The securitisation of project finance debt is a niche investment, and a niche occupation. Loan volumes, as well as the number of banks involved in project finance, have increased sharply in the last several years. Collateralised loan obligations (CLOs) for project debt was billed in 2000, not least in these pages, as a logical next step for project loan portfolio managers, but has had few takers since.

The exceptions have been the two synthetic securitisations of project finance debt – November 2004's Essential Public Infrastructure Capital Ltd (EPIC) and November 2005's Stichting PROFILE Securitisation 1. EPIC was a £391.7 million ($680 million) securitisation of PFI loans that DEPFA Bank had originated, while, Stichting was a £383 million securitisation of SMBC and NIB Capital commitments.

The scope of the two offerings was deliberately limited. Only UK Private Finance Initiative loans were to be included, and the loans are still serviced by their originators, although NIB sold its commitments to SMBC to streamline the Stichting deal. But the deals do transfer the credit and economic risk to the capital markets, and allow their issuers to pursue a greater volume of project finance business than their groups' budgets might allow.

CLOs pool together the repayment obligations on several loans, and issue bonds whose credit is based on that of the underlying pool of loans. They offer an element of capital release, which can improve a bank's return on equity, allow it to set aside less capital for project loans, and provide more loans to projects than previously.

The history

Collateralised loan obligations, at least the large volume ones, go back ten years now, to NatWest's ROSE funding. Open-ended CLOs, those that do not have a fixed reference pool of assets, have become players in the debt markets in their own right. They have, for, example, become important buyers of B loans issued by US power producers.

The first project finance CLO –was Credit Suisse First Boston's $617 million Project Funding Corporation I in 1998. CSFB followed it with Project Funding Corporation 2 in 2000 – a $498.6 million deal – and Citigroup closed a $350 million issue – Project Securitization Co.1 – in August 2001.

Citigroup said at the time that it was considering a number of further issues, and SG and Deutsche both went on record as examining the use of CLOs in distributing their portfolio's credit risk, and SG's then group head, Jay Worenklein, suggested that some kind of synthetic approach to selling on the credit risk – the use of credit derivatives – would be helpful.

But since then, until EPIC, there were no more deals. And the reasons for this lull are legion. The most convincing ones cluster around the difficulties in structuring a CLO, and gaining generous enough capital treatment. EPIC solved some, but not all of these issues.

The lending market tends to plaster banks that do CLOs with a skin of rumour – many in the market assume, rightly, or wrongly, that a CLO signals a bank's exit from the lending business. These accusations have not been completely unfounded – Citi and CSFB are now primarily known as bond, B loan and advisory players.

However, DEPFA and SMBC have been, and continue to be, active players in their markets. Indeed, according to Andrew Bride, who supervises the CLO programme at DEPFA, the ability to recycle capital in this way is an important aspect of the Infrastructure Unit's business model.

In the existing CLO structures, the debt breaks down into three classes – a super-senior, which can be sold to a monoline or other insurer, a mezzanine, which attracts a universe of specialist and high-yield investors, and a first loss piece, which is usually retained by the bank, or sold to a very small group of investors. While the super senior piece is usually over-collateralised, the retained first loss attracts a capital treatment that potentially undermines the economics of a CLO.

Another is the reaction of borrowers to a lending bank's securitisation programme. Most sponsors take a bank's willingness to hold their projects' debt as a sign of their commitment to the sponsor's business. Selling on such commitments will often require a notification, although this condition is not insurmountable. The EPIC deal, for instance, used credit default protection from the Kreditanstalt fur Wideraufbau to mirror the transfer of loans. But the Citi and CSFB deals were partly cash transactions.

Practicalities and possibilities

Then there are the related practical issues associated with assigning security and gaining confidentiality waivers. Assigning security to the underlying lenders, especially in a diverse group of jurisdictions, can be difficult. For this reasons, the first CSFB deal, which consisted overwhelmingly of US loans, and the EPIC and Stichting deals, both confined to the UK, were simpler affairs to structure.

Confidentiality is cited as one of the key factors that has held up further CLOs. Sponsors are frequently unwilling to allow lenders to open up their transactions to outside scrutiny. According to Bob Dewing, the managing director at Citigroup that structured its CLO, this can hit the banks' ability to transfer credit risk, synthetic or real, to an outside vehicle. Dewing says that PPP loans typically come with less onerous conditions.

If this is true, then CLOs will remain a niche within a niche, albeit a valuable one, since PPP loans, which typically feature sub-150bp pricing over long tenors, offer among the poorest returns on equity in a bank's portfolio – DEPFA estimates that a 100bp loan to an infrastructure project, against which a bank must set aside 8% capital, produces a 12.5% return on equity. Bride, asked whether confidentiality is difficult to overcome says that "the relative standardisation of UK PFI means that many people are familiar with the overall structure of the deals, making due diligence and disclosure requirements relatively unobtrusive. This might not be the case with a portfolio with greater geographical or sectoral diversity."

Loan quality and catastrophes

The dominance by transport and PPP deals of project finance lending volumes, however, likely to be mirrored within most bank portfolios. At the same time, many US power deals have migrated to the term B market, where they have become natural assets for open-ended managed CLOs. The collapse in US and UK power markets, while they do not appear to have destabilised the existing CLOs, has still dragged on the credit quality of present-day power financings.

And it has also made banks aware of one alternative to a securitization, at least for the less well-performing credits. The secondary loan market has opened up for project debt in the years since the earlier deals, although this has largely benefited banks with distressed positions, or loans that would be unlikely candidates for securitization in any case.

Again, the collapse in the US power market, as well as the collapse in UK wholesale power prices, were the drivers for banks to offload project finance debt. Dresdner held probably the most high-profile auction of project debt, although a cursory look at the bank list for some recent US power restructurings shows that a majority of the lenders to many distressed US projects sold on their commitments.

At the opposite end of the spectrum, several banks have now been able to take advantage of conduits to sell on export credit agency-covered loans. Citi's GovCo conduit has been in place for several years, and has been used frequently to house US Ex-Im back debt. In October, ABN Amro used the $1 billion SovRisc conduit to securitise ECA-backed loans, and Calyon has used the European Sovereign Funding conduit to fund ECA-guaranteed aircraft loans.

The intervening years have also seen the steady development of markets for project bonds, thus allowing banks to recycle capital on a project-by-project basis. However, this process has not always been smooth – European PPP bonds are still relatively rare, and there are now very few cross-border bonds from emerging markets – a legacy of the currency crises of the late 1990s. But in the US and increasingly the Middle East they have become important structural components of lending markets, as Qatargas 3 and the Skyway refinancing both indicated.

The Basel II effect

But the greatest factor in the development of the CLO markets – for project finance or other loans, has been the treatment accorded to bank lending by the Basel II revised international framework. For project finance lenders, the news since the Basel committee issued the new rules has been largely positive.

The rules, as drafted, initially suggested that project finance, as a type of specialised lending, would require banks to set aside large amounts of capital against such loans. The proposals caused a stir, and if implemented as first proposed would have undoubtedly pushed several banks to sell off their loans in some form or another.

The proposed accords led to a multi-bank study on the default probabilities and loss-given default proportions associated with project finance debt. The study, by Standard & Poor's RiskSolutions group, contained broadly positive news for lenders, and indicated that both default levels and loss given default levels were much below comparable corporate credits. (For more on the Basel treatment of project finance and banks' reaction, search "Basel" at www.projectfinancemagazine.com).

The Basel committee came back with an approach that suggested that banks could use both the external data generated by the S&P study and their own ratings models in calculating the amounts of capital to set aside against their commitments. This concession reduces, although it does not eliminate, the spur towards a securitisation of project debt. DEPFA's deal progressed according to the assumption that the loans would attract a 100% weighting.

Even more EPIC

But DEPFA is going forward with a second EPIC securitisation, one that will likely involve loans from a much wider range of jurisdictions than the first. The groundwork for this development was apparent in the first EPIC deal, since the use of a synthetic structure, and rejection of a true sale approach, was the result of concerns over the withholding tax and security issues that might arise from selling loans to an issuer located outside the location of the borrower.

This second EPIC issue will probably involve the bank's commitments to North American and continental European projects in addition to a further tranche of UK PFI transactions. According to Bride, the deal, whose composition and size are not yet final, will not involve a more complex structure, but will involve a more extensive rating, due diligence and sales process. It may also involve a greater number of loans to projects in construction, an advance that distinguished Stichting from its predecessor.

But DEPFA's second EPIC will not include any assets outside of the bank's business line – PPP and transport. And it will have access to KfW for the purposes of being a swap intermediary. Other banks, particularly those that would like to securitise loans to other sectors, will likely have to choose a funded structure, whereby they would enter into credit default swaps with an investment vehicle that would need to cash collateralise its obligations, and sell credit-linked notes on to investors.

To fund or not to fund

Bride says that DEPFA examined a funded option, but that the use of the KfW as an intermediary and selling the super-senior portion was much cheaper than the spreads available in the market for funded protection. But the price difference between the funded and unfunded options is much smaller than it was in late 2004, and it might be possible to issue a relatively competitive deal in the current interest rate environment. High liquidity levels, and thus supportive conditions for such an issue, will probably not persist.

A follow-on issuer might benefit from the increased sophistication of ratings agencies. The first CSFB and Citibank deals required the agencies to adopt a novel approach to the diversification tests by which they normally assess collateralised debt obligations. This reflected the fact that the earlier deals featured a smaller number of loans and a relatively small number of jurisdictions. The EPIC deal, if anything, possessed this concentration to a much greater degree, and the issuer had to persuade the ratings agencies that the risk of default did not correlate across PFI loans beyond the risk of UK government default.

But ratings agencies have been less forgiving in assigning ratings to low investment grade projects, as some road, and more particularly power, issuers have been known to complain privately. Some have suggested that the capital markets have been willing to price their issues at levels equivalent to a much higher rating. If this divergence in credit assessment could be exploited by a CLO issuer, then securitisation would be more popular. The fact that ratings agencies are integral to the assembly of CLOs, however, would likely dampen any opportunity for arbitrage.

The reasons above all offer some explanations as to why project CLO volume has been so thin, and why DEPFA, SMBC and NIB have been the only takers in the last five years. The final explanation is manpower – many of the banks that looked seriously at CLOs in the intervening years had staff numbers above what dealflow demanded. Few now do, and the process of putting together a transaction is, as DEPFA's Bride notes, "very time-consuming."