the advantages of synthetic structures for originators? Balance sheet synthetic securitisation vehicles present certain advantages over traditional cashflow structures. These include:
For this reason they are increasingly preferred by commercial banking treasury and asset liability management (ALM) desks.
A balance sheet synthetic CDO is employed by banks that wish to manage credit risk and regulatory capital. In a balance sheet CDO, the SPV enters into a credit default swap agreement with the originator, with the specific collateral pool designated as the reference portfolio. The SPV receives the premium payable on the default swap, and thereby provides credit protection on the reference portfolio.
There are three types of CDOs within this structure. A fully synthetic CDO is a completely unfunded structure that uses credit default swaps to transfer the entire credit risk of the reference assets to investors who are protection sellers. In a partially funded CDO, only the highest credit risk segment of the portfolio is transferred.
The cash flow that would be needed to service the synthetic CDO overlying liability is received from the AAA-rated collateral that is purchased by the SPV with the proceeds of an overlying note issue. An originating bank obtains maximum regulatory capital relief by means of a partially funded structure, through a combination of the synthetic CDO and what is known as a super senior swap arrangement with an OECD banking counterparty. A super senior swap provides additional protection to that part of the portfolio, the senior segment that is already protected by the funded portion of the transaction. The sponsor may retain the super senior element or may sell it to a monoline insurance firm or credit default swap provider.
A fully funded CDO is a structure where the credit risk of the entire portfolio is transferred to the SPV via a credit default swap. In a fully funded (or just ‘funded’) synthetic CDO the issuer enters into the credit default swap with the SPV, which itself issues CLNs to the entire value of the assets on which the risk has been transferred. The proceeds from the notes are invested in risk-free government or agency debt, such as gilts, bunds or Pfandbriefe, or in senior unsecured bank debt. Should there be a default on one or more of the underlying assets, the required amount of the collateral is sold and the proceeds from the sale paid to the issuer to recompense for the losses. The premium paid on the credit default swap must be sufficiently high to ensure that it covers the difference in yield between that on the collateral and that on the notes issued by the SPV. Fully funded CDOs are relatively uncommon.
One of the advantages of the partially funded arrangement is that the issuer will pay a lower premium compared to a fully funded synthetic CDO. This is because it is not required to pay the difference between the yield on the collateral and the coupon on the note issue (the unfunded part of the transaction). The downside is that the issuer will receive a reduction in risk weighting for capital purposes to 20 per cent for the risk transferred via the super senior default swap.
The fully unfunded CDO uses only credit derivatives in its structure. The swaps are rated in a similar fashion to notes, and there is usually an ‘equity’ piece that is retained by the originator. The reference portfolio will again be commercial loans, usually 100 per cent risk-weighted, or other assets. The credit rating of the swap tranches is based on the rating of the reference assets, as well as other factors, such as the diversity of the assets and ratings performance correlation. As well as the equity tranche, there will be one or more junior tranches, one or more senior tranches and super-senior tranche. The senior tranches are sold on to AAA-rated banks as a portfolio credit default swap, while the junior tranche is usually sold to an OECD bank.
The credit default swaps are not single-name swaps, but are written on a class of debt. The advantage for the originator is that it can name the reference asset class to investors to investors without having to disclose the name of specific loans. Default swaps are usually cash-settled and not physically settled, so that the reference assets can be replaced with other assets if desired by the sponsor.
The case study below illustrates an innovative structure with a creative combination of securitisation technology and credit derivatives. Analysis of the vehicle shows clearly how a commercial bank can utilise the arrangement to effectively manage its credit risk exposure and optimise balance sheet capital, as well as provide attractive returns for investors. As the market in synthetic credit is frequently more liquid than the cash market for the same reference names, it is reasonable to expect more transactions of this type in the near future.
To illustrate the concept of the static balance sheet synthetic CDO and its application in credit risk management, JP Morgan’s Moorad Choudhry explains how the ALCO 1 structure works. Originated by the Development Bank of Singapore and closed in December 2001, it has led to more rated synthetic balance sheet deals in the Asia-Pacific region.
The ALCO 1 CDO is described as the first rated synthetic balance sheet CDO from a non-Japanese bank3. It is a S$2.8 billion structure sponsored and managed by the Development Bank of Singapore (DBS).
3 Source: Moody’s
The structure allows DBS to shift the credit risk on a S$2.8 billion reference portfolio of mainly Singapore corporate loans to a special purpose vehicle, ALCO 1, using credit default swaps. As a result, DBS can reduce the risk capital it has to hold on the reference loans, without physically moving the assets from its balance sheet. The structure is a S$2.45 billion super-senior tranche – unfunded credit default swap – with S$224 million notes issue and S$126 million first-loss piece retained by DBS.
The notes are issued in six classes, collateralised by Singapore government T-bills and a reserve bank account known as a ‘GIC’ account. There is also a currency and interest-rate swap structure in place for risk hedging, and a put option that covers purchase of assets by arranger if the deal terminates before the expected maturity date. The issuer enters into credit default swaps with specified list of counterparties. The default swap pool is static, but there is a substitution facility for up to 10 per cent of the portfolio. This means that under certain specified conditions up to 10 per cent of the reference loan portfolio may be replaced by loans from outside the vehicle. Other than this the reference portfolio is static.
Name | ALCO 1 Limited |
Originator | Development Bank of Singapore Ltd |
Arrangers | JPMorgan Chase Bank |
DBS Ltd | |
Trustee | Bank of New York |
Closing date | 15 December 2001 |
Maturity | March 2009 |
Portfolio | S$2.8 billion of credit default swaps |
Reference assets | 199 reference obligations (136 obligors) |
Portfolio Administrator | JPMorgan Chase Bank Institutional Trust Services |
Figure 3: ALCO 1 structure and tranching
The first rated synthetic balance sheet deal in Asia, the ALCO 1-type structures have subsequently been adopted by other commercial banks in the region. The principal innovation of the vehicle is the method by which the reference credits are selected. The choice of reference credits on which swaps are written must, as expected with a CDO, follow a number of criteria set by the ratings agency, including diversity score, rating factor, weighted average spread, geographical and industry concentration, among others.
How does it work? The issuer enters into a portfolio credit default swap with DBS as the CDS counterparty to provide credit protection against losses in reference portfolio. The credit default swaps are cash settled. In return for protection premium payments, after aggregate losses exceeding the S$126 million ‘threshold’ amount, the Issuer is obliged to make protection payments to DBS. The maximum obligation is the S$224m note proceeds value. In standard fashion associated with securitised notes, further losses above the threshold amount will be allocated to overlying Notes in their reverse order of seniority. The Note proceeds are then invested in collateral pool comprised initially of Singapore Treasury bills.
During the term of the transaction DBS, as the CDS counterparty, is permitted to remove any eliminated reference obligations that are fully paid, terminated early or otherwise no longer eligible. In addition, DBS has the option to remove up to 10 per cent of the initial aggregate amount of the reference portfolio and substitute new or existing reference names.
For this structure, credit events are defined specifically as 1) failure to pay and 2) bankruptcy. It is important to note how this differs from European market CDOs where the list of defined credit events is invariably longer, frequently including restructuring and credit rating downgrade.
The reference portfolio is an Asian corporate portfolio, but with small percentage of loans originated in Australia. The portfolio is concentrated in Singapore (80 per cent). The weighted average credit quality is Baa3/Ba1, with an average life of three years. The Moody’s diversity score is low (20), reflecting the concentration of loans in Singapore. There is a high industrial concentration, and the total portfolio at inception was 199 reference obligations amongst 136 reference entities (obligors).
By structuring the deal in this way, DBS obtains capital relief on the funded portion of the assets, but at lower cost and less administrative burden than a traditional cashflow securitisation and without having to have a true sale of the assets.