Under the current bank regulatory capital (BIS or "Basle") rules which require banks to set aside capital amounting to 8% of the principal amount of a loan to private corporate borrowers (such as the special purpose companies set up to carry out a PFI project), the returns to be made from infrastructure lending are relatively low. For example, an interest margin of 1% will produce a pre-tax and pre-cost return on equity (RoE) of 12.5% per annum for the lender. If risk provisions are also taken into account this return decreases further.

In recent years, some of the leading banks once very active in lending to the project finance sector generally have withdrawn to a greater or lesser extent. One of the reasons for this has been the low return generated from the lower risk-reward end of the business (such as PFI/infrastructure lending), and the losses from problems in the higher reward but riskier sectors such as merchant power and telecoms.

Whether banks are able to make sufficiently good long-term returns from PFI lending should be a matter of concern to governments, particularly in the UK but also in many other countries where a sufficiently large group of potential lenders for the financing of projects is needed to ensure competition and thereby value for money in the financing terms just as in the other construction and operational components of a project's costs.

Despite the growing involvement of the capital (as opposed to bank debt) market – especially through the involvement of the large monoline insurance companies that specialise in credit-enhancing project bonds with their own AAA credit ratings) – there is a strong continuing need for the banks to remain involved in this sector; not just to provide funding liquidity, but also because of their experience in carrying out the initial due diligence and ongoing monitoring of PFI projects through construction and operation provides government with an important additional layer of protection.

DEPFA Bank has sought to address this issue, through Essential Public Infrastructure Capital plc (EPIC), the special purpose vehicle incorporated for the £391.7 million part-funded synthetic securitisation transaction which achieved financial close in November 2004.

By way of background, the bank's core business is direct lending to central, regional and local government entities. But its ancillary activities in advising the public and private sectors on, and the limited-recourse funding of, the private development of core public infrastructure projects – roads, railways, schools and hospitals – through PPP's, is an area of growing significance as its public sector client base wrestles with the conflicting problem of delivering high quality public services without increasing the financial burden on taxpayers and the government budget deficit.

At the end of March 2005 the Bank had total assets in excess of Eu200 billion, and current commitments to the limited-recourse infrastructure sector of c. Eu2.6 billion.

The repackaging of high quality assets and their sale to investors is a core part of DEPFA Bank's business, carried on for many years through Pfandbriefe AAA rated bond issues in Germany and more recently through Irish Asset Covered Securities AAA rated issues under new Irish legislation. Seeking to securitise a pool of infrastructure finance loans was therefore a natural, although potentially challenging, extension of DEPFA's existing business model. In undertaking this transaction, the objectives were:

(i) relief of regulatory capital in what is the only main business segment of DEPFA where the assets attract a 100% BIS capital risk weighting (lending directly to OECD governments is risk weighted 0%, and lending to local authorities in most OECD countries is risk weighted 20%) ;

(ii) enhancement of return on capital: the Bank's return on equity is currently around 30% with a long-term target in excess of 20% post-tax. From the perspective of DEPFA's senior mangement, there was always a strong desire to improve the return diluting effect of involvement in this sector;

(iii) broadening of the investor base for infrastructure financing: commensurate with DEPFA's exclusive focus on assisting the public sector with all of its funding requirements is using the Bank's structuring and "warehousing" capability to underwrite infrastructure financings and hold them on its balance sheet until the underlying projects reach a proven level of cash-flow generation at which point they are readily refinanceable by the wider institutional investor base;

(iv) transfer of credit risk: the Bank's long-term credit rating is AA- (or equivalent from Standard and Poor's, Moody's and Fitch) and to optimise the Bank's funding cost it is obviously important to maintain this level of rating. Therefore, with most PPP /infrastructure project loans being rated at the BBB level, it was always going to be important to manage the Bank's exposure to this lower-rated (and therefore more expensive to fund) asset class;

(v) creation of a securitisation model with the potential to be replicated for DEPFA's infrastructure loans in other countries in due course.

Selection of transaction structure

2003 was primarily spent planning: having discussions with potential investors, other market players who were contemplating similar transactions and with investment banks interested in arranging and managing such an issue. The aim was to identify a practical structure that would give DEPFA the capital relief it was seeking and that could be implemented without undue difficulty. The choices were essentially between a true sale, a funded synthetic and an un-funded synthetic collateralised loan obligation (CLO) issue.

A true sale or "cash securitisation", which would involve transferring the legal ownership of the loans to a special purpose vehicle would achieve a 0% capital weighting and therefore deliver a number of the required objectives. However, a number of potential practical difficulties were identified. As indicated above, we were seeking a securitisation model that would be capable of replication and application to loans from projects in the full range of markets in which DEPFA operates (Europe, North America and Japan), if not immediately then at least in due course.

The first point of concern with a true sale structure was the identification of a jurisdiction to which loans from a number of countries (the decision to only use a pool of UK PFI loans had not at that point been made) could easily be transferred without withholding tax being levied on them or indeed on payments to investors.

The second concern was the potential difficulty of the legal transfer itself and if not of the transfer of the loans, certainly the transfer of the underlying security. Furthermore, the true sale route had the potential to have an impact on clients, possibly creating concerns both about continued funding of the loans and about their relationship with a lender that was a specialised securitisation vehicle.

A true sale having therefore been ruled out as impractical for this asset pool, the remaining options were the various synthetic structures. These involve different variations on the theme of the Bank entering into a credit default swap (CDS) either with an investment vehicle which would cash collateralise its obligations and hedge itself by issuing credit linked notes (CLNs) to the market (a "funded synthetic") or with another bank which would hedge itself by entering into CDS's on the same risk with investors (an "un-funded synthetic").

A funded synthetic, by providing cash collateral for the amounts payable to DEPFA by the investment vehicle in the event of a loan loss, would achieve the ideal 0% BIS risk weighting and avoid the pitfalls of a true sale structure and therefore initially appeared attractive. However, the pricing for the various classes of CLN indicated by the banks with which DEPFA was discussing the transaction was not appealing: once costs were taken into account, the excess spread would be negligible. This led us to conclude that an un-funded synthetic would be the only feasible route: the pricing offered by super-senior risk takers, essentially the monoline credit insurers, in the unfunded structures, was compellingly lower than that obtainable for CLN's at the same rating levels.

Intermediation by Kreditanstalt fur Wiederaufbau (KfW)

Having concluded that the un-funded synthetic structure was the optimal option given our objectives and having appointed Merrill Lynch as arranger for the issue, an approach was made to KfW (the German government's AAA rated promotional bank) to see if they would be prepared to intermediate the transaction. Their 0% risk weighting (derived from their government-owned status) would bring the capital advantages of a funded structure along with the cost advantages, subject to KfW's intermediation fees, of an un-funded one. KfW agreed to help., principally because they recognised the benefits that this type of transaction could bring to improving bank liquidity in the wider European infrastructure market.

One important stipulation they made was that the tranches of the issue below the super-senior AAA rated tranche taken by the monoline, should be cash funded in order that KfW would have cash collateral for the investors' obligations in those lower rated tranches, where there was a perception that the risk of default in the securitised portfolio and therefore a potential need for reimbursement from investors was higher. The structure therefore became a part-funded synthetic in which DEPFA entered into a CDS with KfW which in turn protected its position, through a combination of CDS's and credit-linked "Schuldscheine", (i.e. a type of promissory note) as set out in Figure 1 above.

Intermediation by KfW also meant that the transaction could benefit from the established securitisation documentation developed by KfW and others in the context of KfW's "Promise" and "Provide" securitisation platforms.

UK PFI or pan-European asset portfolio?

Having settled on the structure, the next major issue was to identify the precise pool of assets to be securitised. DEPFA's infrastructure business covers Europe, Japan and North America. Based largely on the volume of assets, it was decided that a European focus would be appropriate. It is here that the benefit of the synthetic structure came through: DEPFA and Merrill Lynch were able to identify what was felt to be an optimum pool of assets with the only real selection criteria being the views of the rating agencies and the potential appeal to investors. Issues of loan and security transfer, re-registration costs, stamp taxes and withholding taxes could be ignored. Ultimately, it was concluded that a pure UK PFI-based transaction offered the cleanest "story" for investors: in any event the UK would have dominated any pan-European issue, reflecting the far larger volume of transactions in the UK because of the longer history of the PFI in the UK than elsewhere.

Credit ratings

The monoline insurer who would undertake the super-senior swap, required two ratings of the overall transaction. This necessitated getting rating estimates for a large number of the UK loans on DEPFA's books from both agencies (Moody's and Standard and Poors (S&P)). For each loan, the "credit story" had to be traced from the original analysis, through change orders, waivers and all the other events that affect a project on its route from the drawing board to successful operation: a considerable workload burden on both the agencies and DEPFA.

The ratings and ultimate tranching structure were as follows:

Class                  Amount       Standard & Poors         Moody's
                           £ million      Rating                          Rating*
Senior Swap       355.70         AAA                          Aaa
Class A+ Notes  0.25             AAA                          N/R
Class A Notes     17.90          AAA                          N/R
Class B Notes     3.95            AA                             N/R
Class C Notes     3.00            A                                N/R
Class D Notes     3.00            BBB                           N/R
Class E Notes      3.95           BB                              N/R
First Loss (Equity)3.95           N/R                            N/R
Total                     391.70  
*Note: Moody's was not required to rate the tranches below the
  Senior Swap. (N/R = not rated).

In addition to the individual rating estimates of the loans, the other issues that were of importance were those of expected recovery rates and default correlations between loans. In estimating recovery rates, it was recognised that loans to PFI borrowers are fairly carefully protected by a network of direct agreements, termination compensation sums and recourse to defaulting contractors and sub-contractors. These combine to lower the expected level of loss following a default: 78% of the portfolio by value was estimated by S&P to have expected recovery rates of 85-90%. This would mean that a defaulted loan of, say, £20 million would on average cause a loss of no more than £2-3 million.

Initial discussions on default correlations essentially centred on the concept of the PFI as a "sector" and the risk of one loan going in to default once another had done so was high compared to loans to corporates in the same sector. On examination, it was agreed that correlation levels between different PFI loans, were in fact likely to be lower than those of similar corporate borrowers. Each loan effectively lives in its own controlled, contractually created world, for the most part isolated from the volatilities created by changing demand and economic activity. Correlation, to the extent that it exists comes from:
(i) dependence on the UK public sector for ultimate repayment;
(ii) the existence of common construction or operation contractors and sub-contractors; or
(iii) the existence of common shareholders with continuing financial obligations to a project.

Low correlation factors were used in the rating models to provide for these limited common points of interconnection.

In the end, the high recovery rates and low correlation factors were key to the transaction's ratings and capital structure, with its relatively modest risk retention by DEPFA, with first loss "equity" exposure at approximately 1% of the £391.7 million transaction volume.

Loan Eligibility and Replenishment

Having opted for a PFI-based pool of loans, we had to agree criteria which the pool assets would have to meet. Furthermore, given the regularity with which bank-debt PFI loans tend to be refinanced, DEPFA needed to be able to replace the original loans with new ones, of a comparable credit quality and duration. A set of replenishment criteria to define what replacement assets could be put into the pool without creating new "concentration" issues was negotiated.

Although securitisation techniques are increasingly being used for complex asset classes, many of the structures tend to have developed from markets where loan structures are relatively simple, covenants generally fairly straightforward and where the principal event of default is failure to pay.

However, it is widely recognised that many project finance loans, with their complex control matrices, are often in "technical default", at least in some sense, and PFI loans are no exception.

This meant that an eligibility criterion based on there being simply no events of default in respect of any of the loans would not work, principally because the loan agreements do not themselves generally distinguish between "serious" events of default and more "routine" ones. Distinguishing between those events of default that indicate real trouble, and those that simply reflect either bureaucratic inefficiency or the common but ultimately resolvable technical issues or design changes that arise during the course of a construction project, and then trying to reflect this in the documents satisfactorily was one of the more difficult tasks.

"Replenishment" was a more complicated issue. Not only would new loans have to meet the same eligibility criteria as the loans originally in the pool, but they would also have to pass a number of further tests. Investors were concerned that the diverse, mostly investment grade initial pool could tend to drift, as loans were replaced, towards a sub-investment grade, concentrated pool of far greater risk. In order to try and address this, criteria related to, inter alia:
(i) rating estimate,
(ii) repayment profile,
(iii) investor/operator/construction contractor concentration, and
(iv) the percentage of loans in construction
(v) were included to give investors comfort.

Summary and conclusion

Through EPIC, DEPFA has packaged and transferred the risk (and part of the return) of a section of its PFI loan portfolio and achieved its original objectives of releasing capital, improving returns, transferring credit risk and bringing investors into the infrastructure market in a new way that can be replicated in the future. The good quality ratings obtained for the transaction (over 95% of the issue was rated AAA) reflect the comparatively low risks of this type of project finance transaction: typically those that do not involve hi-tech operations or revenues dependent on markets such as those for telecommunications or power.

Although assembling the transaction was complicated, the real proof of whether EPIC will be a success or not will emerge over the next few years. The transaction was designed to be relatively flexible and to allow DEPFA to replace loans as they are refinanced. If replenishment of the pool proves to be impossible, then DEPFA's transaction costs will go up as the senior parts of the financing amortise and its attractions will diminish.

DEPFA's ultimate objective was to create an efficient refinancing template, which could be used repeatedly to refinance the infrastructure assets put onto DEPFA's balance sheet through its successful origination capability. Structurally, we have demonstrated this is possible. However, to generate a substantial improvement in the economic returns from this business sector, much will depend on the development of privately financed infrastructure programmes outside the UK and on the rating agencies' approach both to the individual loans and also to the risks posed by having more than one country involved.

Paul Leatherdale qualified as a Chartered Accountant before moving into banking where, for the last 15 years, he has specialised in project and construction financing. He manages DEPFA's global infrastructure business from Dublin, which in the last 5 years has committed over Eu5 billion to various PPP and concession-based infrastructure projects.

Andrew Bride, also a Chartered Accountant, is based in DEPFA's London office and heads the Infrastructure Finance Unit's activities in Central & Eastern Europe, in addition to managing the infrastructure asset securitisation program.