35 The financial model used for the Comparators produces a range of possible values. The range of the combined Comparators for all three infrastructure contracts is several billion pounds6 between the lowest and highest values. The model also produces single expected values, but we, London Underground and its advisers have concluded that single values should never be relied on in a financial analysis, especially given the particular characteristics of this exercise.
36 As noted in paragraphs 18 to 22, the model depends on fixed inputs of base costs, which are multiplied by inputs of plausible ranges for risk and efficiency factors. These input ranges are used to generate several hundred possible values for the Comparators, from which the output ranges have been generated. The risk and efficiency factors are therefore the drivers of the ranges of values, and reflect London Underground's best judgements about the uncertainties facing a public sector infrastructure operation.
37 The value of each uncertainty factor depends on the elicitation of upper and lower values from engineering experts, as described in paragraphs 18 to 22 above. For example, for signalling investment, engineers anticipated an upper level of overrun of 27 per cent and a lower level of 13 per cent. In the absence of any clearer information, London Underground assumed that the most likely value was the mid-point of the upper and lower range (for example, 20 per cent in the case of signalling investment). As a result of using that assumption, the financial model might be taken to suggest that the most likely value of the public sector costs lies in the middle of the ranges which the model produces. That conclusion would be implausible. As is well known, major projects sometimes are delivered below budget, but very rarely by a wide margin below. But major projects do sometimes overrun their budgets very considerably. That should mean that the most likely outcome would be in the lower half of the range and not in the middle as in this case.
38 Furthermore, the modelling process appears to have considered uncertainty factors on an individual, line-by-line basis, rather than an holistic basis. Because this comprehensive approach has been used, a large number of variables over 30 years have been input into the model. In such circumstances, it is difficult to be sure that there has been no double-counting of uncertainty in different factors. A smaller model, with fewer key variables, might have made possible a more integrated analysis of the risks facing London Underground.
39 As a result of these issues, we and London Underground have concluded that the modelling provides some useful information about the upper and lower bounds of public sector costs over the next thirty years but cannot reliably be taken to produce a single expected value within those boundaries.
_____________________________________________________________________________
6 For reasons explained in paragraph 6, we have not given absolute values. The range referred to here is the range of net present values at a real discount rate of 6 per cent over 30 years.