Social security agency in Scotland: outline business case

5. Financial Case

5. Each of the six options outlined in section B provide the functions needed to administer social security in Scotland under different delivery methods in terms of who delivers and how. This is designed to provide the scope to answer the questions in 1.5 to describe how the social security system operates.

5.1. It is important to remain aware of distinction between the Agency and the System: the agency is the means by which the Scottish Social Security system will be delivered in Scotland. How the overall system will operate is under development after the Public Consultation and the agency, in whichever form it takes, will administer the system.

5.2. This financial case does not contain information on implementation costs nor on transition to full operation. This is because these costs depend on detailed design work that has yet to be undertaken and is so complex it cannot be undertaken until the form of the Agency is determined. However, section 5.7 considers how implementation costs will vary across the options.

5.3. Instead the question that this part of the Outline Business Case addresses is what the on-going commitment that the Scottish Government would be placing on the Scottish Budget under each of the options. However, as demonstrated by Section 5.6, these "steady-state" costs are sensitive to many policy decisions that have not yet been made. As such this OBC does not represent a bid for future budgets but is the evidence base for making a choice between options that represents value for money.

5.4. In terms of the steady state costs, the approach to IT and Digital is different to that used to cost other aspects and is subject to greater variation and uncertainty. The costs are presented on a consistent basis in terms of being point estimates with ranges but the treatment of risk and uncertainty for IT and Digital cost is different and there is a separate analysis in section 5.5 of this. It is crucial to note that the IT and Digital costs should be specifically considered only in terms of the ranges discussed in this analysis.

5.5. The preferred option that forms the value-for-money solution taking into account the advantages and disadvantages of the different options is discussed in section D.

5.6. This part of the OBC contains sensitive information that could negatively impact the value for money achievable through the subsequent procurement processes be they for buildings or IT systems. As discussed in the Foreword, for this reason some financial information, particularly around the breakdown of costs is not included in this document.

5.1 Key assumptions

5.7. In order to provide an informed body of evidence to support the decision of what form the agency will take despite the inherent strategic uncertainty, and to allow a meaningful comparison of the different options, certain assumptions underpinning the presentation of the agency costs have been made:

• A level of activity in 2021-22 which is all benefits administered, and a forecast level of caseload (both new cases per year, maintenance of existing cases). See Annex C for further detail
• The modelling approach has been to analyse the relevant parts of the DWP operational costs structure and use that as a basis to construct an estimate of the costs of administering the Scottish system

5.1.1 Funding

5.8. The establishment and running costs of the agency will be funded from the Scottish Budget. As a result of the Report of the Smith Commission in 2014, HM Treasury provided the Scottish government with funding to implement the newly-devolved powers.

5.9. The Fiscal Framework explains how resources for social security will be transferred ^[32] from UKG: See Figure 12 below.

Figure 12 - Fiscal framework box-out

30. There are administration and implementation costs associated with the powers being devolved. In line with the Smith Commission recommendations the UK government will transfer funding to support a share of the associated implementation and running costs for the functions being devolved.

31. Both Governments have agreed that the UK government will provide £200m to the Scottish Government to support the implementation of new powers. This will represent a one-off (non-baselined) transfer, supplementing the block grant, to support the functions being transferred. The profile of this transfer is to be agreed by the JEC.

32. The Governments have agreed a baseline transfer of £66m to cover the on-going administration costs associated with the new powers. This figure includes the marginal savings realised by the UK government as a result of no longer administering the powers in Scotland devolved under the current Scotland Bill, plus a share of the Scottish Government's running costs. This baseline transfer will be indexed through the normal application of the Barnett formula.

5.10. The steady state running costs presented later in this section show a draw on the Scottish Budget that is greater than the amount being transferred under the Fiscal Framework. The amount to be transferred for administration costs represents a share of administration costs rather than the full value of administering the devolved benefits in Scotland. Management of the total funding required for the steady state running costs will be the subject of future decisions through the Scottish Budget.

5.1.2 Capital Requirement

5.11. The costs involved in setting up the agency will be scored as Capital and Resource DEL. The split will broadly be determined by the nature of the implementation costs and the procurement route.

5.12. The establishment of the agency is not a large capital project compared to others such as those in the infrastructure or health sectors, but there will be capital expenditure on fixed assets such as IT and buildings. These decisions will be made at a later stage.

5.1.3 Net Effect on Prices

5.13. There is no effect on prices (no inflation as a result of the spend) because the scale of the capital and on-going expenditure on non-staff costs is relatively small compared to the wider Scottish Economy. The exception is the potential impact on specialist staff costs around assessments. The impact will be dependent on the model of assessment that is chosen. This is discussed in Section D.

5.1.4 Impact on balance sheet

5.14. The impact on the public sector's balance sheet of setting up the agency will depend on the procurement route selected for the acquisition of assets and services - for example are buildings rented or purchased. These decisions will be made before the Final Business Case is complete.

5.1.5 Impact on Income and Expenditure Account

5.15. As described earlier, this financial case does not contain information on implementation costs nor on transition to full operation. This is because these costs depend on detailed design work that has yet to be undertaken and is so complex it cannot be undertaken until the form of the agency is determined. Therefore, the net revenue costs of the implementation phase, through to reaching steady-state is currently unknown and will be determined when plans are drawn up for the implementation of the preferred option. The Financial case compares the estimated steady-state agency running costs from 2020-21 onwards for each of the options to allow a meaningful comparison to be made, as described in section 5.2 below.

5.1.6 Value Added Tax

5.16. While all costs at this stage are exclusive of VAT, the potential differences between the options in VAT arrangements and in net cost to the public purse due to VAT have been considered. It is anticipated that options 5 and 6 have the potential to be more expensive than the other options in terms of net cost to the public purse from VAT due to the potential involvement of other bodies who may not share the ability to recover VAT for some services in the same manner that an Executive Agency and many other parts of the public sector can. It is anticipated that the net cost to the public purse due to VAT for options 1 through 4 will be broadly similar. We will continue to keep VAT under review as we progress the preferred option to the final business case.

5.17. The analysis undertaken in section 5.6 suggests that whatever the resultant VAT treatment it would not have a material bearing on the choice between options.

5.2 Estimation of Costs

5.18. This section of this report looks in turn at each option estimating how much it would cost to administer social security in this way under 'steady state'. It also looks at the estimated number of full time equivalent staff ( FTE) needed within the social security agency under each option. The costings work is undertaken primarily using the Current Activity Based model ( CAB-M) developed by Communities Analysis.

5.19. CAB-M is an Excel based model which uses detailed activity based information from DWP on the administration (based on Departmental Expenditure Limits - DEL) of the benefits we are receiving as part of further devolution for 2014/15. This activity based information includes both financial resource and full time equivalent ( FTE) resource for that year. It also includes 'fully loaded' costs for the overheads which administering those benefits incurs i.e. use of central functions such as HR. These are then scaled to apply only to Scotland not the rest of the UK. This activity based information from DWP is then used to estimate the activity needed for Scotland to administer these benefits going forward by applying initial caseload forecasts for devolved benefits in Scotland in 2020-21 to estimate steady state costs.

5.20. In addition to this the CAB-M model adjusts the activities needed, constant capabilities and system costs, and the estate costs to fit the administration of the benefits by Scottish Government rather than DWP. This is done for each option within the business case based on how the activities will be undertaken e.g. by local authorities in option 3. More information on the detailed assumptions and specification for the model, and for each of the options, is provided in technical annex B of this OBC.

5.21. This section then goes on to compare each option in terms of the recurrent costs and agency FTE with option 0, which is the status quo of DWP continuing to administer benefits, to show what the additional cost and staffing requirement of devolving social security to Scotland is estimated to be under each option.

5.22. In order to undertake this estimation of the costs of each option, a range of modelling assumptions have been made for each option. These are summarised alongside the costs and FTE for each option in the sections below. FTEs are (in this section) for both the Agency and Scottish Government. Full details can be found in the associated annex.

5.23. In addition to these modelled assumptions, in a number of places a case study approach has been taken to illustrate the resource implications around some of the options rather than providing cost estimates at this stage of the appraisal process, in order to provide further evidence for decision making. When a full business case is developed following a Ministerial decision on the final configuration of the social security system, full costings will be provided including any case study elements which are to be taken forward.

5.24. For each option a similar methodology has been used to estimate digital costs of the enabling systems but as discussed in the introduction to this chapter the approach is different to that taken with other costs.

5.25. In the DWP data, there are costs for "Costs Only IT lines". Those costs relate to the maintenance of existing DWP systems. Given the very varied IT landscape within DWP - and the connectedness of systems that relate to the benefits being devolved to other systems within DWP that relate to reserved benefits - the DWP costs for IT are not a useful comparator for the future on-going IT costs for the social security agency. Because of this, we have removed the DWP IT costs. We have done some early discovery work to understand at high level the type of modern, IT architecture that would be required to support the administration of the devolved benefits. Cost estimates for the on-going operation of these systems have been used instead of the DWP costs. These costs are higher than the DWP costs but are subject to different variance and uncertainty compared with other costs.

5.26. The costs for the above are dependent on the number of staff and number of central offices and further locations, as those have a significant impact on some areas of IT cost ( e.g. numbers of licenses to access systems, or security costs for additional locations). In the point estimates below, VAT, and inflation have been excluded as well as any contingency or optimism bias. This is to provide a consistent approach across all costs. However, given the differences in uncertainty the Digital and IT costs, if considered alone, should be considered in terms of the range of contingency and optimism bias discussed in section 5.5.

5.27. In addition to the above costs, estimates from the CAB-M model of IT Costs that include staff costs are included in the enabling systems costs. This includes staff IT support.

5.28. The CAB-M model will be augmented going forward with a suite of detailed models for each benefit - the Future Activity Based model ( FABM) - that will allow the impact of detailed policy and design changes on costs to be assessed as specific Scottish processes are developed. These models will be an important part of future design development work for both the Agency as a whole and for individual benefits.

5.29. The models initially generate central estimates but ranges of costs are reported in this document. The use of ranges here explore the likely variation of possible costs for each option, taking into uncertainty such as a) the extent to which unit costs derived from the DWP model accurately reflect unit costs in the Scottish delivery system (which is still being designed) and b) uncertainty in the costs for digital systems; these costs could be considerably greater than the point estimate in the model, which includes no contingency cost. The non-digital costs are varied using a normal distribution which is symmetric (and allows costs to be lower or higher). The digital costs are examined with a Weibull distribution which is asymmetric (and only allows costs to be higher).

5.30. It is essential to note that there are still some factors, such as policy decisions on PIP ( e.g., face to face assessments at award review, award durations), whose uncertainty are not reflected in the ranges presented in the table and which will impact the actual realised cost. Further analysis is undertaken in the following sections. This is particularly relevant to digital costs.

5.2.1 Estimating the overall costs of the social security system under the status quo (option 0)

5.31. The costs of option 0 which is the status quo of DWP continuing to administer benefits in Scotland is outlined below. This is based on DWP performing all of the functions set out in Figure 21 above, except those currently undertaken by others either in the 3 ^rd, public or private sector e.g. in the area of pre claim and support services or assessment (known as medical evidence by DWP).

5.32. Based on these parameters and assumptions the annual cost for DWP under option 0 in 2020/21 would be in the region of £150 million. It is not possible/appropriate to estimate the FTE under option 0 as DWP currently does not in the main have separate operations for Scotland with its staff working across the UK on a lot of the functions below.

5.2.2 Estimated costs of option 1 - The agency centrally delivers social security in Scotland

5.33. Based on these parameters and assumptions the annual cost of option 1 in 2020/21 would be in the region of £145 to £180 million with the agency employing around 2,500 FTE staff including around 500 staff administering assessments.

5.2.3 Estimated costs of option 2 - The agency delivers social security in Scotland through local offices

5.34. For cost purposes there are two versions of option 2 with one having some local elements but with a large head office and the other very local with a smaller head office. These are 2a is where some Queries, General Enquiries and Claim Progress; Pre Claim, Support Services and all face-to-face assessment would be spread across the local offices, with all other functions being located at the headquarters and 2b, where administration of the benefits is spread across local offices with only Corporate Functions remaining at the headquarters. In addition 2a has the telephony function by central staff, 2b has the telephony function by local staff. Therefore 2a and 2b can be seen as extremes in terms of the balance of staff located locally and staff located centrally to explore the impact that this has.

5.35. Based on these parameters and assumptions, the annual cost for option 2a in 2020/21 would be in the region of £170 to £210 million, with the agency employing around 3000 FTE staff including around 500 staff undertaking assessments (Administration - assessments). Estates costs for these assessments (within the Estates cost line) are additional and higher under option 2a than option 1 (which had no physical space for assessments). Some functions (within Centralised Functions) would be undertaken by Scottish Government using their existing IT systems, with nearly 100 staff providing strategic and policy development as well as management information .

5.36. Further Costs specific to option 2a include costs for local building staff (building manager, front of house, reception/post, local IT support) and additional local face to face advisors.

5.37. Based on these parameters and assumptions, the annual cost for option 2b in 2020/21 would be in the region of £165 to £205 million, with the agency again employing around 3,000 FTE staff including around 500 staff undertaking assessments (Administration - assessments) at a cost of just over £25 million. Estates costs for these assessments (within the Estates cost line) are additional and higher under option 2b than option 1 (which had no physical space for assessments). Some functions (within Centralised Functions) would be undertaken by Scottish Government using their existing IT systems, with nearly 100 staff providing strategic and policy development as well as management information.

5.2.4 Estimated costs of option 3 - The agency delivers most benefits, but local authorities provide the face-to-face contact for the social security system and additional benefits based on local need

5.38. Based on these parameters and assumptions the annual cost for the agency and Scottish Government under option 3 in 2020/21 would be in the region of £145 to £180 million, with the agency employing around 1,800 FTE staff . Some functions (within Centralised Functions) would be undertaken by Scottish Government using their existing IT systems, with nearly 100 staff providing strategic and policy development as well as management information. The costs which would fall to local authorities under this option are included in this.

5.2.5 Estimated costs of option 4 - The agency delivers cash and benefits in kind as goods, services or concessions.

5.39. Based on these parameters and assumptions the annual cost of option 4 in 2020/21 would be in the region of £150 to £190 million with the agency employing around 2600 FTE staff including around 500 staff undertaking assessments (Administration - assessments) at a cost of just over £25 million. In addition the cost of providing 'in kind' benefits would be in the region of £4 million including 25 staff - this only includes costs for concession cards for those receiving disability and ill health or carer's benefits. Other in-kind costs are described in the case study section below. Some functions (within Centralised Functions) would be undertaken by Scottish Government using their existing IT systems, with nearly 100 staff providing strategic and policy development as well as management information.

5.40. These costs are shown with in-kind options as a "bolt-on" to the option 1 model. A further variant has been developed, which adds in-kind options as a bolt on to option 2b, but costs are not shown below.

5.2.5.1 Option 4 case storage, distribution and fitting of goods

5.41. In addition to the in-kind costs that form part of the estimate above, it is possible under option 4 to look at the costs of existing 'in kind' benefits provided by the Scottish Welfare Fund and other areas. There is a discussion of this in annex B.

5.2.6 Estimated costs of option 5 - The agency provides governance but the delivery of social security is done by others e.g. via procurement or a Service Level Agreement

5.42. Based on these parameters and assumptions the annual cost of option 5 in 2020/21 would be in the region of £165 to £200 million with the agency employing around 1,300 FTE staff. Assessments (Administration - assessments) are undertaken by others at a cost of nearly £38 million (including costs for IT systems, accommodation and estimated contractor profit). Some functions would be undertaken by Scottish Government using their existing IT systems with nearly 100 staff providing strategic and policy development as well as management information.

5.2.7 Estimated costs of option 6 - Social security is embedded in a range of existing public services with the agency providing governance.

5.43. Based on these parameters and assumptions the annual cost for the option 6 in 2020/21 would be in the region of £225 to £275 million with the agency employing around 600 FTE staff). Some functions would be undertaken by Scottish Government using their existing IT systems with 150 sta ff providing strategic and policy development as well as management information at an estimate cost of £35m.

5.44. The costs which would fall to the wider public sector under this option are included here in this as an estimate and are explored in the case study in the Annex.

5.3 Summary of cost ranges

5.45. The range of probable costs for option 1- 6 are set out in Table 7 below.

Table 7 - Comparison of costs

	Cost Estimate
Option 0	£155m	Represents an SG estimate of the costs of the DWP calculated on an equivalent basis to the costings of the other options.

	Range of Cost Estimates
Option 1	£145m	to	£180m
Option 2a	£170m	to	£210m
Option 2b	£165m	to	£205m
Option 3	£145m	to	£180m
Option 4	£150m	to	£190m
Option 5	£165m	to	£200m
Option 6	£225m	to	£275m

5.46. Based on the range of costs for options 1, 2, 4 and 5, a reasonable estimate of the annual cost of running the social security agency under a number of configurations is between £145-£275 million. Option 1 and option 3 are likely to be the cheapest options when compared to options 2, 4, 5, and 6, at an estimated annual cost of between £145-£180 million. Note that costs for option 6 (the most expensive option) are particularly difficult to assess - for a full discussion, see Annex B, Section 11.7 "Agency and Scottish Government Costs under option 6".

5.47. Point estimates of the FTE employees in the agency or Scottish Government are shown in Figure 13 - FTEs in each option (Agency/ SG and public sector). For option 6, wider public sector figures relate to all those with a social security role.

Figure 13 - FTEs in each option (Agency/ SG and public sector)

5.4 Formal presentation of present value of costs for each option, NPV and BCR

5.48. Table 7 above shows the steady state on-going costs for the options.

5.49. In order to present the costs of each option in a standard format, a 30-year time horizon is used from the point (2020-21) that the agency is in place. The core assumption is that on-going costs rise in line with population growth (of the average order, as projected by National Records of Scotland ^[33] ) of 0.3% per annum. Two sensitivity tests are made to the calculation by using a constant population and the projected growth in the over 65 population (which may be a better proxy for caseload of disability benefits) which is significantly higher at around 3%.

5.50. Note that, in accordance with standard methodology, the analysis is conducted in current prices i.e. no inflation. Applying a discount rate of 3.5% in line with HMT Green Book methodology gives the following results for the Present Value of Costs ( PVC) shown in Table 8.

Table 8 - Present Value of costs ( PVC) and Present Value of Benefits ( PVB)

	PV of costs (£m)							PVB (£m)
Option	1	2a	2b	3	4	5	6
Population growth (0.3%)	2,630	3,020	2,950	2,620	2,740	2,870	4,000	8,300
Constant population (0% growth)	2,530	2,910	2,850	2,520	2,640	2,770	3,860	8,010
O65 population (3% growth)	3,720	4,280	4,190	3,710	3,880	4,070	5,670	11,170

5.51. A similar approach is undertaken for the monetised benefits estimated in section 4.1. The present value of the benefits ( PVB) (of the benefit expenditure) is shown in the last column of Table 8. The Net Present Value ( NPV) for each option under each scenario and the overall Benefit Cost Ratio ( BCR) are shown in Table 9. The Benefit Cost Ratio is calculated by dividing the PVB by the PVC. An alternative formulation is BCR = ( NPV+ PVC)/ PVC. Note it is also assumed that the distribution of income remains constant over time.

Table 9 - Net present value ( NPV) and Benefit Cost ratio ( BCR)

	Overall NPV and BCR
	Overall NPV ((£million) BCR(= PVB/ PVC)
Option	1	2a	2b	3	4	5	6
Population growth (0.3%)	5,680 3.16	5,290 2.75	5,360 2.81	5,690 3.17	5,570 3.03	5,440 2.89	4,310 2.08
Constant population (0% growth)	5,480 3.16	5,100 2.75	5,160 2.81	5,490 3.17	5,370 3.03	5,240 2.89	4,150 2.08
O65 population (3% growth)	8,050 3.00	7,490 2.61	7,580 2.67	8,060 3.01	7,890 2.87	7,700 2.74	6,100 1.97

5.52. This analysis, whilst limited in its power to choose between options as it is unable (as discussed in the socio-economic case) to capture the differences between options, is a strong demonstration of the overall value of the system. In terms of the BCR, in pure monetary terms any of the options represents strong technical value for money. By way of comparison, a BCR of greater than one demonstrates that the benefits are greater than the costs and a BCR greater than around 1.3 does so whilst taking into account (in a similar manner to the distributional analysis used to generate the PVB) that government expenditure needs (to a greater or lesser extent at the UK level) to be funded by tax receipts. The worst case - option 6 assuming expenditure rises with 065 population growth - has a benefit cost ratio of 1.97 which is still very good value. Option 1 and 3 achieve BCR of above 3.

5.53. Section D of this OBC considers the implications of the costs of each option along with the remaining four cases and derives a preferred option.

5.5 Optimism Bias and contingency for Digital

5.54. The NPV of costs in the preceding section took the point estimates of the steady state costs This section considers the steady-state costs in terms of wider guidance on optimism bias. The key factor driving this section is the accepted variability and sensitivity of the estimates of IT costs.

5.55. HMT guidance on Optimism bias ^[34] states:

Project appraisers have the tendency to be over optimistic. Explicit adjustments should therefore be made to the estimates of a project's costs, benefits and duration, which should be based on data from past or similar projects, and adjusted for the unique characteristics of the project in hand.

This guidance provides cost and time uplift percentages for generic project categories which should be used in the absence of more robust primary data.

5.56. The sensitivity analysis provided in section 5.6 and the Monte Carlo analysis that generates a range of costs should be thought of as replacing more generic analysis (as described above). However, there remains the issue of potential optimism bias around the IT costs. HMT suggested initial optimism bias uplifts for a range of project types are detailed in Table 10 below.

Table 10 - Standard OB uplifts

Project Type	Optimism Bias (%)
	Works Duration		Capital Expenditure
	Upper	Lower	Upper	Lower
Standard Buildings	4	1	24	2
Non-standard Buildings	39	2	51	4
Standard Civil Engineering	20	1	44	3
Non-standard Civil Engineering	25	3	66	6
Equipment/Development	54	10	200	10
Outsourcing	N/A	N/A	41*	0*

* The optimism bias for outsourcing projects is measured for operating expenditure.

5.57. It is also worth noting that analysis of the variation of costs within categories shows extensive variance. For instance the 66% uplift for Non-standard civil engineering projects is an average cost overrun across a large number of capital projects but this average is considerable influenced by very, very large variations in a few large scale projects. See Mott MacDonald (2002), Review of Large Public Procurement in the UK.

5.58. The Equipment/Development category:

Equipment & development projects: Projects that are concerned with the provision of equipment and/or development of software and systems ( i.e. manufactured equipment, Information and Communication Technology ( ICT) development projects) or leading edge projects.

is clearly the appropriate one to use.

5.59. Given the early stage of the Digital cost estimates this suggests a 200% uplift (a cost level 3 times the initial estimate) would be appropriate as an illustration of potential levels of optimism bias at this stage. An alternative approach would be to consider a 100% contingency on the digital costs at this stage. Both have been undertaken but for reasons discussed in the foreword are not reported in detail here. However, it is possible to show the impact of the OB assumption without directly revealing commercially sensitive information that could have an impact on future value for money.

5.60. Significant detailed design is on-going and will form a key part of the final business case material. In line with the discussion in the Strategic case, this design will include co-production with users and the final costs will be dependent on design specifications that have yet to be determined. Part of the reassurance around this aspect of the costs is the analysis presented in section 5.4 that demonstrated a strong NPV and BCR in pure economic terms. Updating the analysis to include OB for the population growth scenario gives the following as shown in Table 11. This continues to give strong evidence of overall VFM.

Table 11 - BCR including 200% digital Optimism Bias

	BCR (£million) with 200% digital OB (£m)
	Overall NPV
Option	1	2a	2b	3	4	5	6
BCR Population growth (0.3%)	2.58	2.16	2.19	2.62	2.46	2.44	1.68

5.6 Sensitivity of costs to other factors

5.61. As a way of assessing the sensitivity of the costs to other factors, Table 12 shows the percentage changes is the costs of individual capabilities, across all the options, that are required to change the ranking of options by cost. As option 1 and option 3 are quite close in costs, the table shows the changes required for 1 change in ranks (two options switch places) and for more than 1 change in ranks (more than 2 options switch places). This is a way of assessing the potential impact of bias in the costings of any of the capabilities on the overall performance of the options.

Table 12 - Sensitivity of cost rankings to changes in costs of individual capabilities

Capability	%age change in costs of capability required to change rank of costs	%age change in costs of capability required to change rank of costs of more than 2 options
Queries / General Enquiries / Claim Progress chase etc.	35%	68%
Pre Claim / Support Services	36%	236%
Administration - assessments	21%	32%
Administration - Other	100%	101%
Dispute Resolution	N/A	N/A
Error Management / Investigation	150%	350%
Centralised Functions	100%	234%
Management Information	N/A	N/A
Enabling Systems & Constant Capabilities (Digital)	401%	801%
Other - Estates	219%	241%

5.62. The results confirm that the costs are not particularly sensitive to changes in the costs of specific capabilities. This is particularly the case for digital. It takes a 800% increase (significantly greater than the OB and contingency levels in the previous section combined to change the rank more than 1 pair of options).

5.63. Of the other capabilities, Administration - assessments is the most sensitive to changes. This suggests (as is discussed elsewhere) that assessments are a key driver of costs. Dispute resolution and management information are so similar across options that no change in the costs of the capability change any of the option ranks.

5.7 Potential impact of implementation

5.64. UKG agreed that at total of £200m will be available through the Fiscal Framework to cover implementation of the devolved powers. SG has always been aware that implementation costs are likely to be higher than this amount but of a similar order of magnitude. As discussed above, this OBC is for the agency at steady state.

5.65. There are some areas where implementation costs for any option are likely to be significant - particularly on estates and on IT. Work has been undertaken to begin to cost the implementation of information technology through initial discovery and alpha work. While many of the IT implementation costs would be similar across all of the options considered here, there is some variance - primarily associated with additional security costs associated with larger numbers of separate locations, or greater levels of interaction with systems elsewhere.

5.66. Other implementation costs will depend on decisions yet to be taken about the detailed form of the agency, the procurement route chosen and the nature of the assets to be procured. As section 5.5 demonstrates, the impact of long-run steady state costs is of a significantly greater order than the likely range of implementation costs. For the purposes of this OBC, what is of prime importance is how implementation costs might affect the choice of option. Given the discussion above they are likely to be broadly of the form:

Option 5< Option 1 or 4 < Option 3 < Option 2 < Option 6

5.67. In words, option 5 will be cheaper than option 1 or 4 which will be cheaper than option 3 etc. However, the differences will be small compared to the differences in NPV shown in Table 9. Option 5 is likely to the lowest cost implementation wise and option 6 the most expensive. The differences between the other options will be smaller.