III. The information economics of securitisation.

The mechanics of bank-based loan securitisation lend themselves to models of information economics, as the sale of selected asset claims by issuers raises issues involving concepts of asymmetric information and decision-making under uncertainty. As much as the non-diversifiable idiosyncratic risk inflicts some degree of illiquidity on the reference portfolio of bank loans, the fundamental motivation of loan securitisation substantiates this notion. At bottom, the conduct of CLOs garners issuers with a range of options in improving the credit quality of their loans by means of incorporating structural and credit enhancement, such that investment grade debt securities can be issued to capital markets.

However, private information about the credit quality of loans restricts the scale of securitisation in view of the way information asymmetries adversely impact on the marketability of bank loans. Illiquidity fuels the most intuitive, though paradox, objection to an efficient securitisation of loans, notwithstanding the fact that the complete absence of asymmetries would render the securitisation of illiquid assets unprofitable, as it scuppers efforts to diversify bad risk across a selected asset pool. Loans are non-standardised, non-commoditised claims due to opaque nature of the lender-borrower relationship.

For illiquidity trims the market value of asset claims, the securitisation structure of a CLO could mute such adverse effect on the value of the reference portfolio. By extension, the securitisation increases the average value of the reference portfolio to a selling price beyond what would be deemed necessary to at least offset the management cost associated with a securitisation. Hence, the detrimental effect of illiquid assets on the bank balance sheet can be extenuated by virtue of securitisation structures. However, their efficiency-improving effect is conditioned on the "capitalisation" of the financial system of the respective jurisdiction, which arguably signals the importance of market transparency of borrower fundamentals in external finance (e.g. relationship lending, etc.). In general terms, the economic effects induced by information asymmetries and illiquidity of the securitised collateral portfolio will inevitably determine the security design of the CLO transaction. Generally speaking, market implications of private information, i.e. adverse selection and moral hazard, as well as trading costs, are the sources of illiquidity, which impose limits to the degree of securitisation of loans.

A. Adverse Selection

Since the presence of asymmetric information qualifies as an element of uncertainty, investors assume the existence of adverse selection to occur in the spirit of the lemons market a la Akerlof (1970). (24) Such beliefs are compounded in their effect on asset pricing of securitisation transactions by the attendant degree of private information associated with loans, amid the apprehensive stance of banks towards disclosing their credit risk assessment methodology applied in assessing the creditworthiness of debtors. Assuming unilateral information advantage by issuers induces ex ante moral hazard in the asset selection process (see also Exhibit 19 in section III.B), rational investors anticipate being misled by issuers of a securitisation transaction, who are sure to be in a better position to judge the true credit quality of the reference portfolio.

[ILLUSTRATION OMITTED]

Given some uncertainty about the true value of the credit quality of the underlying reference pool of loans, investors will expect adverse selection and merely offer a price (average market price) that is on average below the true market price of the reference portfolio under symmetric information. Thus, the estimated value of such private information imposes a lemons premium on the issuer, who could either retain the reference portfolio of loans or sell it by means of securitisation. Even though issuers seek to counteract this effect by bundling assets and then further tranching these bundles before they are sold in capital markets as debt securities, the degree of private information is sanctioned by investors. Conversely, the ability of the issuer to establish maximum transparency about asset quality bears out the discount investors would command as compensation in return for uncertainty about the true value of the reference portfolio.

CLO transactions cannot exhaustively guard investors against the danger of adverse selection arising from the illiquidity of bank loans. In cognisance of the agency cost of adverse selection issuers of CLOs could suppress the pecuniary charge associated with the lemons premium by soliciting a higher valuation of the reference portfolio. They retain a claim in the performance of the collateral (reference portfolio) as a sign of asset quality in order to overcome the information problem. Since adverse selection can only arise in relation to the downside risk of default risk, the tranching and the allocative mechanism of losses in the structure of a CLO transaction bears critical importance, as they signal the absorption of loan default risk within the transaction. However, only if these provisions help discriminate good from poor issues, market separation through increased transparency can come about. (25)

Generally, issuers would opt for a combination of both (i) the concentration of credit risk of the underlying reference portfolio in a structural enhancement (see Exhibit 10) and (ii) the tranching of the debt securities issued to investors. In the context of subparticipation, the so-called loss cascading mechanism ensures that small junior tranches find most of the default risk allotted to them, leaving hardly any credit risk to large senior tranches, which could be sold to investors without suffering from price discounting due to adverse selection.

[ILLUSTRATION OMITTED]

In order to achieve high ratings for the senior securities, the conduit must commit to obtaining credit enhancements, which insulate senior securities from the risk of fluctuating payment patterns and excessive default on the underlying loan pool. Credit enhancement is defined as a contractual provision (such as asset retention) to reduce default loss from the reference portfolio eventually borne by the investor. Rating agencies typically require credit enhancement to cover the difference of default probability between the securitised collateral and the desired structured rating of the securitisation transaction.

One possibility of credit enhancement, for instance, would be if the sponsoring bank of the CLO transaction retains the most junior tranche, which attracts the highest lemons premium from adverse selection, as first loss position (credit enhancement) to cover all expected default loss of the underlying reference portfolio, and possibly accepts further stakes in subsequent tranches of higher seniority (second loss position). In return for credit enhancement as well as the loan origination and servicing functions the sponsor of the transaction appropriates whatever return is to be had from the securitisation net of prior claims by issued debt securities. That is, the gain from securitisation lies in the residual spread between the yield from underlying loans and the interest and non-interest costs of the conduit, net of any losses on pool assets covered by credit enhancements.

Due to the inherently illiquid nature of the loan pool and the high risk associated with the most junior tranche as the first-loss piece ("equity note"), the marketability of such unrated credit enhancements is limited (Herrmann and Tierney 1999). However, so-called interest subparticipation allows issuers to trade credit enhancements. The mechanism of interest subparticipation has been devised by issuers to reduce the illiquidity of the first loss piece of securitisation transactions in order to ameliorate the marketability of the credit enhancement held as an equity tranche by the sponsor of the transaction. Payments out of available interest generated from the overall reference portfolio are partially used to offset first losses of noteholders of the first loss position. By doing so, the principal amount of the outstanding first loss piece is reduced through the amount of interest subparticipation, in an amount equal to the allocated realised losses. Even though the claim of first loss noteholders to the interest subparticipation is an unsecured claim against the issuer, the economic rationale behind this concept is regulatory capital relief, as no capital has to be held against interest income under the current regulator), standards. Since the first loss piece achieves the rating of the issuer, the placement of" credit enhancement under interest subparticipation is cost efficient. However, the capital efficiency derived from such an arrangement is associated with substantial institutional risk in view of potential future changes in the regulatory framework, which has hitherto not given clear guidance on the capital treatment of the concept of interest subparticipation in the provision of credit enhancement. The new proposal for a revision of the Basel Accord indicates the possibility that the fist loss position will most likely be subjected to a full deduction from capital in this thinly regulated area of structured finance. Given present regulatory uncertainty as to the future capital treatment of structural provisions, such credit enhancement and the interest subparticipation, it is worthwhile incorporating a regulatory call of the first loss piece, which allows for the possible restructuring and subsequent sale of the most junior tranche to capital market investors.

Nonetheless, retention of credit enhancement--as a sign of willingness to shoulder significant credit risk--could not only allow the sponsor to allay adverse effects of private information associated with asset illiquidity, provided that issuers are prepared to go the extra mile by slightly exceeding minimum requirements of essential credit enhancement by rating agencies. (26) By the same token, credit enhancement also furnishes investors with additional comfort that the issuing bank has proper incentives to maintain effective loan servicing in the sense of mitigating ex post moral hazard as a disciplining device.

In most transactions credit enhancement comes in a variety of forms, where two or more structural provision of internal and external credit support are combined. Although the asset-backed (ABS) market has increasingly resorted to new structural features, such as aforementioned credit derivatives and subordination for the longest time (List, 2001), (27) a great number of transactions still rely on third-party support in providing the payment of debt (see Exhibit 10 below). This obligation might be a letter of credit (LOC), a standby bond purchase agreement, an irrevocably revolving credit agreement, a well-kept agreement or a guarantee (Deutsche Bank Global Markets, 2001). In the European context, typically insurance companies, swap providers or liquidity providers are the sort of agents that tend to commit themselves to third-party obligations as credit enhancements. By definition, credit and structural support in the form of (bond) insurance or guarantees, represent one of the key features in the security design, which distinguish asset-backed debt securities from unsecured or plain vanilla bonds.

Nonetheless, since credit enhancement remains to be an issue of great uncertainty, traditional devices of credit support, such as letters of credit and cash collateral, have recently been substituted for subordination with the well-known issuers only. Despite the growing attention devoted to subordination, many smaller issuers used to be confined to monoline policy in the form of the aforementioned insolvency insurance. The beauty of hard insurance, though admittedly more costly to the issuer, feeds on the capacity to reduce possible downward risk emanating from the deterioration of the loan pool or servicer quality (third party effect). Recently, even non-investment grade issuers have begun to rely on subordination as a means of substantiating credit enhancements, in order to acquire the right to be reimbursed for credit losses in excess of the first loss position (credit enhancement). The broader application of soft insurance in asset securitisation confirms a growing preference for subordinating investors' claims on the reference portfolio over third party support mechanisms and establishes an alternative route towards credit support of securitisation transactions. However, the attractiveness of subordination has major implications on the assessment of the implied credit risk in a structured finance transaction in way that reconciles discrepancies between internal credit ratings and external ratings of the loan pool underlying a securitisation transaction. Thus, with issuers militating towards soft forms of credit support, structured ratings are expected to display higher degrees of volatility in the future.

In summary, we distinguish the following types of internal/external credit and liquidity support are possible in a security design of securitisation transactions to protect investors from a deterioration of the reference portfolio underlying the securitisation transaction (see Exhibit 10). (28) Internal credit/liquidity support: (i) senior/subordinated structure and over-collateralisation, (ii) reserve fund, (iii) yield spread (excess servicing), (iv) "turboing", and (v) "commingling". External credit/ liquidity support: (i) third-party and parental guarantee, (ii) bond insurance, (iii) letters of credit (LOC), (iv) bank facility, (iv) cash collateral account (CCA), and (v) collateral invested amount (CIA).

A senior/subordinated structure, a popular type of internal credit support, represents an over-collateralisation (29) of the transaction--funded by the proceeds received from subordinated tranches of issued debt securities--which covers all estimated credit losses incurred by the reference loan pool. As defaults drain the value of the reference portfolio the loss burden is not equally shared amongst tranches. Instead, the subordination scheme allocates some interest proceeds--which would otherwise be distributed to subordinated debt if no distinction were made between tranches in terms of seniority--to be earmarked as payments to senior debt. This payment settlement process to senior and subordinated noteholders as well as third parties requires that any payment to subordinated noteholders is made only unless such disbursement reduces any funds contractually assigned to other creditors, whose credit support the issuer relies upon, to the extent that nonpayment of these funds to creditors would jeopardise the issuer's solvency. This provision serves to prevent that payment to senior investors are prejudiced by any payments made to the subordinated investors. A subordination of payment claims as means of coverage for both credit losses and liquidity shortfalls tends to be more costly than the acquisition of liquidity protection from a third party, as subordinated debtholders must be offered a higher interest rate in return for greater exposure to credit risk. Any potential rights of the subordinated noteholders to the underlying asset pool and its attendant revenues--before and after any default of issued debt securities (notes) occurs--further guide the compensation for this excess risk. As opposed to transaction structures that feature credit support from a creditworthy third party a senior/subordinated structure is less susceptible to a rating downgrade.

In practice, subordination is characterised by a loss cascading mechanism that involves a senior (or "A") class of securities and one or more subordinated (B, C, etc.) classes that function as the protective layers for the "A tranche". If a loan in the pool defaults, any loss thus incurred is absorbed by the subordinated securities. The "A tranche" is unaffected unless losses exceed the amount of the subordinated tranches. The senior securities are the portion of the ABS issue that is typically rated "triple-A", while the lower quality (but presumably higher-yield) subordinated classes receive a lower rating or are not rated.

The use of a reserve fund is a popular alternative to a bank facility in senior/subordinated structures in order to finance timely payments on outstanding debt of the securitisation transaction. A reserve fund, a separate fund created by the issuer, guarantees such credit support as it reimburses the trust for losses up to the amount of the reserve (Giddy, 2002). It is often used in combination with other types of enhancement. This form of credit support draws its prime benefit from the permanent coverage of asset losses, as it is required to be sufficiently liquid (held on the issuer's bank account) to ensure its availability whenever necessary. Moreover, issuers forgo the cost of maintaining a bank facility and incurring interest on any drafts made. Nonetheless, notwithstanding these inherent benefits, the cost associated with its funding, such as bond proceeds or a loan whose accrued interest must be repaid with surplus funds held by the issuer, have to taken into account in benchmarking the reserve fired mechanism with a bank facility. Since the issuer cannot release the surplus unless the reserve fund is sustained at its contractually required size, the risk of a rating downgrade of an issue is mitigated.

The excess spread (30) from the difference between the cash proceeds generated from the debt issuance on the underlying assets and the repayments on the issuer's assets can also be employed as credit coverage and liquidity support. (31) In other words, excess spread represents the net amount of interest payments from the underlying assets after bondholders and expenses have been paid. Most commonly, monthly excess spread is used to cover current-period losses and may be paid into a reserve fund to boost credit enhancement (Giddy, 2002). In the case of so-called turboing, excess servicing is applied to outstanding tranches as principal. Any excess spread must cover financial shortfall arising from the combination of credit loss, in the worst-case scenario of both prepayments and termination rates on asset claims, and maximum payments to debtholders. Additionally, taxation of any excess spread further reduces the amount available to the issuer. Nonetheless, in some cases a portion of the excess spread might be trapped, i.e. it is stricter from being released by the issuer, as it stands to be available for future needs.

In cases where collections of interest and principal on assets are pooled in a general account of the servicer and commingled with its other funds (especially in cases of mortgage-backed securities) (32) before these payments are passed on to the issuer of the securitisation transaction (commingling), the risk of the servicer to retain such payments in cases of insolvency or bankruptcy poses a persistent contingency on appropriate credit coverage. Based on the legal opinion from the issuer's counsel as to whether the loss of funds would be temporary (liquidity stress) or permanent (credit loss), the availability of sufficient funds to cover credit losses has to be guaranteed. In the move to evade negative implications of commingling as regards credit coverage, any payments received from assets should be redirected to the issuer, such as the SPV. Hence, the amount of funds likely to be drawn into any bankruptcy or insolvency resolution process is minimised.

In addition to internal credit and liquidity supports, also external credit enhancement from a third party represents an alternative means of shielding investors from expected credit loss. Under a third-party or parental guarantee an external party (such as an insurance company, parent company of the servicer/issuer of the transaction, etc.) enters into a contractual commitment to reimburse the issuer for losses up to a predetermined notional amount. Such a guarantee agreement could also be extended to include the obligations of advancing principal and interest to investors in a trustee-like fashion (see Exhibit 1) and/or buy back defaulted loans (Giddy, 2002; The Bond Market Association, 1998).

[ILLUSTRATION OMITTED]

Bond insurance (through surety bonds) can serve as a vehicle of specialised third-party credit/liquidity support. It is provided by a rated monoline insurance companies (generated triple-A rated), which guarantees full payment of principal and interest to noteholders of the transaction, as it reimburses the issuer of the transaction for any losses incurred. Even though issuers are able to achieve an "AAA" rating for "insured" tranches, bond insurance is a credit enhancement much less prevalent as a means of credit support in securitisation transactions than subordination due to higher cost. The higher expense associated with this form of credit coverage stems not only from the cost of insurance but also from the requirement of the underlying reference portfolio to be drawn on a loan pool of a sufficient investment-grade rating level. In most cases the insurer provides guarantees only to securities already of at least investment-grade quality (that is, BBB/Baa or equivalent). Hence, the insurance-based credit/liquidity support disciplines issuers to carefully balance both the level of credit enhancement needed for a desired structured rating of a designated reference portfolio and their financial capacity to provide such enhancement if they so desire. So monoline insurance tends to require one or more levels of credit enhancement that will cover losses before the insurance policy (Giddy, 2002). Rating agencies quantify the risk posed to the bond insurer by determining the capital charge on the exposure of the reference pool. Only sufficient financial capacity to meet the financial exposure (claims paying ability) merits continuation of the bond insurer's (i.e. no rating downgrade due to the prospect of failure to maintain the claim-paying ability), whilst the "insured" receivables of a securitisation transaction bought by investors are rated equal to the rating-assessed claims-paying ability of the insurance company (typically triple-A), because the insurance company guarantees the timely payment of principle and interest on the outstanding securities of the transaction.

Letters of credit (LOCs) are the surety bond-equivalent in regards to non-insurance financial institutions are guarantors, where typically banks promise to cover any amount of losses up to the level of credit enhancement needed for a given portfolio quality of the underlying reference pool of assets.

Third-party guarantees, bond insurance and letters of credit as forms of external credit enhancement expose the security level rating of securitisation transactions to the claims paying ability of the institutions providing enhancement as we need to think of these provisions as pledges of cash in keeping with some guarantor obligations, devoid of actual cash transfer or other payments. Hence, the character of such external credit enhancements does not betray any hint of downgrade risk independent of the actual time-varying loan performance of the underlying reference portfolio.

A bank facility represents another possibility of external liquidity support for a securitisation transaction, as the issuer can draw and redraw on the facility as and when needed, with repayment of drawn amounts being made when sufficient Rinds are held by the issuer of the transaction. Continuity of a standing bank facility is only guaranteed if the rights of the facility provider to termination are limited to cases of issuer's bankruptcy, whereby the lender is prohibited from petitioning the issuer into bankruptcy given that any utilisation of the facility does not constitute an act of insolvency. However, under the provisions of a bank facility the issuer ought to be entitled to terminate the facility agreement if the lender's rating is downgraded or, if specially agreed, has been downgraded such that future drawing rights can no longer be guaranteed.

This impediment to third-party risk is obviated by a cash collateral account (CCA). In this case, the issuer borrows the required amount of first loss provision (credit enhancement) from a commercial bank only to purchase a corresponding amount of highest-rated short-term (one-month) commercial paper. Unlike in the case of third-party guarantees, CCA represents an actual deposit of cash rather than a pledge of cash only, and, thus, the downgrade risk of the securitisation transaction remains unaffected by a rating change of CCA providers.

Finally, the collateral investment amount (CIA) concludes this diverse group of possible forms of credit and liquidity support. The CIA, akin to a subordinated tranche of a transaction, is either purchased on a negotiated basis by a single third-party credit enhancer or securitised as a private placement and sold to several investors. By common consent the attendant benefits of the CIA lies in asset retention as a form of partaking in portfolio performance (without downgrade risk of guarantor uncertainty) through first loss provision as credit enhancement.

Since credit ratings in securitisation transactions reflect the likelihood of full and timely payment of principal and interest to debt holders and expenses of other third parties, rating agencies need to examine whether investors are sufficiently shielded from losses of the underlying reference portfolio and cash flow interruptions or outright defaults caused by delinquencies, defaults and any insolvency of the loan servicer. Mind you that the support of a transaction critically depends on the availability, preference, advantages, and costs to the issuer, as well as on the sophistication of the market. Assignment of a certain structured rating to a tranche primarily hinges on whether the rating agency confidently deems the issuer sufficiently fit to ensure full and timely debt service at a level commensurate to the respective default expectations on the debt (see Exhibit 11). Depending on the quality of any credit support provided by the issuer and the sponsor respectively, external structured ratings are assigned to the various tranches of the transaction. (33) The important role of rating agencies in the securitisation process and the determination of structured ratings is summarised in Exhibit 12.

Exhibit 11. Data requirements by rating agencies for loan
securitisiation

INFORMATION ABOUT THE ISSUER

Operational data of the issuer including financial
development, organizational structure recent
development

Market competition aria market snare

Credit and termination policy

* Overview of departments and business practices
* Experience of employees
* Insolvency procedures, depreciation

Portfolio management

Realistion of assets/claims (billing)

Illustration of the life-cycle of bank assets/claims
(including IT systems and infrastructure)

Depreciation policy

Overview of dilutions including description of the
causes for dilution

Cash management

Concentration of the portfolio (granularity) and
management of clusters

INFORMATION ABOUT THE REFERENCE
PORTFOLIO ("THE COLLATERAL")

Historical information of collateral performance on a
monthly basis (time horizon 3 years)

Asset/Claim stock, expected/scheduled payments,
actual payments, prepayments, default rate, recovery
rates/loss given default, new acquisitions/claims

Delinquencies, maturity and termination rates
Dilutions (reduction of return)

Structure of the collateral portfolio

* Distribution of nominal balances/claims, maturity
and weighted maturity, type of claim, regional/
industry concentration, sales of debtor

* Breakdown of major debtors and their proportional
share in the portfolio

Source: Deutsche Bank Global Markets (UK) Ltd. (2001)

Exhibit 12. Data requirements by rating agencies for loan
securitisation

Rating agencies analyse the structure and the underlying reference
portfolio of the securitisation transaction.

They help investors understand and manage structured credit risk
as they:

* reduce uncertainty and widening of investment horizons: rating
agencies promote informed investment and assist investors in the
acquiring expertise or knowledge base to analyse complex
transaction types

* set credit risk limits: e.g. benchmark for a pension fund

* determine credit risk-adjusted yields: ratings also used to assess
the yield the investor should demand to be compensated for expected
credit loss

Rationale: viable debt rating system contributes to a sound
functioning of capital markets as it encourages greater market
efficiency and increased liquidity through transparency.

Although the retention of some assets reduces the collateral base of the transaction, the efficiency increase through a mitigated adverse selection premium more than compensates for the opportunity cost of partial nonsecuritisation ex ceteris paribus. (34) The retention of a first loss piece as credit enhancement in the loan securitisation, however, poses regulatory problems. The concentration of risk in the lowest tranche of the transaction is a pro forma provision for estimated (scheduled) loan default, whose deviation from expectations is not covered in the transaction structure on economic and regulatory grounds. In spite of the deduction of the first loss position from the issuer's capital base, as required by the Capital Adequacy framework of the Basle Accord, the effects excess default still pose a liability on such conventional regulatory provisions.

The introduction of increased transparency qualifies as another way of dodging the consequences of adverse selection, if issuers impart more detailed information about collateral quality of the underlying loan pool on investors and supporting agents in the security design of securitisation transactions. At some threshold level of available information in the bid for fair asset pricing of the loan pool, however, the effect of marginal disclosure of information would be strictly negative, as the insurance effect of asymmetric information markets is gradually eroded. The securitisation market would be prone to collapsing. Issuers with high quality reference portfolios could forgo any bundling and structuring of loan claims and sell loans directly to the market through straightforward loan sale or completely retain their reference portfolio on the loan book. Increased transparency in the valuation of the collateral quality also connotes the transition from the conventional type of securitisation to a synthetic structure, which is only hypothetically backed by the assets in reference portfolio. We distinguish between traditional and synthetic transactions.

a. Traditional securitisation

Traditional securitisation involves the "legal or economic transfer of assets or obligations to a third party that issues asset-backed securities (ABS) [, which] are claims against specific asset pools" (Basic Committee, 2001). (35) In its second working paper on the treatment of asset-backed securities the Basle Committee on Banking Supervision (2002a) defines traditional securitisation as a structured finance transaction that "involves the economic transfer of assets and other exposures through pooling and repackaging by a special purpose entity (SPE) into securities which can be sold to investors. This may be accomplished by legally isolating the underlying exposures from the originating bank through subparticipation."

The conventional type of loan securitisation is always predicated on a clean break between the bank originating the assets and the securitisation transaction itself, i.e. it epitomises the legal and economic separation of the seller from the securitised assets via a true sale (novation, assignment, declaration of trust or subparticipation). Granting regulatory capital relief through the transfer of assets off the balance sheet in standard transactions represents the most fundamental regulatory issue for the originating bank of a securitisation transaction. According to the revised proposal of the Basle Committee (2001) regulatory capital relief by means of removing assets from the balance sheet for purposes of determining minimum capital requirements takes effect once the following minimum conditions are satisfied (36):

* the transferred assets have been legally isolated from the transferor; that is, the assets are put beyond the reach of the transferor and its creditors, even in bankruptcy or receivership. This must be supported by a legal opinion,

* the transferee is a qualifying special-purpose vehicle (SPV) and the holders of the beneficial interests in that entity have the right to pledge or exchange those interests, and

* the transferor does not maintain effective or indirect control over the transferred assets.

These conditions are essentially the equivalent to the provisions in IAS 39/FASB 140/FASB 125, and therefore, there is no new restriction or qualifying condition being put up by the regulators. Unless the three previously listed conditions are met, the Basle Committee proposes to retain the respective assets on the books of the originating bank for regulatory accounting purposes (RAP), even if the assets are removed from the books in compliance with GAAP.

b. Synthetic securitisation

In the wake managing regulatory and risk capital banks and financial services companies increasingly turn to what is frequently termed the newest wrinkle of securitisation and structured finance--the synthetic security (Meissmer, 2000). An increasing number of structured finance transactions are such compound products, which amalgamate properties of both asset-backed securitisation and credit derivatives (37) in one coherent structure.

According to the Basle Committee on Banking Supervision (2002a and 2002b) synthetic securitisation "generally involves the transfer of credit risk though the use of funded (e.g. credit-linked notes) or unfunded (e.g. credit default swaps) credit derivatives or guarantees that serve to hedge the credit risk to which the originator is exposed."

In defection from conventional forms of selling claims on a reference pool of assets, synthetics effectively sidestep the legal quagmires, mainly because most or all of the assets are never sold to capital market investors. Under this scheme of loan securitisation the originating bank merely transfers the inherent credit risk of the loan book by means of a credit default swap, in which the counterparty agrees upon specific contractual covenants to cover a predetermined amount of losses in the loan pool. A significant portion of the global $300 billion business of risk transfer comes from collateralised debt obligations (CDOs), whose prime sub-categories are forms of synthetic and traditional CLO structures (The Economist, 2002a). Apart from this credit derivative, also credit-linked notes, credit spread options and total return swaps are further financial instruments, which allow issuers to shift isolated credit risk to guarantors, thereby making the risks marketable while leaving the original lender-borrower relationship untouched (Burghardt, 2001), as the reference asset is the loan pool retained by the bank. In case a sale does not come about, many of the bankruptcy and other securities laws become moot.

As the credit risk of the loans is transferred to a special purpose vehicle (SPV) and from there on to the investors, the originating bank (the sponsor of the transaction) achieves regulatory capital relief through a transfer of credit risk the underlying loan portfolio, which would otherwise qualify for a minimum capital requirement to cover credit risk exposure. The SPV as securitisation conduit does not purchase the reference portfolio of securitised bank assets and, hence, forgoes financial outlay in raising funds for financing what is considered to fall outside the definition of an off-balance sheet transfer of assets. Usually a synthetic transaction is complemented by a third-party credit default swap agreement, which protects the sponsor against asset default of the reference portfolio in nature much akin to an insurance contract with exogenised damage claims. In synthetic structures credit derivatives, e.g. credit-linked notes (CLNs), credit spread options, credit default swaps and total return swaps, are used as vehicles to shift isolated credit risk to guarantors, thereby making credit risk marketable, while the original credit relationship between creditor and debtor remains unchanged (Burghardt 2001). Since bank assets are retained on the balance sheet, a synthetic CLO transaction does not constitute a credit de-linkage between the servicer of the loan pool and the issuer of the tranches offered to investors. At the same time, the credit default swap removes large portions of credit risk from the balance sheet of the sponsor and mitigates the minimum capital requirements for credit risk cover, albeit the issuer sells credit-linked notes in the capital market. So "synthetication" is a more fine-tuned approach, which only addresses the regulatory sensitive element of the loan book, namely credit risk.

Generally, synthetic securitisation amounts to on-balance sheet credit hedging by means of a transmission mechanism of payment claims on a portfolio of assets that defy conventional forms of securitisation. Among the reasons certainly are restrictive provisions that prevent transfer or assignment, i.e. the sale of loans to an issuer, the SPV, might compromise client relationships or restrictive contractual covenants on the transfer of the underlying loans. Concerns surrounding the retention of client relationships associated with certain bank loans lead issuers to include those assets in the reference pool of synthetic transactions, whose off-balance sheet treatment would impede the generation of future business. If credit-linked notes are issued the legal integrity of perfected security interest in the reference portfolio underlying these debt securities does not only give rise to the benefit of regulatory capital relief; as investors assume a synthetic and prioritised share in default loss. The "synthetication" of structured claims for CLOs also wins out over conventional, true sale securitisation transactions (traditional CLOs) in terms of efficiency gains from the legal and economic treatment of the reference portfolio. Possible heterogeneity of loan characteristics, which would otherwise entail legal obstacles, complicates the legal definition of a true sale and its effective completion. In synthetic transactions, however, the absence of an outright transfer of legal title to the loan pool purports to a reduction of structural risk and administrative cost of CLOs.

For loss of loan transfer to a bankruptcy-remote special purpose vehicle (SPV) the legal issues associated with the notification of obligors and the perfection of legal transfer are evaded altogether in establishing both bankruptcy remoteness (perfected security interest) and true sale properties, essential to conventional transactions. The ability of the sponsor to retain legal title in the framework of a synthetic securitisation particularly lends itself to loans that have been originated in different jurisdictions. Consequently, issuers avoid the cost of complex transfer arrangements of loans that do not lend themselves to a straightforward sale. As collateral assets of synthetic transactions are frequently unfunded, the popularity of synthetic structures as a carrier of regulatory capital mitigation is largely due to the favourable funding properties of large banks, which typically have access to on-balance sheet funds at competitive spreads especially in the area of mortgage-based financing and Pfandbrief issues.

Consequently, the "synthetication" of structured claims squares with both regulatory arbitrage and improved risk-adjusted returns, as the diversification effect of risk transfer by means of credit derivatives requires enhanced internal pricing methods of expected default loss (Rosch, 2001). Even if proposed regulator), changes to the standard credit risk weightings for bank loans (as foundation balance sheet restructuring effect of securitisation) renders the regulatory arbitrage aspect of securitisation obsolete, it constitutes no rebuttal to the benefits associated with loan securitisation per se, as efforts of boosting the economic rents from loan origination are not scuppered.

c. Distinguishing conventional and synthetic CLOs

Although both types of securitisation pursue broadly similar economic objectives in terms of balance sheet restructuring and increasing efficiency of banking operations, significantly different exposures to explicit and implicit risks warrant a careful distinction as to their effects on the structural make-up of the securitisation process and security design of CLOs.

The Basic Committee on Banking Supervision (2001 and 2002) addresses this aspect in the tentative regulatory treatment stipulated in Basle Consultative Paper on Securitisation (see section VII.A.1.i), (38) which discusses the two broad types of securitisation structures separately in two sections of its new proposal for revision and augmentation of the Basle Accord of 1988, and the Second Working Paper on the Treatment of Asset-Backed Securitisation. The schematic illustration of the contractual and financial relationships involved in the completion of a CLO transaction highlights the properties of loan securitisation, on the one hand, and aids understanding of the distinct features of conventional (see Exhibit 13) and synthetic transactions (see Exhibits 14 and 15), on the other hand

[ILLUSTRATIONS OMITTED]

In a conventional balance sheet CLO, the sponsor of the transaction is in charge of packaging (selection and structuring) the asset claims to be transferred to a bankruptcy-remote special purpose vehicle (SPV), which issues securities on the underlying reference portfolio of loans ("the collateral"). The securities are structured in credit tranches, where a prioritisation of claims and loss cascading guarantee that senior tranches carry a high investment-grade rating (triple-A or double-A rating), provided sufficient collateral quality and the sufficient availability of mezzanine and junior tranches in the CLO structure. These tranches are needed to shield more senior tranches from credit losses. The process of asset transfer to a SPV in a balance sheet CLO involves significant administrative effort in a loan-by-loan review to ensure compliance of each collateral asset with the stipulated eligibility criteria of the respective securitisation structure. Also the existence of contractual restrictions and special covenants prohibiting the transfer of ownership of the loan must be examined, whilst the continued servicing of the transferred assets by the sponsor of a balance sheet CLO does not attract major legal and administrative enquiry and verification. The latter feature of traditional loan securitisation is advantageous to both the sponsor, who receives earning fee income, and the creditor, as the client relationship is not compromised.

Some of the fees received by the sponsoring bank tend to be used to offset the cost of a commitment device in securitisation. As shown in theory, the originating bank (the sponsor) retains an equity claim as credit enhancement, whose nominal amount is directly deducted from its capital base for regulatory purposes. Credit enhancement represents the sponsor's willingness to mitigate the adverse selection effects of private information associated with inherent illiquidity of the reference (loan) portfolio. Investors can draw comfort from such a provision as it goes to show that the bank has installed proper incentives for effectively servicing the loan assets. Moreover, many transactions incorporate fixed-to-float interest rate swaps, which are used to hedge the interest rate risk of any fixed-rate loans such that credit risk remains the only investment risk (as described in the definition of cash flow CDOs).

Proper transmission of asset losses and the distribution of proceeds to investors by the issuer is supervised by the trustee, who task is particularly sensitive in times of premature determination of the transaction through early amortisation or excessive unexpected losses in the reference portfolio of the transaction. The trustee of the transaction, acting on behalf of the SPV, must also have the ability to hold perfected security interest for each loan asset. In balance sheet CLO structures this role is critical in compliance with regulatory statues governing the transfer of loan assets with reference to borrower confidentiality.

In contrast to conventional securitisation, synthetic securitisation represents a structured finance transaction where only credit default risk of a reference portfolio is transferred to a third party by means of credit derivatives without credit de-linkage between the servicer of the loan pool and the issuer of the tranches offered to investors in steps 2-4 as shown in Exhibit 15 below. Instead of the assets being sold by the originator, credit risk is transferred through a credit default swap (Exhibit 15, Step 2). So any resulting capital relief for mitigated risk exposure does not stem from the actual transfer of assets but the acquisition of credit protection from counterparties by means of credit derivatives.

Sellers of the credit protection receive a premium for their obligation of compensating buyers for any loss suffered on the assets underlying the credit derivative. This property, of synthetic CLOs is attractive to large banks, which tend to have access to on-balance sheet assets at competitive spreads.

Since a synthetic securitisation can be conducted with or without a SPV (see Exhibit 14 and 15 respectively), the general description above warrants refinement as to the specific mechanism governing the completion of synthetic CLOs with SPV. The direct issuance of credit-linked notes (CLNs) by the sponsor in a synthetic CLO transaction can alternatively be completed by an intermediating securitisation conduit, such as a special purpose vehicle (SPV). Provided that the sponsor of a synthetic transaction incorporates a SPV as the issuer ofa CLO, the latter has little or no need to raise funds, because it is not required to purchase the underlying loan pool. Similar to traditional schemes of securitisation the seller of a transaction transfers credit risk of a given asset portfolio through a specified conduit. The latter issues credit-linked notes to investors and retains the proceeds to invest in highly rated investment-grade securities as collateral for secured credit-linked notes (Exhibit 15, Step 3).

Exhibit 14. Structure of a synthetic collateralised loan obligation
(CLO) without special purpose vehicle (SPV)

[ILLUSTRATION OMITTED]

Step 1: The protection buyer/originating bank selects a reference loan
portfolio and structures expected interest and principle
repayment such that it can issue credit-linked notes (CLNS)
to investors in return for receipt of cash proceeds. On the
maturity of the notes, principle (net of allocated losses, if
any) will be repaid along with the redemption proceeds of the
collateral.

Step 2: The sponsoring bank transfers the risk in the "super-senior
tranche" to an OECD bank by means of a credit default swap
(CDS).

Step 3: The sponsor issues secured obligations as direct obligations of
the sponsor, whose structured claims are collateralised by
long-term risk-free government bonds.

Step 4: The sponsor issues unsecured obligations as direct obligations
of the sponsor, whose structured claims are not collateralised.

Step 5: The originating bank may also act as investor to the equity
note as first loss position (credit enhancement). This equity
claim is the first tranche to absorb credit losses before more
senior tranches are effected by unscheduled default in the
reference portfolio.

Step 6: The trustee oversees the assets of the SPV and protects the
interest of the noteholders and the supersenior counterparty.
The occurrence of credit default requires the trustee to
oversee the premature amortization of the transaction by
redeeming the outstanding CLNs through collateral sale. Aside
from the importance of its timing during the workout process,
a guaranteed minimum value can be generated from the selling
collateral to fund full note redemption. In the event of an
issuer downgrade, a put option allows for an at price for
government bonds (collateral plus accrued interest.

Exhibit 15. Structure of a synthetic collateralised loan obligation
(CLO) with special purpose vehicle (SPV)

[ILLUSTRATION OMITTED]

Step 1: The orginating bank (protection buyer) pays to the SPV a
premium in return for first loss protection on the reference
portfolio. The SPV uses the premium and the interest proceeds
from collateral to fund the spreads on the notes issued to
investors. If credit losses occur on the reference portfolio,
the SPV pays such amount to the protection buyer, subject to a
maximum payment equal to the sum total of the face values of
the notes (including the equity note).

Step 2: The SPV issues notes to investors and receives cash proceeds.
The originating bank may also act as an investor to the equity
note.

Step 3: The SPV purchases 0% risk weighted collateral ("collateral") to
collateralise the notes. On the maturity of the notes,
principal (net of allocated losses, if any) will be repaid
along with the redemption proceeds of the collateral.

Step 4: The risk of the "superior-senior piece" is transfered to an
OECD bank via a credit default swap or bank guarantee. The
originating bank will pay a default premium to the OECD bank.
If losses on the reference portfolio occur in excess of the sum
total of the face value of the notes issued by the SPV
(including the equity note), the OECD bank with compensate the
originating bank for such excess.

Step 5: The trustee oversees the assets of the SPV and protects the
interests of the noteholders and the "supersenior"
counterparty.

The SPV gains in the reallocation of investment funds generated from buyers of debt securities, collateralised by Pfandbriefe or similarly highly rates sovereign or corporate debt securities, and finances the additional spread for CLO notes by the swap premium paid by the sponsor (excluding an administrative charge). This collateralisation of claims ensures timely repayment of principal and interest to investors. In return, the SPV assumes a proportion of underlying collateral credit risk by entering into a credit default agreement with the sponsoring bank (Exhibit 15, Step 2 and 5), which remains the servicer of the underlying loan portfolio. The sponsor compensates the swap counterparty by paying a premium for the credit default swap. In the case of unexpected credit default of the underlying loan portfolio, the bank seeks recourse with the SPV as protection provider. If total accumulated losses incurred in credit events do not exceed scheduled losses of the reference portfolio, i.e. funds held by the SPV are exhausted by compensatory payments to the originating bank (protection buyer), capital market investors have a prioritised claim on both

* expected returns from investments financed by the proceeds generated from the debt securities issued by the SPV as well as

* the total premium of the credit default swap paid by the sponsoring bank for credit protection, minus some administrative charge levied by the SPV.

Despite of regulatory arbitrage becoming less likely in the view of an internal ratings-based approach to risk-weighted capital requirements under the revised proposal for a new capital adequacy framework, synthetic securitisation still steals a march from an economic perspective if we consider the balance sheet entries of both sponsor and originator of the securitisation transaction. The sponsoring bank substitutes the payment of a credit swap premium, the reduction of minimum capital requirements and a potential increase in risk-adjusted returns (due to greater asset base and higher diversification through re-composing the loan book) for the present level of either regulatory capital or economic capital (whichever one is higher). While the servicing function of the sponsor of the transaction remains unaffected, the generation of interest income from loans does not enter this trade-off consideration. The same applies to the cost of capital. Thus, the key benefit from synthetic securitisation does not tally with the main argument of securitisation--exclusive regulatory arbitrage as the only true benefit. As arbitrage fades, increased economic efficiency in reducing economic capital edges out as prime incentive.

1. Structural comparison of traditional and synthetic (with SPV) CLOs

The major differences between conventional (true sale) CLOs and synthetic CLOs can be illustrated in terms of various stakeholder issues and the security design of the loan securitisation process. Whereas traditional transaction is predicated on the selection of a loan portfolio to be transferred into a special purpose vehicle, synthetic CLOs do not promulgate restrained balance sheet growth by means of true sales of on-balance sheet assets. However, the credit-linkage between the issuer of the transaction and issued debt securities in synthetic CLOs means that information about the potential exposure to credit risk of the collateral (reference pool of loans) is acutely relevant in comparing synthetic and traditional CLOs.

The accuracy in assessing information about the actual quality of the reference portfolio ultimately affects the credit rating of the issued notes, barring any mitigating effect of structural enhancements, which could absorb deteriorating collateral quality. Needless to say, due to its heightened sensitivity to collateral performance, synthetic CLOs prerequisite a greater degree of information disclosure of collateral quality (to the risk transfer counterparty) by the sponsoring bank, which is recognised by rating agencies in a more painstaking and rigorous examination of the issuer's ability to meet their obligation of promised investor returns. While the collateral pool of synthetic CLO without SPV is held by the sponsoring bank, which issues credit-linked notes (CLNs), the incorporation of a SPV results in collateralisation of issued debt securities by whatever collateral asset the issuer is willing to choose. This arrangement leaves the credit-linkage with the sponsor intact. Exhibit 16 below shows the distribution of senior note collateral in European synthetic CLOs.

Collateralisation with government securities greatly limits the credit risk suffered by noteholders. The popular use of non-government securities (such as Pfandbriefe) as collateral must not be overlooked, for it might frustrate efforts to substantiate senior claims on the securitised reference portfolio. Although the latter form of collateralisation serves the triple-A credit rating to translate into a similarly high rating of senior CLO notes, (39) The credit volatility of non-government securities (which are not zero-risk weighted--for regulatory purposes--compared to government bonds from OECD member states) is higher than the exposure to unexpected loss in the case of risk-free sovereign debt. As its default probability cannot be insulated from the rating of the issuing entity, this exogenous collateral risk poses a significant structural challenge to synthetic CLOs. If the rating of non-sovereign security collateral fails to withstand issuer deterioration or an adverse change in the legal or regulatory framework pertinent to these instruments, the outstanding CLO notes might be subjected to a downward rating drift, with the sponsor's loan portfolio remaining unchanged in asset quality. Nonetheless, in keeping with this structural exposure to synthetic CLOs, the decrease of funds held by the originating SPV reigns supreme in assessing the credit risk of a CLO transaction.

As much as the development of collateral quality held by the SPV could fail to deflect structural exposure to the sponsor's rating, the sponsoring bank itself as a protection buyer can be a source of credit risk exposure in a synthetic CLO transaction. The credit risk exposure stemming from the defined role of the sponsoring bank might impede on the scheduled performance of synthetic transactions as to the initiation of early amortisation features (e.g. economic and structural triggers). Since a premature deposit by sponsoring bank--as the protection buyer in a credit default swap--can be contractually arranged, such that the risk of payment obligations does not add to the overall exposure of the respective synthetic CLO transaction (be it with or without a SPV) as, other functions of the sponsor might very well represent an exposure.

Whenever the collateral is credit-linked to the sponsoring bank, i.e. the "synthetication" forgoes the incorporation of a SPV as CLO originator, investors are directly exposed to the insolvency of the sponsor as a form of structural risk (Anonymous, 2001). However, not only a collateral pool (comprised of bonds) with material credit-linkage to the sponsoring bank establishes a perfected security, interest that exposes investors to the sponsor's credit rating (Batchvarov et al., 2000). In synthetic CLOs with SPV, even high investment-grade non-government bonds are sensitive to the credit performance of the sponsoring bank, despite being devoid of any credit-linkage therewith. Due to multiple reliance on the sponsor, issuers of such synthetic CLOs carefully observe this source of exposure by taking an appropriate hedging position. In the state of seller deterioration or insolvency, the cost of such a provision might eventually compromise the originator's ability to pay down outstanding notes issued on the underlying collateral, provided that the collateral has some relationship with the sponsor. If this was to occur, the apparent credit dependence of the issuer on the sponsoring bank could render impossible the market sale or marked-to-market payment of notes outstanding. Hence, the physical delivery of collateral in lieu of redemption is warranted.

In addition to both payment of credit protection and credit profile, the role of a repurchase transaction (repo) (40) or hedging counterparty represents another area of potential exposure, resulting from the role of the sponsoring bank within the CLO structure. The deterioration of the sponsor's credit quality in a marked-to-market hedging arrangement adversely affects issued CLO debt securities, unless structural provisions have been adopted, e.g. a put option on sponsor downgrade or declining collateral balance. Whereas the rating on outstanding notes is likely to remain unaffected (provided the collateral value remains unchanged), failure to do implement remedial action in improving the sponsor's credit standing induces an early amortisation of collateral through early liquidation.

As sponsors of synthetic CLOs forgo transferring the loan servicing function to another party in the bid for capital relief, the need of protecting noteholder interest by trustees appears less pressing compared to traditional loan securitisation. In a conventional transaction, the search for a proficient substitute servicer--upon the originator declaring insolvency--leads to heightened structural exposure. This limited role of the trustee in shielding investors from bankruptcy risk of the CLO originator does not extend to other aspects of synthetic CLOs. The continuity of loan servicing in synthetic deals does not rule out possible incentive incompatibility between originators and investors of CLOs in terms of

* the monitoring and sale of the reference portfolio/collateral,

* the determination of amortisation triggers and their initiation threshold, as well as

* the verification of credit default loss and its prioritised allocation to investors. Thus, synthetic CLOs heavily rely on the prudential vigilance of trustees, whose conduct in monitoring both sponsors and issuers of CLOs is vital in ascertaining their adherence to binding contractual and regulatory terms and conditions set forth in CLO transactions.

The degree of leverage in the security design denotes the funding level, i.e. the proportion of the reference portfolio that is not subject to the structural claim of the CLO transaction. As the sponsor relinquishes loan servicing through asset transfer (true sale), traditional cash flow transactions are always fully funded, i.e. the value of issued collateralised notes tallies with the underlying reference portfolio. Synthetic transactions hardly are. They do not even need to be necessarily partially funded; however, if they are, the presence of partial funding adds to the existent leverage of senior noteholders from the prioritisation of losses in subparticipation. Hence, any increase in the leverage entails greater relative losses (as a proportion of the reference pool) for senior tranches of the CLO transaction compared to fully funded CLO structures, provided that accumulated credit losses have depleted the first loss provision of the subordinated tranches. Although senior tranches might gain from diversification effects as their expected returns are leveraged on a larger underlying pool of loans, by the same token, they are exposed to higher potential default loss (loss severity) from a larger collection of loans.

The marginal difference in senior risk exposure between partially funded synthetic securitisation and traditional securitisation does not extent to junior noteholders with subordinated security interest. While partial funding structures bear more risk emerging from the sponsor's role, the credit enhancement (first loss provision) and subsequent junior tranches (the second loss position) are no more exposed to credit risk in synthetic deals than they are in traditional CLOs.

The above analysis cumulates in a synthesis of benchmark characteristics (see Exhibit 17) pertinent to an ideal synthetic loan securitisation (in contrast to traditional securitisation), based on areas of distinctive deviation from traditional CLOs, such as the role of the sponsoring bank, the credit dependence of the collateral, the structural provisions for cash transfer as well as the interpretation of credit events and subsequent administration of transaction workout involving earl), amortisation triggers and the redemption of collateralised notes issued. The formulation of an optimal combination of these characteristics caters to a proper assessment of the complex security design of CLOs for structural and pricing purposes. Some of the benefits and drawbacks associated with both structures of CLOs are summarised below (see Exhibit 18). Given the varying approaches taken by rating agencies in deriving structured ratings, a comparative perspective on the basis of general benchmark criteria has been strongly encouraged by major CLO issuers in building investor confidence in "synthetication".

Exhibit 18. Benefits and drawbacks of structural alternatives
between synthetic and traditional securitisation.

Benefits

Conventional

* tried and tested structure, well familiar to both rating
agencies and investors

* allows for relief of both regulatory and economic capital

* "True sale" transfer of assets allows for off-balance sheet
accounting treatment of reference portfolio

* tighter trading spreads as opposed to synthetic structures
due to mature market (investors, rating agencies, etc.)

Synthetic

* efficient transfer of isolated credit risk, esp. of
non-transferable assets and less restrictive than "true sale"
requiremetns in traditional transactions

* reversibility of transfers to the reference portfolio

* possibility of combining synthetic structures with other
transactions to create compound structures (e.g. Pfandbrief,
etc.)

* allows for evolution in the loan portfolio without incurring
the cost of re-structuring (as opposed to conventional deals)

* structural simplicity and greater transparency of asset
terms of eligibility for securitisation

* shorter implementation timetable

* also unfunded and/or partially funded structures are possible

Drawbacks

Conventional

* high demands on reporting capability of loan portfolio
information systems

* eligibility of asset transfer depends on loan terms and
conditions and jurisdictional constraints

* in some jurisdiction the possibility of continued servicing
requires the sponsor to seek permission by the issuer of the
transaction

* moderately longer implementation time-table than synthetic
structures

* higher administrative cost than synthetic structures

* only funded obligations are issued

Synthetic

* allows primarily for economic capital relief only

* "leveraged" structures, as notes issued generally amount to no
more than 10% of the reference portfolio

* issue of defining credit event and work-out procedures in
cases of default

* less "informed buyers" in the market leads to historically
wider spreadds in comparison to tradition transactions

In conventional cash flow CLOs the solid quality of the reference pool, structural provisions for possible credit dependence of the collateral, the various roles of the sponsoring bank, the determination of cash and asset transfer as well as the function of credit enhancements establish the prime requirements flowing into its credit assessment: (i) stable and projectable cash flows, (ii) availability of historic information about the collateral portfolio (delinquencies, terminations, defaults, dilutions, etc.), (iii) homogenous, broadly diversified portfolio, (iv) no legal obstacles in transfer of ownership (assignment, novelation, subparticipation, etc.), and (v) minimum portfolio size of 50m [euro] ("step-up" possible).

According to the European Securitisation Group of Deutsche Bank (42) a benchmark collateral portfolio of securitised loans would roughly match the following characteristics:

--reference portfolio: mainly loans with variable interest rate and credit derivatives

--rating: B and better (on average BB and better)

--average maturity: 5-8 years

--number of assets: 150-500

--portfolio diversification: according to industry sectors and location--Moody's diversity score (30-70)

--average loan volume: 600,000 [euro]-15m [euro]

--AAA (tranche) enhancement: 10%-15%

--bond spreads: LIBOR +25 bps (AAA) to +250 bps (BB)

--estimated default frequency (EDF) p.a.: 0.05%-0.25%

This array of properties does, however, fall short of addressing the special nature of synthetic transactions. Rating agencies and investors general require additional criteria to be met for a synthetic loan securitisation to come together.

* in synthetic CLOs with SPV the collateral of synthetic CLOs is comprised of sovereign debt securities, i.e. government bonds or capital market paper with similar credit risk weighting, whose credit volatility is lower than collateral exposure from collateralisation with non-government bonds. Independence of collateral performance from the sponsor's credit profile ensures its insulation from structural exposure that would otherwise compromise its zero-risk weighting. The secondary market for the chosen collateral (sovereign debt securities) needs to be sufficiently liquid, so that the adverse price impact of collateral sale for purposes of efficient notes redemption can be fully absorbed (without distorting market prices);

* any structural provisions that involve the participation of the sponsoring bank in terms of advance payment facilities/liquidity support or revolving credit agreements are kept at a minimum and/or are curtailed by safeguarding mechanism of bankruptcy remoteness, such as advance funding of protection payment obligation or letters of credit/guarantees of proficient counterparties;

* the integration of a first loss position retained by the servicer of the reference portfolio underlying a synthetic CLO serves as an incentive compatibility device in maximising recoveries of defaulted collateral. This template requires credit enhancement in the form of an issued equity claim on the reference portfolio, sufficiently large to cover almost all credit risk before senior tranches suffer from expected loss and subparticipation of interest proceeds from collateral to junior noteholders is warranted. The settlement of noteholders' claims in the case of termination ought to be conducted at the end of a well-defined workout process, whose duration being as it is consistent with maximum collateral recovery rather than a stipulated time horizon after default. By common consent, the recovery rate under enforcement or workout process is held to generate higher prices for distressed assets than soon aster default; (43)

* the definition and validation of credit events of the reference portfolio is vital in the estimation of expected losses and possible recovery of bad debt in synthetic CLO structures. Only an agreed threshold on portfolio losses--mapped to a certain workout process after early amortisation has been triggered--maximises investor returns in case of distressed collateral;

* the degree of leverage in the funding structure should not be in conflict with the default tolerance of assigned (senior) tranche ratings. Since synthetic transactions are hardly fully funded, i.e. the notional value of issues notes does not square with the valuation of the underlying reference portfolio of bank loans, the implicit marginal increase of credit risk premium on senior tranches merits structural compensation (additional credit support, etc.); and

* the leverage of synthetic deals coupled with the retention of loan servicing by the sponsor requires the exclusion of accrued interest on the defaulted portion of the reference portfolio due to the synthetic interest claim of investors. In traditional CLOs the extent to which notes are redeemed is solely contingent on the recovery rate as interest payments on defaulted loans cease. In contrast, the structured claim of investors in synthetic deals allows for the allocation of losses at deal termination. In this way, the full note payment to investors is not compromised unless excessive collateral losses are charged up against proceeds from interest and principal.

2. Differences and similarities between ABS and CLOs

After having considered the differences between conventional and synthetic sccuritisation it is also worthwhile to compare asset-backed securitisation and traditional (balance sheet) CLOs (Deutsche Bank Global Markets, 2000). Generally, they display the several common structural properties. First, assets from the reference portfolio (of marketable loans in the case of CLOs) are transferred to a bankruptcy-remote, limited purpose entity (special purpose vehicle (SPV)), whose business function is restricted to (i) the acquisition and holding of the designated loan pool as collateral as well as (ii) the issuing of debt securities collateralised by assets in the reference portfolio. The SPV prioritises these debt claims in order of seniority by issuing different tranches of debt securities, including one or more investment grade classes and an equity tranche or reserve fund (cash reserve). Senior claims to the reference portfolio are largely insulated from default risk to the extent that subordinated tranches of the transaction absorb credit losses. Such junior tranches, i.e. equity claims and/or excess cash balances, tend to carry investment grade ratings above and beyond the rating of the underlying reference portfolio of securitised assets. The detailed configuration of issued tranches might vary with the need of the SPV to enter into interest rate swaps to mitigate asset mismatches of heterogeneous reference portfolios. This provision pertains to interest rate transformation from fixed rate assets to floating rate debt securities and the reconciliation of different interest indices used as base rates for the calculation of interest proceeds in the case of loans. By the same token, the SPV ensures that related administrative functions pertinent to loan servicing and asset transfer are carried out in compliance with contractual provisions. Second, in keeping with its monitoring task, the trustee of the transactions holds the reference portfolio as collateral on behalf of debt holders to ensure proper servicing of assets and allocates proceeds from principal and interest repayment as well as default losses to investors. Third, the servicer (and issuer if securitisation structures dispenses of a special purpose vehicle) is in charge of administering the collection of proceeds from obligors of the securitised reference portfolio on behalf of investors. Proper servicing of the underlying asset pool also includes the preservation of estimated asset value in the event of delinquency or termination (default) through maximising asset recovery in the workout process.

However, also significant differences exist between ABS and traditional (balance sheet) CLOs. First, the reference portfolio of ABS transactions is more homogenous as the majority of assets is originated by a single lender as opposed to CLO transactions, whose heterogeneous asset pool comprised of non-standard items with little or no market liquidity requires a high degree of diversification by obligor and industry in order to control exposure to default risk correlation among obligors. Moreover, prepayments and premature calls of assets in the reference portfolio are frequently observed for CLO transactions and result in erratic portfolio amortisation, which defies traditional pool level actuarial forecasting of diversification effects in homogenous portfolios.

B. Moral Hazard

Moral hazard is the second effect of market imperfection due to information asymmetry in the securitisation of loans. The effort level of the CLO originator before and after the issue date might not be incentive compatible if an insufficient proportion of net gains from collateral proceeds is allocated to investors, i.e. an issuer could be tempted to retain a large share of the high-quality portion of the collateral pool (reference portfolio) as an "effort choice" and neglect the costly enforcement of contractual restrictions imposed on loan obligors of the collateral portfolio (ex ante moral hazard of selective bias in the security design of the CLO if the issuer (i) engages in selective bias as to the loan composition of the reference portfolio and/or (ii) anticipates the opportunity of reduced effort levels after issuance of the transaction). The predicament of the resulting moral hazard is generally resolved through the separation of proceeds and losses in the allocative mechanism of loss cascading and prioritisation of claims by means of different tranches. Since large senior tranches with a high probability of full repayment would delude any incentive of both reduced effort and the inclusion of poor asset quality in the reference portfolio, issuers securitise a large proportion of interest-generating asset claims. At the same time, they substantiate the disincentive of ex ante and ex post moral hazard by retaining the most junior tranche as a commitment to bearing some credit risk.

Note in this case that we need to carefully distinguish between remedial structural measures in relation to the kind of information problem at hand. Both adverse selection and moral hazard impose agency cost of asymmetric information. Whereas the former requires issuers to increases investor information about the actual quality of the transaction (in order to achieve market separation), the latter case typically calls for some disciplining mechanism that ensure incentive compatible behaviour of issuers. Hence, the rating process of transactions and various means of internal and external credit support listed in section III. A increase market transparency, whilst essential credit enhancement required by rating agencies to at least cover expected default loss of the securitised reference portfolio clearly serves as a commitment device by issuers to mitigate moral hazard (see Exhibit 19 above). In the securitisation process the sources of agency cost are highly inter-related, e.g. ex ante moral hazard of biased asset selection--paired with some information advantage by issuers--could give rise to rational investor beliefs about adverse selection.

Originators might also be tempted to select assets of the reference portfolio according to own private information, and thus, transfer a pool of securitisable loans, which is not reflective of the general average asset quality of the loan book. This so-called cherry picking would manifest itself as ex ante moral hazard in matters of asset selection for CLOs, similar to the aforementioned first instance moral hazard, where the motivation of issuers could be to misrepresent the average loan book quality by including over-priced, low-quality loans in the reference portfolio. Such a conduct of securitisation would eventuate a gradual deterioration of the residual value of the loan book. Alternatively, cherry picking could arise as ex post moral hazard in terms of biased asset sorting, as maturing loans of the reference portfolio are replenished by the sponsor. If managers of CLO portfolios fail to successfully negotiate the structuring process, the average loan quality of the reference portfolio might deteriorate in excess of natural attrition due to prepayment of loans. Insufficient effort in extracting additional loans from the loan book to replace loans amortising prior to the maturity date of the CLO transaction ("portfolio replenishment") could be the source of such a scenario. Even though cherry picking is prohibited by national regulators, which have adopted statutory objectives in supervising the securitisation process, testing the adherence to this requirement in the bid to guard against selective bias of issuers is riddled with methodological and administrative difficulties.

A final source of information asymmetry and uncertainty surrounding the proper administration of the securitisation process is the impending danger of front running in arbitrage structures of CLOs, i.e. traders in market value portfolios prefer to trade on their own account rather than allocating the traded assets to a reference portfolio of a CLO or CDO securitisation in general. This occurs if the benefit from trading activities exceeds the gains to be generated from securitising assets based on these trading activities. Certain non incentive-compatible trading behaviour that gives rise to a situation comparable to the principal-agent problem in the corporate finance setting (Weiss, 1999), the illiquid nature of the collateral pool could also result in decreased valuation since heightened transaction cost attract allocational inefficiencies.

C. Trading Cost

Depending on the composition of its loan book, the CLO originator might be faced with the prospect of high trading cost associated with the reference portfolio of loans, as the market for potential buyers and sellers may be small (Duffie and Garleanu, 2001) and uncertainty about the true value of the collateral governs investors' beliefs in the face of adverse selection. The combination of higher searching cost and a limited pool of potential buyers of an illiquid collateral pool compel originators to offer the structured asset claims to the highest bidder at relatively short notice. Moreover, illiquidity would entail a clear shift of negotiation power from sellers to buyers in loan securitisation markets. Market-making investors recognise the risk involved in future resale of securitised claims to non-transparent reference portfolios and discount the current valuation in addition to the adverse selection effects. Originators of CLOs, however, are able to strengthen their negotiation position to a level akin to a market of actively traded assets.

The structural design of the securitisation can be geared as to flexibly remedy the illiquid nature of the reference portfolio of loans. For one, issuers might consider improving the overall average rating of the collateral pool through structural enhancements, such as counterparty guarantees, draft facilities and monoline insurance against credit loss. Additionally, issuers of CLOs reduce the net cost of bearing illiquid collateral through a particular subparticipation of issued tranches. Hence, issuers include highly liquid structured claims, i.e. large, homogenous senior tranches in the structure of the CLO transaction. Latest advancements in security design of CLOs also feature the incorporation of super-senior tranches, secured by a credit default swap as a means of improving the marketability of issued claims.

In order to keep information imperfections from compromising the proper valuation of the reference portfolio, various structural provisions are feasible. Issuers are keen to demonstrate a sufficient degree of commitment to the future performance of the reference portfolio in order to counter adversity of discounting from non-verifiability due to existing private information. In order to allay fears of investors in being "picked off" in the acquisition of a structured claim on illiquid assets, issuers of CLO transactions retain a small but most junior tranche, which carries almost all expected first loss risk. This form of adding liquidity to the more senior tranches held by investors could prevent adverse selection through signalling of quality, as only issuers of low-risk collateral portfolios will be able to afford to cover some significant amount of credit risk given the lower probability of loan default compared to poor reference portfolios. Innes (1990) proffers a model for security design, which addresses the merits of subordination to the extent that the prioritisation of claims and the bottom-up cascading of losses produces a higher structured rating and a higher valuation of tranches compared to straight pass-through securitisation structures (see Exhibit 20 below).

[ILLUSTRATION OMITTED]

However, this retention effect of the most junior tranche as credit enhancement only holds true unless the standard deviation of unexpected losses from expected losses increases over time, i.e. default losses in senior (investor) tranches and in the first loss piece develop in the same fashion over the lifetime of the transaction. Otherwise, the senior tranche would bear a gradually increasing, implicit share of total unscheduled losses. Hence, rational investors would command excess spreads on issued senior tranches as adverse selection and investor uncertainty reduce the valuation of the transaction.

Also the issue of moral hazard is addressed thereby. Managers might reduce due diligence in both asset selection for replenishment and monitoring of debtors. Just as much as the liquidity of senior tranches in the transaction assures investors of the reduced risk (of transaction cost) involved, the retention of concentrated risk in junior tranches together with the securitisation of the maximum portion of proceeds from the underlying reference portfolio mitigates such possible moral hazard of CLO managers, though agency problems of this kind cannot be entirely removed from the securitisation process. Schoring and Weinreich (1998) suggest that the principal agent problem pertinent to arbitrage securitisation could be resolved by means of subordinating a significant proportion of management fees to the issued tranches. The qualitative consequences of these structural features reflect how issuers negotiate the exercise of reputation building (reputation cost) in a market for illiquid assets such as corporate loans. As the valuation of a transaction will be significantly driven by investor confidence in fair pricing--according to the mean-variance theorem of efficient markets--issuers need to implement a structural design that prevents them from entertaining the idea of extracting informational rents.

The information asymmetries involved an illiquid reference portfolio of corporate loans have profound economic consequences for the administration of securitisation transactions. Generally speaking, the functional aspects of CLOs will make credit become a commodity that can be modelled to fit any situation of banking business as it reflects an alternative to traditional asset funding by means of deposit-taking. The importance of securitisation will almost surely grow as market participants' understanding of the process improves. Additionally, CLOs allow banks to capitalise on their core competencies in loan origination, i.e. screening, servicing and monitoring of debtors (Freixas and Rochet, 1997), whose degree of sophistication defines the margin to be generated from securitisation. In turn, rather than being event-driven in their asset allocation, banks are be able to resort to a broader portfolio of diversification practices through securitisation without venturing beyond their traditional client bases. By way of implementing credit guarantees on specific instruments and cash flows as well as broader cross-default protection in securitisation, banks are able to better specify total or partial recovery and fine-tune their capital provisions for credit risk. As banks derive a greater measure of accuracy in actively balancing credit exposures, they are evidently pressed to shore up their client relationships, which might improve the debtors' ability to attract funds. More efficient asset funding also causes various formats of loan securitisation to have far-reaching consequences for the workings of capital markets. As the likelihood of a portfolio to cause financial strain is less likely the greater the diversification of credit risks earmarked for securitisation, banks will require more favourable capital adequacy treatment. Moreover, individual incentives of securitisation lead to an aggregate effect of improved market efficiency as the existence of CLOs aids the completeness of capital markets a la Arrow-Debreu (1954).

Exhibit 16. Breakdown of senior note collateral in European
synthetic CLOs (Batchvarov et al., 2000)

Other Government             4%
CLOs                         3%
OATs                         2%
BUNDs                        5%
Others                       2%
U.S. Treasuries             26%
Public Sector/Pfandbriefe   59%

Note: Table made from pie chart.

Exhibit 17. Comparison of traditional/conventional
and synthetic CLOs (41)

                   Traditional CLO

Reference          * historical performance and diversification
Pool                 effects of loans
Credit             * credit rating of obligors and type of asset
Quality
                   * portfolio composition and payment rate
                   * margin on reference loans

Structural         * credit enhancement (retention of first
Considerations       loss provision) and liquidity support
                     (advanced payment facilities, etc.) from
                     servicer or third parties (hedging, swap,
                     guarantee, etc.) as cusion ofr potential
                     losses, including excess spread in
                     traditional CLOs)
                   * loan pool selection and replenishment
                     criteria (eligibility & substitution
                     criteria)
                   * verification of credit event, workout
                     process and redemption of tranches
                     under consideration of early amortisation
                     triggers
                   * prioritisation of losses & proceeds under
                     different scenarios

                   * legal integrity and regulatory compliance
                     (bankruptcy remoteness of SPV) of asset
                     transfer ("true sale")
                   * risks from commingling, set-off, interest rate,
                     maturity mismatch, etc.)

Originator/        * impact of loan servicing and reference pool
Servicer             charateristics on the originator underwriting
Considerations       standares, effectivness of operational
                     routines in dealing with delinquencies,
                     deptor relationship

                   * servicing competencies' administration of
                     collections and payment remittances, policies
                     of bad debt allowance, adherence to servicing
                     duties and obligations (eligibility criteria)
                   * availability of substiute servicer and call
                     provision of notes

Collateral
Risk

Sponsoring Bank    * ability to perform as swap
Credit Risk          conterparty (if applicable)

Capital Market     * sensitivity of structured rating and
Considerations       spreads to event risk and so-called
                     "headline risk"
                   * bond progile-pass through structure
                     vs. bullet structure, integration of
                     call option for premature redemption
                     of notes
                   * secondary liquidity

                   Synthetic CLO

Reference
Pool
Credit
Quality

Structural         * legal integrity of structured
Considerations       claim with/without SPV as
                     originator ("perfection of security
                     interest in collateral, etc.")
                   * loss determination and timing of
                     settlement, allocation
                   * exposure from credit deterioration
                     of sponsoring bank
                   * illiquidity and value deterioration
                     of collateral

Originator/
Servicer
Considerations

Collateral         * marketability (liquidity) and scope
Risk                 for rating downgrade/volatility
                   * potential linkage to the sponsoring
                     bank and market value

Sponsoring Bank    * ability to pay credit protection
Credit Risk          premia for interest deficiency
                   * potential exposure from other roles
                     performed in transation
                     (for e.g. a repo or hedge
                     counterparty)

Capital Market     * degree of senior note leverage in
Considerations       partially funded structures
                   * reliance on trustee and rating agencies
                     in being able to safeguard investor
                     interests (allocation of proceeds and credit
                     losses) and credit assessment respectively

Andreas A. Jobst

London School of Economics and Political Science (LSE) and J.W. Goethe Universitat Frankfurt am Main