1 We reviewed a representative sample of 15 project savings from HM Treasury's dataset of £1.6 billion of reported savings. We selected the six largest savings and then a further nine savings randomly selected using a monetary unit sampling technique to get good coverage by department and type of saving. Our sample represents 86 per cent of savings by value and 25 per cent of the number of projects reporting signed savings. We excluded the smallest savings from the sample population before sampling to avoid the selection of small savings, which would incur an excessive audit burden.
2 We reviewed the evidence underpinning each reported saving, such as signed contract variations, benchmarking and other calculations. We assessed the assumptions made and the level of prudence exercised to assess whether the reported saving was sufficiently supported. We also reviewed evidence relating to the maintenance of frontline services to distinguish between 'cash-releasing efficiency savings, 'other-efficiency savings' which have an unquantifiable cash benefit and 'cash-releasing scope reductions.
3 Some authorities have reported a number of different types of saving in a package. This form of reporting has made it difficult for the Treasury to identify the type of savings clearly. We have had to make some assumptions about which of our three types of saving categories is the most relevant where packages of savings have been reported.
4 We conducted semi-structured interviews with representatives of the authorities reporting the six largest savings and four of the smaller savings in our sample, to assess whether they had identified the risks to the valuation of their saving. We satisfied ourselves that these risks were considered in financial modelling and brought to the attention of the competent decision-making body. We did not form our own view on the value for money of accepting these risks because to do so would have required extensive fieldwork and analysis beyond the scope of our work.
5 The level of evidence to support each saving varied as the savings are not all cash-releasing, most have not yet been fully realised and some are based on estimates.
6 We categorised savings into one of three assurance categories to indicate the level of confidence we had in the reported savings:
• Sufficient evidence. The calculations for the saving are clear, the evidence base is available, any assumptions made are realistic, measurement uses strong benchmarks (e.g. recent, actual cost data) and sufficient prudence has been exercised.
• Partial evidence. The saving has some uncertainty or the evidence base is not clear. For example if there is no recent or actual cost information to draw on, a best estimate of future avoided cost has been used.
• Insufficient evidence. The saving is not supported by documentary evidence, or the estimates or assumptions are unrealistic or contested.
7 Because the savings we looked at were largely forecast savings rather than realised savings, our review does not do the following:
• Provide assurance that the savings reported will in fact be realised because the savings are, in the majority of cases, forecasts and are therefore subject to uncertainty.
• Confirm that frontline services have not been affected by the savings. Our work was limited to interviews and reviewing documentary evidence about predicted impact on frontline service delivery. Our work gives no assurance as to the actual impact on service delivery of these changes.
• Some risks that transfer to authorities as a result of these savings may materialise and may increase authorities' costs. We have assured ourselves that authorities have considered the risks associated with these savings and their potential effects, but we have not audited the authorities' governance or management of these risks and their associated costs.
• We did not examine the impact of departmental reviews of contracts prior to signature, which in some cases have resulted in better value contracts.