Jim and Margaret Cuthbert show how much of the PFI debt has been hidden

There is a massive hole in the public finances, but it’s not the one politicians are talking about. A typical PFI project involves a long term (25 or 30 year) contract between the public sector and a private consortium, for the provision of a serviced asset – like a school or hospital. The public sector makes a regular payment, known as the unitary charge, throughout the operational phase of the contract. This payment covers both the use of the capital asset, and the provision of the services specified in the contract. When it comes to the government accounting for PFI schemes, the key question is – should the capital asset involved in any particular scheme appear on the books of the public sector? In fact, there are actually two different sets of government accounts in which a given PFI asset might appear, and hence two different on/off book decisions to be made. First of all, should the asset appear in the accounts of the specific government department involved: and secondly, should it appear in the national accounts compiled by the Office for National Statistics, (ONS). Both of these sets of accounts are very important – for different reasons. Departmental level is where expenditure control actually bites: a project will not go ahead unless there is sufficient provision in the budget of the relevant department. The national accounts matter because they are the key focus for determining the overall sustainability of the public finances.

A bizarre and little appreciated point is that in the UK these two different on/off book decisions are in fact governed by different accounting standards. Departmental accounts are governed by rules for government financial reporting, designed to be consistent with International Accounting Standards and International Capital Reporting Standards. National accounts, however, are governed by the rules of the UN System of National Accounts and the European System of Accounts, as interpreted by ONS. The two sets of standards are significantly different: national accounting standards generally take a more restrictive view of what constitutes a liability for the public sector. However, up until 2009, the difference between the two sets of standards did not matter in practical terms as regards PFI: both standards led essentially to the same test for a PFI scheme, based on the degree of risk transferred to the private sector. If sufficient of the risk involved in the project was transferred to the private sector, then the capital asset associated with the scheme would appear neither in the relevant departmental account, nor as a government asset in the national accounts: that is, the asset would be off book as regards both sets of accounts.

From the early days of PFI, the government’s attitude was perfectly clear: PFI schemes should be designed so that they were ‘off the books’. There were a number of apparent advantages to this. Departments were able to get the benefits of new capital expenditure without breaching their capital expenditure control limits: and government was able to set about renewing the infrastructure of the state without adverse effects on measures of fiscal sustainability. Of course, these particular advantages are more presentational than real. But nevertheless, those involved in capital procurement in the public sector knew that PFI schemes were very unlikely to be approved unless they were ‘off the books’. So government set about designing PFI schemes which would pass the off-book test. This involved, for example, developing schemes where the provision of the capital asset was inseparably bound up with delivery of associated services. These schemes were known as ‘non-separable’ schemes: and the point of non-separability was that the external auditor who was charged with classifying the scheme as on or off the books, would not be able to regard the delivery of the capital asset as a self standing project on its own – and so would be much less likely to class the asset as ‘on book’.

It is at this point that an important boundary is crossed, where accounting treatment starts to affect the real world. The need for PFI schemes to be non-separable has a number of adverse consequences. Non-separable schemes are, by definition, complex, and probably large – so reducing the number of firms which can compete for such projects, and hence reducing the competitiveness of the market. Moreover, complex, non- separable contracts are inherently more difficult for the public sector client to scrutinise effectively. Both aspects, therefore, are likely to reduce value for money for the public sector. In addition, the large size of non-separable projects reduces the number of local firms who can compete – with adverse effects for the local economy. In terms of getting schemes off the books, the government’s strategy was very successful. Almost all PFI schemes were initially classed by auditors as off book: and while some of these judgements were later reversed on more detailed consideration, the upshot was that a large majority of schemes remained off book. As a result, of the more than £60 billion of capital assets in signed PFI deals by 2009, only £5 billion of this was reflected in departmental accounts, or in the public sector net debt in the national accounts. This is slightly to understate the eventual effect on public sector net debt, because there are timing effects at work here as well: schemes don’t come into the accounts even if classed as on book until construction is completed. Nevertheless, it is clear that, under the present accounting treatment, PFI has apparently had a minimal effect in increasing the overall financial liability of the public sector: the important word here, of course, is ‘apparently’.

It rapidly became clear that there were nonsensical aspects to this PFI accounting regime. For one thing, while most PFI schemes were off the government’s books, many schemes were also classed by the private sector operators as being off their books too, since there were tax advantages in doing this. As a result, the capital assets of a large number of schemes were being accounted for in the books of neither the public nor private sectors. Further, it also became clear that the assessments of risk transfer which were being made were, in many cases, highly questionable. For example, it was very suspicious that the amount of risk transferred often turned out to be just sufficient to make the PFI option marginally cheaper than the public sector comparator. Following criticism along these lines, there was a general welcome for the announcement made in the March 2007 budget that government would be altering the way in which departmental accounts are compiled. Technically, what the government announced was that it was moving to International Financial Reporting Standards as a basis for compiling departmental accounts, rather than the approach laid down by International Accounting Standards. For PFI schemes, this meant that the old risk-based test would be replaced by another test, under which the capital asset would come on- book if either the public sector retained a substantial right in the residual asset at the end of the concession period, or if the public sector controlled the terms on which the service associated with the PFI scheme was delivered to the public. Since most PFI schemes will satisfy both of these criteria, it was clear that the adoption of the new approach as from 2009 would bring almost all PFI schemes on-book as regards departmental accounting.

It was, however, premature to assume that this change was going to solve any of the substantive problems surrounding PFI accounting. First of all, as we have explained, departmental accounts are compiled to different standards than those ONS uses in compiling the national accounts. ONS quickly made it clear that the adoption of a new approach for departmental accounting had no implications for their handling of the national accounts – and that the new approach would not result in any greater number of PFI assets being included in the national accounts. Even at the departmental level, the Treasury announced in 2009 that it was breaking the link between departmental accounting and budgeting. Henceforth departments would have to keep two sets of books: for the purpose of producing their annual accounts, PFI assets would indeed be included, but as regards capital controls and budgets, the old risk-based test for PFI assets should continue: departments therefore have the same incentive to classify schemes as off book on the risk- based test, in order to avoid capital budget constraints. So, in fact, the widely heralded change announced in 2007 has solved none of the problems with PFI accounting. What has happened is a classic example of government having its cake and eating it: that is, professing the highest accounting standards, while acting in such a way that the effect of the standards is actually circumvented. It is interesting to note that the House of Lords Select Committee on Economic Affairs, which could not in any sense be described as an anti-PFI body, was nevertheless heavily critical of the way departments will in future have to run two sets of books for PFI.

We now turn to another important area of PFI accounting. When a PFI asset is accepted on the books, there also has to be a measure of the liability which the public sector has taken on to pay for that asset: how should this liability be assessed? At present, what effectively happens for on-book schemes, in both departmental and national accounts, is that the public sector accepts a liability equal to the capital value of the asset. But suppose that the stream of payments which the government had contracted to pay for the availability of the capital asset was actually worth more than the value of the asset? In these circumstances, counting just the capital value as the liability would understate the actual extent of the public sector’s liability. At this point, we are entering into an area which can only be examined by looking at how PFI schemes behave in practice: and it is also an area where we ourselves have carried out some relevant research. Detailed information on how PFI schemes behave is very difficult to obtain because it is classed as commercial in confidence. We were fortunate to obtain, by means of Freedom of Information, the detailed financial projections produced by the operating consortia for eight PFI projects at the time when the final contracts for the schemes were signed. (We are grateful to Unison for obtaining some of these projections). One of the things we were able to do with these detailed projections was to split down the stream of unitary charge payments into two components: one covering the cost of the services which will be provided as part of the PFI contract: and the other being essentially the payments which will fund the capital used in the project. It is this latter stream of payments which represents the liability which the public sector has undertaken in order to secure the availability of the capital asset. The problem is – how should this stream of payments be converted into a single figure representing the cost to the public sector? The appropriate way to do this is to calculate how much the public sector could have borrowed, for the same cost as the stream of payments, if it had gone down the normal public sector route of borrowing from the National Loan Fund. When we did the appropriate calculations, the results were fascinating. In each of the eight cases for which we have data, we found that the cost of the liability the public sector was taking on is much greater than the cost of the capital asset. In fact, in six of the eight cases, the liability was more than one and a half times the cost of the capital asset: and in three cases, the liability was effectively twice the cost of the capital asset.

These results do not mean that PFI is necessarily twice as expensive as public procurement: the public is also getting an element of risk transfer for the stream of capital payments. But for present purposes, what matters is that the liability being taken on by the public sector on PFI contracts is clearly commonly very much larger than the cost of the capital asset being provided. So even for on book schemes, which do feature in the national accounts, these schemes appear in a way which grossly understates the relevant liability. . (For the avoidance of doubt, we should stress the point that the liability we are talking about here relates solely to the provision of the capital asset: we have taken payments for services right out of our calculation. There are, of course, significant contractual liabilities attaching to future PFI service provision too: but that is not the subject of this article.)

The PFI projection data we have obtained also shows something else very relevant: namely, just how profitable PFI schemes can be for those providing the equity component. In all of the eight schemes, the annual return on the combined input of subordinate debt and pure equity was 15% or more: and this was commonly earned on an average outstanding debt which was more than twice the capital actually invested. The relevance of this is that the equity capital providers are the primary risk takers: and if their returns are very large, the extent to which they can be said to be truly bearing risk is very limited. The phrase “having a flutter with public money” comes to mind, rather than meaningful risk transfer. The implication is that the extent of projected profit should be taken into account in the risk transfer test: in which case, many more PFI schemes would presumably come on book in national accounts terms.

Overall, what we have here is a sorry tale. Government, while professing adherence to the highest international accounting standards, has so managed things that the public finance liability for PFI assets has been grossly understated. Far from the trivial £5 billion PFI liability which currently appears, a reasonable view of the actual liability would be greater than £60 billion – perhaps much greater. The warning signal, which should have indicated the extent of this deepening liability as it was incurred, has been over-ridden.