The risk referred to is that of overruns on construction cost as a result of what is commonly called 'optimism bias'. Supporters of PFI projects emphasise that publicly-financed projects tend to suffer considerably from 'optimism bias' and that the projects invariably cost more and take longer to build than budgeted. Thus, in some notes on the Private Finance Initiative (PFI), the Norfolk and Norwich University Hospital (NNUH) Trust contrasted the building of the Scottish Parliament with that of the NNUH. The Trust stated that the Scottish Parliament was completed some three years behind schedule and ten times over budget whereas the NNUH was delivered 20 weeks ahead of schedule and on budget (NNUHT, September 2004,1). However this is a highly misleading comparison since the Scottish Parliament was a one-off project whereas the NNUH was, and is a fairly standard District General Hospital (DGH).
So what allowance for optimism bias is reasonable for a fairly standard DGH? In its notes on the PFI, the Trust referred to a report carried out for the Treasury by Mott MacDonald, a firm of consulting engineers. The Trust correctly pointed out that Mott MacDonald had reported that publicly-financed projects tended to underestimate costs and project duration. However what the NNUH Trust did not point out is that the Mott MacDonald report, produced in 2002, also made the following important points;
• the degree of optimism bias varied enormously between different projects and different types of projects (Mott MacDonald, 2002, table 2);
• lessons were being learned by the public sector from past projects and improvements had been made in the estimation of project costs and timings with the result that optimism bias was getting smaller and smaller (page 21);
• the degree of optimism bias changed over the stages of the project from the outline business case to the contract award (page 15).
Given these three points, it is hardly surprising that the optimism bias varied enormously between the various studies carried out for the Government (see Appendix G, Mott MacDonald 2002). For standard buildings (such as District General Hospitals) it is worth noting that the 'optimism bias' guidelines given by Mott MacDonald were between 2% and 24%, the average being 13% (Mott MacDonald 2002, page S-2). However a study by Pollock, Price and Player in 2007 has been highly critical of the Mott MacDonald study, arguing that it was deficient in its sampling techniques and that it contained serious measurement bias (Pollock A et al, 2007 pages 6 and 7). Furthermore in 1999, the House of Commons Public Accounts Committee had given lower rates for cost overruns stating that between 1988 and 1991, the cost overrun on major capital projects in the NHS was averaging about 13 per cent and that this had since declined to about 7 per cent (quoted in Sussex 2003, 65). And as pointed out in chapter 5 above, in 1999, Gaffney et al pointed out the "average increase in cost over approved tender sums for NHS capital projects has been between 6.3% and 8.4% in the 1990s" (Gaffney et al 1999). As Sussex 2001 put it; "The myth of continuing rampant cost and time overruns on conventionally procured major hospital projects is just that; a myth" (page 50)
But even if we are generous to the PFI supporters and assume an average cost overrun (as in the Mott MacDonald report) of 13 per cent, this is much, much smaller than the excess capital+financing cost of the private sector implied by the Andersen model. The latter, as have seen, is over 60% which is a very high price to pay to avoid an average 'optimism bias' of 13%.