Despite the closure of the capital markets and the demise of some of the traditional homes for project debt risk – monolines and hedge funds – 2008 has proved to be the busiest year yet for project debt securitizations. Five of a total of 14 project debt securitizations ever, closed this year. Santander, IKB and SMBC are among those currently working on deals. Other large project finance lenders with $1 billion-plus loan books, such as RBS, HSBC and Calyon, are thought to be considering securitizations.

The deals stretch back to a static cash CLO of CSFB's loan book in March 1998 (see box). The reason for this increase in activity is the contraction in banks' balance sheets and their inability to recycle capital in syndication and through non-distressed secondary sales.

Project debt securitization roll call

Bank(s)	Name of transaction	Date	Amount	Reference assets	Type
CSFB	Project Funding Co I	March 1998	$617 million	Predominately US power B loans	Static Cash CLO
CSFB	PFC 2	March 2000	$487 million	US and emerging market PF debt	Static Cash CLO
Citi	Project Securitization	August 2001	$350 million	Global PF debt	Static Cash CLO
Depfa	EPIC	November 2004	£391.7 million	UK PPP loans	Synthetic CLO
SMBC-NIBC	PROFILE	November 2005	£383 million	UK PPP loans	Synthetic CLO*
Depfa	EPIC II	June 2006	Eu900 million	PPP loans. Europe/Asia	Synthetic CLO
Dexia	WISE	December 2006	£1.47 billion	Wrapped PFI/ utilities bonds (80% UK)	Synthetic CBO
SMBC	SMART	March 2007	£388.8 million	UK PPP loans	Synthetic CLO
Depfa	EPIC III	December 2007	Eu666 million	Wrapped UK PFI/ utilities bonds	Synthetic CLO
ABN Amro; Banco Espirito Santo	Lusitano	December 2007	Eu1.1 billion	PPP, UK, Spain, Portugal and Hungary	Cash CLO
NIBC	Adriana	April 2008	Eu963 million	European PPP/infra	Cash CLO
SG	COSMOS I	August 2008	Eu940 million	Global PF loan book	Synthetic first loss CDS
Lloyds TSB	Exeter Blue	October 2008	Eu1 billion	PPP, UK, Western Europe, Aus.	Synthetic CLO
Depfa	Sirrah Funding 2	November 2008	Eu2.68 billion	UK, Aus Utilities/PPP	Cash CLO

*NIBC benefited from a fully funded transaction, selling 41.5% of the initial portfolio to SMBC.
Source: Project Finance Magazine

While the frequency of the deals has changed, so has their structuring. The first three portfolio debt securitizations – by CSFB (twice) and Citi – were fully funded cash deals; all of the remainder were synthetic deals, until late 2007 when the tide turned in favour of cash deals.

The reason synthetic deals have fallen in and out of vogue has broadly matched the credit cycle. In the good times banks were primarily interested in releasing regulatory capital; synthetic CLOs are much more efficient at releasing capital at advantageous pricing. Now, in the leaner times banks are more interested in liquidity and favour the cash deals.

Spreads have widened considerably, making both synthetic and cash deals less efficient. Wide spreads are a function of the lack of investor appetite. The traditional providers of the super senior swaps on synthetic deals, monolines, are no longer active buyers and many of the hedge funds, which were active traders of the subordinated tranches, are no longer in business.

"Without a functioning securitization market the traditional synthetic CLO model is broken," says Mark Gordon, head of securitization at SMBC. "What we are seeing in the market at the moment are legacy deals. Putting together CLOs is a time-consuming and expensive process; no-one will start a new rated, project finance CLO in the current market."

Instead of arranging CLOs to pitch to the general market, banks are courting investors with bilateral agreements. Lloyds TSB closed just such a synthetic securitization with a 'key strategic investor' in October. Lloyds' Eu1.062 billion CLO references 49 senior secured loans for PPP/PFI, transport, power, energy and regulated utilities projects, predominantly in UK, Western Europe and North America.

          Size         Fitch       Credit              Final 
Class (EUR m) Rating     Enhancement    Maturity             Coupon
A       31.875    AAA        8.85%            Oct 2025           3mE+125bp
B       31.875    AA           5.85%            Oct 2025           3mE+225bp
C       26.5625  A              3.35%            Oct 2025           3mE+325bp
D       10.625    BBB          2.35%           Oct 2025           3mE+600bp
E         8.5         BB            1.55%           Oct 2025           3mE+1000bp

The issuer of the Eu126 million notes is a special purpose vehicle, Exeter Blue Limited, based in Jersey. Lloyds has entered into six credit default swaps (CDS) with the issuer against the super-senior tranche and A-E tranches. The bonds in each tranche are sold to the investor, with the proceeds held in deposit by Lloyds. The notes benefit from credit enhancement through over-collaterisation and 50bp excess spread. (Excess spread is the issuer's profit and the first layer of protection for noteholders; it makes the deal more expensive for the originator but improves tranching efficiency.)

A Eu936.6 million super-senior portion (representing 88.1% of the capital structure) and a first-loss equity piece of Eu16.47 million (1.55%) is retained by Lloyds, presumably because the pricing was too high.

The demise of monolines, the usual providers of super senior protection, has had a direct impact on the efficiency of synthetic CLO structures. However, this is tempered by the new treatment of super senior tranches under Basel 2. Already some banks have begun to call the super senior tranches on their securitsations because it is more efficient to bring them on balance sheet. Under Basel 1 the super senior tranche is 100% risk weighted, as credit ratings have no effect on regulatory capital, but under Basel 2 with standard compliance this drops to 20%.

As an illustration of how far the market has moved, Depfa's Eu900 million EPIC 2 synthetic deal, which closed in June 2006, was structured with just Eu79 million in notes between AAA-BB (8.8% vs. 11.9% on Exeter Blue) despite similarly sized super senior (89%) and first loss (1.75%) pieces.

True, there are differences between EPIC and Exeter: the reference asset class is different (EPIC was backed by global PPP debt) and the EPIC structure featured KfW as swap intermediary. Nevertheless, the tranching differences and huge differences in pricing shows that synthetic CLOs have become a lot less efficient. The pricing on EPIC 2 was AAA-38bp, AA-60bp, A-90bp, BBB-190bp.

It is even possible that the published margins of the Lloyds CLO do not reflect the true cost of the Exeter Blue securitsation if the investor bought the debt below par value.

Engineering CLOs: a new paradigm

In the current climate the most prudent route for banks wishing to free up regulatory capital through a synthetic CLO or improve liquidity with a cash CLO is to secure a preliminary agreement with an investor before starting the structuring. The Lloyds deal is believed to have been placed with a pension fund.

Project finance should be a relatively easy sell, benefiting from low default rates and low rates of loss given default (senior debt has a global average 72% recovery in projects that have defaulted in the last 15 years). One banker said that he was fielding reverse enquiries from investors new to the infrastructure sector about taking a first loss piece in a portfolio of project debt.

Banks that are in talks with potential investors are keeping their contacts close to their chests. It is known that a number of banks are in talks with non-bank investors outside the usual buyers of project risk such as pension funds and life companies. These investors do not mark their assets to market.

In August Societe Generale (SG) closed a Eu940 million synthetic deal, COSMOS I, which involved the purchase of protection via a credit default swap (CDS) with a strategic investor, thought to be a pension fund. The reference portfolio includes 53 project finance and renewable energy deals in the US, UK, Europe, Canada and Australia.

"The difficulty with tranching at the moment is that the pricing of the AAA piece is excruciatingly high, and that eats into the equity return, and makes previously executed synthetic and cash CLOs unattractive," says Matthew Vickerstaff, global head of infrastructure and asset based finance at SG.

While pre-crunch super senior pricing came in at around 10bp, the £1 million portion of Depfa's EPIC 3 which closed in December 2007 came in at 65bp for insurance from Assured Guaranty.

SG initially contemplated a cash CLO but adapted its strategy in early 2008, according to Vickerstaff, "as the AAA piece became far too expensive." In a discussion with investors SG decided to proceed with a first loss CDS. The investor forms a view on the SG management team, its risk assessment practices, and – because the deal is replenishable – its origination capabilities. SG will keep a tranche of each underlying project financing so that it has a vested interest in monitoring the projects. "SocGen's and the investor's interests are fully aligned, which is key."

"There is a class of investor such as pension funds and life companies who are interested in the infrastructure market, and are keen to diversify their exposure across a number of deals," adds Vickerstaff. "And for a bank it is effectively another way of bringing in highly desirable regulatory capital."

Lloyds and SG have avoided going to the wider market with a synthetic deal by opening their loan books to select investors. Both deals, although engineered differently, are driven principally by freeing up regulatory capital.

Cash deals

But what about the banks whose primary concern is liquidity? Unfortunately, there does not appear to be an active market of buyers for true sale project debt securitizations. Rather, the banks that have arranged cash CLOs have almost certainly posted the most senior tranche as collateral with the European Central Bank (ECB) under its asset backed repurchase agreement (repo) arrangements.

In December 2007, ABN Amro and Banco Espirito Santo (BES) closed a Eu1.1 billion cash CLO. BES retained the notes from the CLO because of turmoil in the market, with the option of posting the AAA portion of the securitization, equivalent to 82.5% of its face value, with the ECB as it needs funding.

NIBC followed suit in April with a Eu1.1 billion cash CLO, and again a Eu963 million (A3) senior portion priced at 50bp was believed to have been placed as collateral with the ECB. The issuer acquired 70% of the assets in its portfolio at close and have until May 2009 to buy the rest.

The market has known for some time that NIBC needed liquidity. In early April 2008, NIBC Holding announced it had written off Eu300 million from its US commercial real estate portfolio after Eu255 million in losses on the assets in the previous year. An equity injection by shareholders of Eu400 million was also announced to help it maintain its Baa1/BBB+/A- rating.

Similarly, Depfa and its parent Hypo Real Estate are taking measures to strengthen its balance sheet after Hypo's Eu50 billion bailout. A pioneer of the synthetic CLO, Depfa closed a Eu2.68 billion project debt cash CLO in November 2008, with the senior portion almost certainly placed with the ECB.

A Eu403 million unrated subordinate B tranche, which Depfa retains, provides support for the Eu2.28 billion senior tranche's A rating. Tranche A has a margin of 200bp over three-month Euribor, and tranche B 600bp over six-month Euribor. Depfa used to originate PPP deals at sometimes less than half 200bp, the deal shows how desperate Depfa and its parent need term funding.

Class A note holders benefit from an over-collateralisation test that is based on the ratio of the par value of the reference pool to the size of the debt. As the deal is a single-rated tranche structure and the class B notes are solely for the protection of the class A notes, the class A over-collateralisation trigger is set very close to the closing value. The par value at closing is 117.65% and the over-collateralisation trigger is 117.6%.

As well as providing funding, ECB-eligible collateral securities are treated as liquid for the purposes of Basel 2, with the senior tranche conferring up to 24% regulatory capital relief on the tranche for the bank.

Around Eu400 billion has been placed by European banks under the ECB's ABS repo programme. However, this cannot continue indefinitely because – as the EU recognises – the ECB distorts the functioning of the asset-backed market. The ECB arrangement must be viewed as an interim fix.

Moves to wean the market off the ECB credit line are already afoot. The ECB currently applies a valuation haircut of 2% on floating rate asset backed securities, but from February these rules will be less generous. If the value of a security is based on a theoretical level rather than a market price, and without active trading all senior tranche cash CLOs will be valued theoretically, the ECB will apply a 5% haircut before applying a higher 12% haircut to any and all ABS used as collateral.

CRAs – harsh or prudent?

One of the key criteria for ECB repo eligibility is that the senior tranche is rated at least A-/A3 by at least one of S&P, Moody's and Fitch. The ratings agencies have been present on all project debt securitsations to date. The agencies stress test the reference portfolio according to their methodologies, set the attachment points for the differently rated tranches and preside over the required over-collaterisation, excess spread and other credit enhancements needed to achieve given ratings.

Ratings agencies will still be needed on cash deals: at least one rating is needed for the ECB repo agreement, and two are usually required to pitch the paper to the wider market. However, for synthetic deals that are not marketed but arranged on the back of the bilateral agreement between a project bank and a non-bank investor, rating agencies could become redundant.

Banks want to drop rating agencies because of the changes in their methodologies regarding project CDOs. "The interface between agencies' structured finance and project teams has always been a bit of a mess," one source snipes.

The agencies' methodologies have become more conservative because they have increased the correlation among project financings in their models. The consequence of these changes is that tranching has become less efficient and expensive because greater credit enhancement is required.

For instance the class A (A-rated) notes on Depfa's recent cash CLO have an over-collateralisation trigger of 17.6%. Before the credit crunch this would have been nearer 5%.

New rating agency methodologies have landed banks wanting to structure CDOs with both dwindling investor appetite and more onerous credit enhancement requirements.

Some bankers, with some justification, have called the ratings approach to project debt "reactionary". More conservative methodologies were always likely, given how badly the agencies underestimated the correlation among reference assets in mortgage securitisation and – as the credit crisis has bitten – corporate debt CDOs.

The ratings agencies are in an unenviable position as arbiters of risk; no matter how many Monte Carlo scenarios they run based on past default rates and similar credit events, they are easily subject to the empiricist's trap. That is, just because an event has never happened before, does not mean it will not happen in the future. As the current financial crisis has shown, large outlying, systemic risks are impossible to predict or measure.

Because the market for synthetic CDO risk has fallen as bilateral arrangements have risen, if the investor is comfortable the rating agencies can be omitted. "They bring nothing new to the table," says one source.

At least three project CLOs have been either dropped or revised because of tranching inefficiency caused by changes in ratings agency approach: SMBC has dropped its Middle East debt securitization, which it originally began working on with Calyon. Santander was believed to have been working on a synthetic CLO but has amended its plans. SG was working on a cash CLO but revised its plans in early 2008 when it was clear the senior piece was too expensive.

Although SG used a rating agency to rate its underlying portfolio, Matthew Vickerstaff is adamant that future application of a first loss CDS is not reliant on a rating: "We are not beholden to the rating agencies."

Presently banks considering a securitization of their project portfolio have two choices, depending on whether their primary goal is to free up regulatory capital (CDS) or obtain extra liquidity (true sale).

Synthetic deals which are agreed bilaterally with a non-bank investor that has done its due diligence or has confidence in the originating bank may get done without credit rating input. As banks move to advanced Basel 2 compliance and can set aside regulatory capital according to their own internal credit ratings, synthetic securitizations involving an investor and a third-party accountant to verify bank risk procedures may be how future deals are engineered. For cash deals, ratings agencies will still be required.

The golden path

The two goals of regulatory capital release and liquidity remain largely mutually exclusive using a single securitization structure. "There are broadly three goals for a securitization of project debt," says Matthew Job, senior counsel at Linklaters. "One, liquidity; two, risk transfer and the release of regulatory capital and; three, overall pricing. These are not mutually consistent and there is an inevitable compromise between the three – it is hard to see how one could structure a deal which fully achieves all three simultaneously."

A synthetic deal efficiently shifts the risk, but is largely unfunded, and a cash CLO provides liquidity but is relatively expensive. Banks generally have a cost of funds below Libor but cash CLOs have a far higher cost of funding. Depfa's recent cash CLO, for example, has a weighted average cost above 200bp. Cash CLOs are also less efficient from a regulatory capital point of view because banks tend to keep the junior tranche.

As an illustration, before the credit crunch a synthetic CLO's super senior tranche, rated higher than AAA (theoretically more secure than US and UK sovereign risk) would comprise up to 90% of the capital structure and be priced at below 10bp. This would give a weighted average cost of the whole capital structure of around 20bp.

For the same collateral pool a cash CLO typically has a different tranche structure. For instance, 85% of the capital structure would be AAA notes, 10% single A and the remaining 5% junior-most equity. Pre-crunch pricing would typically give a weighted average cost of around 70bp for the entire capital structure of a cash deal; Depfa's cash deal has shown that now this margin has roughly trebled.

It is possible to combine a synthetic CLO and cash CLO on the same pool of assets to receive term funding and efficient regulatory capital relief. This dual approach is effectively equivalent to debt syndication or a secondary sale. However banks are tweaking this model before it is implemented and actively looking at alternatives to cash CLOs to provide funding at better pricing.

SMBC is currently working on a covered bond securitization partially backed by a portfolio of UK PPP/PFI debt. Covered bonds benefit from recourse to the issuer – unsecured corporate credit – with preferential recourse to a pool of reference assets. Covered bonds should offer a pricing advantage over cash CLOs because they benefit from double default protection and the associated higher credit ratings. They can also be priced over mid-swaps rather than Euribor.