Direct government support

Large projects, particularly those beyond the current capacity of the bank market, require direct government support in order to be financed in current market conditions. Such support would take the form either of the government providing some form of guarantee to the project's debt financiers, or of it providing finance directly to PPP Co.

Several commentators have suggested that governments should provide a full financial guarantee of the project's debt, enabling access to a much wider pool of investors. This guarantee would be a contingent liability of the government, and so need not appear directly on its balance sheet. The problems with this approach are that:

•  it removes most of the discipline brought by private sector debt providers to the project, substantially reducing risk transfer to the project, and hence largely removing the rationale for undertaking the project as a PPP

•  once the government starts to provide guarantees, it will be difficult for it to stop providing them.

A few projects in the UK have had partial government guarantees (termed underpinning) of their debts. In these projects, the compensation paid by the government following early termination of the project for PPP Co default has a fixed minimum (for example, 95 percent of senior debt in the London Underground infrastructure projects, and a proposed 60 percent for the M25 project). However, too high a level still removes private sector discipline, whilst too low a level will be ineffective or even counterproductive (if lenders see the guarantee as being for less than their expected recovery following default termination). Under this structure, senior debt can be split into two tranches: the first fully guaranteed and the remainder not guaranteed, and thus effectively subordinated. This structure thus becomes similar to the Queensland supported debt model described below, and suffers from the same problems, albeit the guaranteed lenders have different interests to the government.

A better alternative is for the government to finance the project directly. In its simplest form, direct government finance could take the form of a direct capital grant to PPP Co to cover part of the capital cost of the project. Such grants have been used both in Australia and elsewhere for economic infrastructure projects, which otherwise would not have been commercially viable, though they have not generally been used for social infrastructure projects. There is a risk that such a grant either concentrates risk on the remaining private finance, making the project unfinanceable, or puts back substantial project risk to the government at the other extreme.

For example, lenders might assess a $1 billion project as having the risk of a $200 million loss due to higher than expected life cycle costs and performance deductions (borne by PPP Co). The finance for this project might normally be $100 million of equity and $900 million of debt. The potential loss to debt providers is 11 percent ($200 million loss less $100 million of equity, divided by $900 million). If the government provides a capital grant of $700 million, which replaces debt, with project risk allocation remaining the same, this potential loss increases to 50 percent. Consequently, the project may have become unfinanceable: at the least, the finance cost will have increased.

Alternatively, the government could take back some project risk such that the private finance for the project becomes $30 million of equity and $270 million of debt, with a potential loss of $60 million, leaving the potential loss to debt providers unchanged at 11 percent.

However, governments can avoid the problems of either higher finance costs or risk take-back by ensuring:

•  that the grant is for significantly less than 50 percent of the project's capital cost (which could result in a funding shortfall for the largest projects of over $1 billion)

•  in social infrastructure projects, that operational leverage is not increased, by making payment deductions for unavailability or poor performance proportionate to the amount of private finance rather than to the capital cost of the project (albeit diluting the risk transfer to the private sector).

If capital grants are used, they should be pre-agreed and based on the achievement of major milestones (such as the completion of discrete phases of the project), ensuring that the construction risk remains with PPP Co as much as possible.6

Alternatively, governments could provide a direct loan to PPP Co. In the UK, the government recently established its own virtual bank to provide such loans on a pari passu basis with senior debt. Such a loan would not affect the basic allocation of project risks, and would retain private sector financeability.

Problems with this approach are that:

•  private sector financiers may be concerned about the conflict of interest between the government as procuring authority and the government as lender

•  proper credit risk assessment and monitoring of the project would require experienced staff that the government may not have (the government should not just rely on other lenders' credit assessments)

•  depending on accounting treatment, the project's impact on the government's balance sheet may be doubled-up.

However, the government may be able to have its loan repaid by other financiers when market capacity increases in the future, reversing any adverse balance sheet impact.

This approach may be the only practical way to finance the largest PPP projects.

A variant approach is the supported debt model used for the recently closed South East Queensland Schools project. Under this approach, Queensland Treasury Corporation will refinance 70 percent of a project's financing following successful commissioning, charging a low margin over its own cost of finance (and hence being substantially cheaper than commercial banks). QTC chose the 70 percent figure as being the level where full recovery is highly likely should the project ever go into default. The remainder of the finance (debt and equity) is subordinated to the QTC debt. The private sector debt component thus bears substantially more project risk than normal senior debt. This model has additional problems to the UK model:

•  the private sector still needs to finance and bear the risk of the full project during its construction and commissioning phase, so market capacity remains a problem for larger projects

•  the subordination of the private sector debt after construction substantially limits the amount of finance available (some major banks are not willing to lend under this model)

•  the subordination leads to higher pricing of private sector debt, which may more than offset the benefit of the low cost of the QTC debt

•  the inter-creditor issues between QTC and the subordinated debt providers are highly complex (much more than if the government lends on a pari passu basis) and require extensive negotiation.

Another variant approach is the credit guaranteed finance (CGF) model tried in the UK in 2003/04. This model directly addresses the underlying problems with the current financial markets by separating funding from risk bearing. Under this model, the government provides the debt finance for the project, but in turn benefits from a full financial guarantee of its loan provided by either a syndicate of banks or a monoline insurer. The main argument for this approach was that it improved value for money (though the calculation ignored the guarantor credit risk borne by the government, so the VFM benefits may not have been present in reality). However, the UK Government only used the CGF model for two projects, as its accounting treatment doubled the project's impact on the government's balance sheet without any real offsetting benefits. This approach does little to address any shortfall in market capacity. Banks may be more willing to provide larger guarantees than loans, as they do not require any access to limited funding, but their regulatory capital requirements are the same, limiting any expansion in market capacity.

However, governments should consider a variant of the CGF model, which we have termed the counter-indemnity approach.7. This approach enables the project to access finance from the natural providers of long-term debt for PPP projects, the superannuation funds and life assurance companies.

Under this approach, the government effectively would guarantee payment of a substantial portion of the Service Charge in full, and repayment of debt in full on early termination of the project. The guaranteed portion of the service charge would be sufficient to cover all debt service payments of interest and principal. Loans made against this revenue stream therefore would effectively be government risk. Any deductions from the service charge for poor performance or unavailability would first be made against the remaining portion of the service charge (notionally allocated to operating and maintenance costs and to equity returns). Any deductions greater than this portion of the service charge would be covered by a full counter-indemnity to the government in the form of a guarantee of all project risks by a club of banks, a monoline insurer, or some other creditworthy vehicle established for the purpose. PPP Co would support this guarantee by cash collateral. The guarantee would also cover the full amount of outstanding debt following early termination of the project (netted-off against any compensation due to PPP Co from the government). In the unlikely event of early termination of the project for PPP Co default, the government would repay the lenders in full and reclaim the difference between that amount and the compensation it would have paid under the standard PPP model from PPP Co, the cash collateral and the guarantors.

The success of this approach relies on several factors, shown in Table 3.

This model could include a government loan for part of the finance, should market capacity be insufficient to finance the full project. However, such a loan may double-up the project's impact on the government's balance sheet. Another possible problem is that, even if market capacity for the project guarantees is higher than that for project loans, it is still insufficient to cover the full project. The government might then choose to accept only a guarantee for a proportion of the project's value that, in practice, would only result in a residual risk to government in the event of early default termination of the project. However, the guarantors would still bear full project risk, which would lead to higher pricing and might limit market capacity.

The Dutch and French governments have used a variant approach, under which they make a large proportion (60-70 percent) of the Service Charge not subject to any form of deductions following successful commissioning of the projects. As a result, the private debt finance for the project is split into two components: a tranche secured on that part of the Service Charge that is effectively government-guaranteed, and a subordinated tranche onto which project risk is concentrated. This approach in essence is similar to both the Queensland supported debt model and the underpinning model, and suffers from much the same problems, namely market capacity, pricing and inter-creditor issues; though there is no risk of a double impact on the government's balance sheet.

__________________________________________________________________________________

6  As a result, project documentation will become more complex, needing to define each milestone and its "completion' requirements.

7  This approach draws on a concept originally developed by ABN Amro Australia, with a contribution from Partnerships Victoria.