5. When assessing the value for money of a PFI deal departments are advised to produce public sector comparators (PSCs) to estimate the cost of an equivalent conventionally procured project. In this procurement MOD's final PSC showed a central estimate of £746.2 million. This compared to a deal cost of £746.1 million.2
6. MOD considered that the closeness of the comparison was due to the fact that in the final negotiations it only agreed to close the deal when Modus had dropped its price to a level which was less than the PSC. Modus had not known the PSC cost estimate and MOD considered it had got the maximum price reduction that could be negotiated.3 Even where there may be valid reasons for a PFI price being marginally less than a PSC there is still a risk that presenting figures with a high degree of accuracy, such that minor cost differences on large value projects appear important, may lead to a misinterpretation of the financial comparison. The appropriate conclusion is that the two routes are similar in cost terms. MOD considered that non quantified factors tipped the balance if favour of the PFI deal. These were the fixed price of the contract, the risk of overruns being passed to the contractor, incentives for the project to come in on time, the fact that the contractors' money is put at risk, and the incentive scheme for the contractor to perform.4
7. The inherent uncertainties in a financial comparison of this kind are highlighted by this deal. The MOD's central estimate of the PSC cost of £746.2 million included an adjustment for risk of £102.9 million or 17.2% (Figure 1). It is right to take differences in risk into account in these comparisons since it is not possible to predict conventional procurement costs with absolute accuracy. The outcome of the financial comparison was, however, very sensitive to the costing of risk. Figure 2 shows that the decision on the extent of the risk adjustment for capital expenditure had a significant effect on the financial comparison. For example, a change in the risk adjustment for capital expenditure from MOD'S assumption of 29.5% to 29% reduces the PSC by £1 million. MOD's own calculations showed that, depending on the assumptions that were made for all the various risks, the costs of conventional procurement could fall anywhere within the range of £690 million to £807 million.5
8. MOD considered that its risk assessments were not excessive. It noted that its capital expenditure risk adjustment of 29.5% was lower than the risk factor of 35% that Treasury guidance at the time suggested should be applied to construction costs in conventionally procured projects.6 A more recent study of public building projects commissioned by the Treasury estimated a range for construction cost overruns of between 2 and 24% in standard building projects and between 4 and 51% in those involving non standard buildings.7 The extent of these ranges highlights how difficult it is to predict with certainty what the costs for a conventionally procured project might have been.8
Figure 1: The effect of risk assumptions on the main components of MOD's PSC
NPV (£ million at 2000 prices) Base costs | Base costs | Risk | Risk as % of base costs |
| £ million | £ million |
|
Capital Expenditure | 208.6 | 61.5 | 29.5 |
Capital replacements during contract period | 95.2 | 22.2 | 23.3 |
Operating costs | 202.4 | 8.0 | 4.0 |
Rent, contribution in lieu of rates, and leasehold risks | 133.1 | 0.0 |
|
Other | -2.5 | 17.7 |
|
Total | 639.9 | 109.3 | 17.2 |
Total PSC (base costs and risk adjustments) |
|
|
|
Source: C&AG 's Report (Figure 9, p24)
Figure 2: The financial comparison with different risk adjustments for capital expenditure
Capital cost risk factor | Equivalent PSC cost |
35% | £757.7 m |
29.5% (MOD's central assumption) | £746.2 m |
PFI price | £746.1 m |
29% | £745.2 m |
25% | £736.9 m |
20% | £726.4 m |
Source: National Audit Office9
9. Given these uncertainties in estimating what a project might have cost under conventional procurement, a precisely calculated PSC should not be the main basis for making decisions on value for money. MOD told us that it had to abide by the rules set down by the Treasury that required a PSC using these techniques.10 We note that the Treasury's guidance specifically warns against the pursuit of spurious precision, and that it describes the PSC as an aid to judgement and not a pass or fail test.11
10. The PSC was used by MOD as a negotiating tool to reduce the deal price by £4 million on the day of financial close. Knowing that the price had become higher than the PSC, MOD told Modus that MOD could not sign the deal unless the price was reduced. Modus did not know how much it needed to reduce its price by in order to offer a price lower than the PSC. MOD had given Modus certain information on base costs but had not disclosed the full PSC including risk adjustments. Modus eventually reduced its bid price to £100,000 below the PSC. MOD would still have signed the deal had the price been above the PSC because of the fixed price and other benefits the deal provided. MOD would be prepared to walk away from a project but would not normally do so on a PFI project once the initial evaluation and approval had been completed. It had only once walked away from a PFI project after it had reached the preferred bidder stage.12
11. In all deals there is a risk the price will increase if the deal is not closed quickly once the preferred bidder has been appointed following the bidding competition. In this deal MOD appointed Modus as preferred bidder in January 1999 after Modus had bid a proposed contract price of £647 million. It took MOD 16 months to finalise the contract. By the time the contract was let in May 2000 the deal price had increased by £99 million (15%). This was mainly due to increased financing costs and additional work required to Main Building that was only discovered during detailed surveys undertaken after Modus became preferred bidder. At the time it appointed Modus preferred bidder MOD had been expecting savings of £25 million compared with its PSC estimate of the cost of a conventionally procured project. These savings were not realised. It was only through the final negotiations on the day of closing the deal that MOD avoided the deal price being higher than its PSC.13
12. £60 million of the £99 million price increase was due to increases in the cost of financing the project. These were mainly due to interest rate increases and other movements in the financial markets (including a loss of income as an initial assumption that surplus funds could be invested at a profit could not be realised)14 (Figure 3). These factors caused an increase of close to 50% to the financing costs to £185 million.15 As is normal in PFI contracts the financing cost risk remained with the public sector. MOD was exposed to this risk during the 16 months it took to close the deal with Modus. The increase in financing costs added 10% to Modus's bid price. Commercial companies often employ hedging strategies when closing such large deals to avoid being exposed to such movements in financial markets. MOD did not hedge against movements in the financing markets as this is not government policy, but had also been advised that it would not have been possible to hedge the financing risk as there was no market in hedging instruments for a deal of this size at the time.16
Figure 3 Reasons for increase in financing costs
| £m |
Increase in long term borrowing rates | 27 |
Assumption that surplus funds could be invested at a profit not realised | 25 |
Other factors (note 1) | 8 |
| 60 |
Note 1 : Other factors includes £2 million in respect of debt funding for additional work, £3 million additional equity costs and £3 million additional costs related to working capital and reserves. | |
Source: National Audit Office17
13. MOD was exposed to increases in long term borrowing rates which occurred during the 16 months it took to close the deal. Part of this increase was due to the fact that MOD remained committed to a decision taken five months into this period that the deal should be bank financed. Long term borrowing rates in the bank financing markets then became relatively more expensive compared to those in the bond markets during the remaining eleven months MOD required to close the deal.18
14. MOD had asked its final bidders to provide two bids, one based on bank finance and one based on bond finance. At this stage, Modus's bond financed solution was expected to be £25 million cheaper than its bank financed solution (mainly due to an assumption that surplus funds from a bond issue could be invested, short term, at a profit). Modus kept the two financing routes open, but was concerned about the high costs of doing so, and in June 1999 pressed MOD for a decision on the method of financing. MOD considered it important to balance the cost of keeping the two options open against the potential benefit of having financing alternatives available. It did not, however, calculate what the costs and benefits might have been of keeping both financing options open.19
15. Movements in the markets meant that the expected difference in cost terms between bank and bond finance became marginal. Modus proposed that bank finance should be used to finance the deal. MOD considered that, in the absence of a clear cost difference, there were qualitative reasons for choosing bank finance. These included: greater flexibility to cope with any contract variations that would require changes to the financing; lower costs to MOD in the event of an early termination of the contract; concerns about the security issues attached to the disclosure levels of a public bond; and that the underwriters of a bond might not accept Modus being exposed to the same level of penalties for poor performance as could be applied in a bank financed deal.20
16. In the subsequent eleven months up to the deal's financial close the markets moved so that the cost of bond financing became again generally favourable compared to bank financing. MOD stayed with its decision that bank finance should be used, however, as it could not see a clear case for change. It considered the movements in the relative costs of bank and bond finance were volatile, so any cost differential in favour of bond finance could not be certain to exist at financial close. It did not wish to delay the deal by changing the financing arrangements and it continued to prefer bank finance on qualitative grounds. There is evidence to suggest that when the deal was closed in May 2000 bond finance may well have been cheaper. Based on other bond financed PFI building projects signed around the same time, if a bond had been issued to finance the MOD deal the financing might have been up to £22 million cheaper.21
17. Given the fluctuations that can occur in financing rates it makes sense to make a final decision on financing as late as possible. MOD told us that, in the same circumstances, it would in future run a funding competition before closing the deal, like that used in the Treasury Building project. It considered that in 2000 it would, however, have been wrong to trial the use of a funding competition, which had not then been used before, on the large MOD Building deal. Since the financing markets had moved in favour of bond finance, and bond finance had been actively considered by MOD earlier in the procurement, the bond option might usefully have been reassessed before this deal was closed.22
18. In a bond financed project funds are drawn down from the bond issue at the beginning of the deal and are placed on deposit to earn interest until they are required. Modus's bid assumed that, based on interest rates ruling at that time, monies from a bond issue could be placed on short term deposit at rates that would be higher than the long term interest rate payable to bondholders. In the event this benefit was not realised as short term interest rates decreased and MOD chose bank finance where funds are drawn down as and when required reducing the likelihood of there being surplus monies to place on deposit. These factors accounted for £24 million of the increase in financing costs after Modus became preferred bidder. At the time Modus was appointed preferred bidder MOD estimated that Modus's bid would achieve savings of £25 million compared with MOD's PSC estimate of the cost of conventional procurement. These estimated savings, which are not now expected to arise from the deal, were therefore largely based on the speculative assumption that surplus funds from a bond issue could be invested at a profit.23
19. On the day of financial close, 4 May 2000, bank rates increased because the market knew that the MOD deal was coming.24 MOD raised all the bank finance for the deal, some £500 million, on the day of financial close. In addition there had been a number of other large PFI projects that came to the market around that time. These factors increased the price that both MOD and other departments were charged for finance at this time. MOD agreed it would have been sensible if the large projects had sought finance at different times. To that end, OGC is now co-ordinating the PFI activities of different departments.25
20. When MOD chose Modus as the preferred bidder in December 1998, bidders had only had access to high level surveys undertaken by MOD into the condition of the building.28 MOD decided to wait until the preferred bidder stage to allow more detailed surveys of the building as it considered these would be unacceptably intrusive on the working environment, and would increase the bid costs if carried out before the selection of preferred bidder. Modem, the other bidder, had already threatened to pull out of the deal due to the cost of bidding. By deferring the detailed survey work until the preferred bidder stage, MOD was, however, exposing itself to the risk that further work would be required as a result of the surveys, which would be priced without competitive tension.26
21. In the event, the detailed survey programme revealed that the building was in a worse condition than Modus had assumed in its bid. The detailed surveys identified problems with the water supply system, the quality of the roof and asbestos in the building. Modus asked for an additional £60 million to cover additional work to deal with these problems. To counter the lack of competitive tension in the pricing of this additional work MOD used its advisers, Bernard Williams Associates, to cost each element of the additional work. MOD then negotiated with Modus and reduced the price increase Modus was seeking from £60 million to £37 million. MOD considered the £37 million price increase for the additional work was fair. Nevertheless, if MOD had arranged these detailed surveys during the final bidding round the work would have been priced competitively and the period during which it was exposed to variations in financing costs would have been reduced.27
22. As the deal is funded by bank finance, there is the possibility that it will be refinanced at some time in the future. As our predecessors found, refinancing can significantly increase the contractors' returns from a PFI project.28 MOD does not have a specific contractual right in this deal to share any gain that might arise from a refinancing. However, Modus agreed with MOD that if Modus wanted to change its financing arrangements to effect a refinancing it would have to seek MOD's approval, thus enabling MOD to negotiate a share of the refinancing gains.
_____________________________________________________________________________________________________
2 C&AG's Report, para 2.48. The total cash cost of the PFI deal is around £2.4 billion (Qq 88-89)
3 Ibid, paras 2.59-2.60; Qq 16-18, 75-81, 159, 162-163
4 C&AG's Report, para 15, Table 1, p 7; Q 2
5 C&AG's Report, para 2.47; Q 76
6 Qq 86-87, 124; Ev 22, para 13
7 Ev 22, para 13
8 C&AG's Report, paras 2.49-2.50; Qq 123-125
9 Ev22, para 13
10 Qq 86-87
11 Treasury Task Force Private Finance, Technical Note No 5, How to construct a Public Sector Comparator, para 2.2.3 and p 67
12 C&AG's Report, paras 2.58-2.60; Qq 77-80, 144-145, 160-161 (and footnote), 176-181
13 C&AG's Report, paras 2.24, 2.58, Figure 12, p 27; Qq 5-6, 75-79
14 C&AG's Report, para 2.24; Ev 22, paras 9-12
15 discounted at the Treasury discount rate of 6% real (Q 109)
16 Qq 57, 88-96 (and footnote)
18 C&AG's Report, paras 2.31-2.39; Ev 22, para 10
20 C&AG's Report, para 2.32; Qq 66-67, 69-70, 74, 88-89
21 C&AG's Report, paras 2.40, 2.42, 2.44. The possible savings are based on the bonds used to finance the GCHQ and Treasury Building deals in June and May 2000 respectively.
25 C&AG's Report, Figure 8, p 22; Qq 19-23, 44, 122
27 C&AG's Report, para 2.2.5; Qq 167-175
28 13th Report from the Committee of Public Accounts, The refinancing of the Fazakerley PFI prison contract (HC 995-I, Session 1999-2000)