Social security agency in Scotland: outline business case

Collection of the analysis and evidence behind the Ministerial Statement on Scotland's social security agency, made on 27 April 2017.

9. Assessment of the hybrids and the preferred option

9. This section examines the two pairs of three hybrids discussed in section 8.4. Section 9.1 considers the 3 delivery hybrids under the MCA lens and section 9.2 does the same for the 3 assessment models. Section 9.3 considers the costs of the hybrids including sensitivity analysis and, finally, section 9.4 considers the preferred option that should be taken forward to Final Business Case.

9.1 MCA assessment of Hybrids

9.1. Hybrids X and Z are more likely to successfully deliver Dignity and Respect outcomes than Hybrid Y. Hybrid X offers greater flexibility of the type of advice and support and better alignment with the existing advice and support around other public services. Hybrid Z provides a greater local presence and more comprehensive face-to-face advice and support, which together with local case management and decision-making may be more preferable for some people. Hybrid Y is the weakest of the three hybrids, as people would be required to travel greater distances to access advice and support services. Both Hybrids Y and Z have advantages in terms of offering more scope to set and maintain a certain organisational culture and provide a more transparent system from the accountability perspective.

9.2. Hybrid X is considered the strongest option in terms of Equality and Poverty, largely because more comprehensive phone and online support would allow more tailored interaction with specific groups. Accessing support online or over the phone may be a better option for those on low incomes than travelling to local offices. Hybrid Y is the weakest option because of travel that may be required for some people to access face-to-face support, with Hybrid Z providing a more accessible system in comparison. Hybrid Z could provide more consistent advice and support service as it would be delivered from agency's own offices.

9.3. Hybrid X and Hybrid Z both have strengths and weaknesses in relation to different aspects of Efficiency and Alignment, but on balance are expected to be similar in how they perform on this measure and are considered to be somewhat stronger than Hybrid Y. Hybrid X allows for more scope to explore efficiencies within existing public services by co-locating advice and support functions and potentially delivering the same services with fewer resources. With Hybrid Z, local case management could speed up certain functions and create a more streamlined model and with one local office per Local Authority, some efficiencies could be explored through co-location, which cannot be achieved to the same extent with Hybrid Y.

9.4. In terms of Implementability and Risk, Hybrid Y ranks higher than the other two hybrids, although the differences in the delivery challenges presented make comparisons difficult. Hybrid X presents a greater implementability challenge because of risks around ensuring effective co-location and co-production and functional phone and online support. However, Hybrid X could be delivered quicker, as it does not require local office estate procurement. Hybrid Z is expected to take the longest to become operational because of the number of local offices.

9.5. On balance, it is deemed that Hybrid Z is marginally stronger than other hybrids on Economy and Environment, despite poorer performance against resource consumption related criteria. This is because it less likely to have a direct impact on existing public sector services, which could be negative due to capacity issues in some areas under Hybrid X. Hybrid Y is considered to be the weakest, partly because it could increase the burden on existing third sector advice services if advice and support is limited to 8 local offices.

9.6. The results are summed up in Figure 18 below.

Figure 18 - MCA for hybrid delivery structures

Figure 18 - MCA for hybrid delivery structures

9.2 MCA assessment of assessment models

9.7. Overall, model MA2 is considered to be the best model when ranked against the Dignity and Respect criteria. Conducting face-to-face assessments from fixed locations, with outreach services for certain cases and most assessors being solely employed be the agency provides the most flexibility and choice. Model MA3 has the advantage of NHS professionals with the right level of knowledge providing the evidence for assessments which could improve decision-making, as well as wider public perception about benefit claimants. However, this model may not be sufficiently flexible in terms of timing and location of assessments and accountability lines may become blurred between the agency and the NHS from the perspective of those being assessed. Model MA1 is considered the weakest model of the three because of the lack of flexibility it offers. Home-based face-to-face assessments may suit some people, especially those living in remote rural areas, as well as those who cannot travel because of their disability. However, default home-based assessment is unlikely to suit everyone's needs because of differences in suitability of people's homes for assessment and could result in assessments being of lower quality and consistency.

9.8. On balance, models MA1 and MA3 are considered to be somewhat more favourable in terms of Equality and Poverty, although on balance the differences between the three models are likely to be small. Model MA3 could encourage greater take-up of benefits by reducing stigma and fear of the assessment process. Model MA1 eliminates the requirement to travel to face-to-face assessment location which can be a considerable upfront cost for those on low income even if expenses are reimbursed. Although weaker on this criteria set than the other two options, model MA2 is most likely to deliver better outcomes in terms of the fairness element, as the process of assessment and skills of assessors are more likely to be consistent across cases.

9.9. There was found to be little difference between models MA1 and MA2 in terms of Efficiency and Alignment. Both are likely to be more efficient models from the process perspective than model MA3. It is likely that the assessment process would be less simple and streamlined with model MA3, not least because the NHS professionals could only work on cases on a part-time basis. Model MA3 may also present difficulties around Ministerial ability to implement policy changes around assessments.

9.10. Model MA2 performs best against the Implementability and Risk criteria, as it most closely resembles the model that is in place at the moment, without the element of private sector involvement. The biggest challenge in model MA2 is finding suitable premises for conducting assessments, which is considered to be of smaller difficulty than those presented by models MA1 and MA3, which are largely around recruiting and retaining a sufficient number of assessors. The mobile worker model in MA1 may not be sufficiently attractive as an employment opportunity and there is an additional risk associated with assessments being done through home visits not being implementable because of unsuitability of some people's homes for such a purpose. Despite potentially positive public perception at the outset, model MA3 is considered to be the weakest in terms of Implementability and Risk, as the existing NHS workforce may not have a sufficient level of inclination and/or capacity to meet the assessment requirements and any incentives that displace existing resource away from healthcare to social security may prove difficult.

9.11. On balance, there is not a substantial difference in how the three models of assessment perform against the Economy and Environment criteria, although the models differ in their performance on individual aspects of this criteria set. Model MA1 is the weakest in terms of employee satisfaction as the mobile worker model could involve a lot of travel for the assessors and current public attitudes to the benefits assessment process may make home visits a stressful element of the job. Model MA2 requires both parties to travel to the assessment location and requires more resource consumption because of the use of fixed locations for assessments. Model MA3 has the potential to have a detrimental impact on the performance of the NHS.

9.12. The results are summed up in Figure 19 below.

Figure 19 - MCA results for assessments

Figure 19 - MCA results for assessments

9.3 Costs of the hybrids

9.13. The majority of the social security system does not alter between the hybrids, with the agency undertaking the functions around administration, dispute resolution, error management / investigation and management information in the same way for all the hybrids.

9.14. Where the functions differ is around pre-claims support and advice and assessment. Varying the approach to these functions does however have knock-on effects on the constant capabilities and enabling systems required, the services that support these functions, and the estates footprint needed to deliver them. This means that the overall estimates of steady-state running costs vary not only between the three hybrid options, but also for each hybrid and assessment variant.

9.15. There are therefore separate estimates of cost for each of 9 scenarios - each of the three hybrids X, Y and Z with each of the assessment models MA1, MA2 and MA3.

9.3.1 Point estimates and Average estimates

9.16. The cost model is built up by estimating the expenditure required for each individual function, combined with assumptions about how that function will be delivered under each hybrid, and assumptions about the work required to manage that function. For each benefit, we have assumed that the caseload remains the same as it currently stands, that the number of new claims (in-flows) and the number of claims ending (out-flows) are similar to current patterns, and that the associated administrative requirements remain the same as they are now.

9.17. For each function, we can estimate not only a point estimate of costs, but also a likely distribution of costs, based around likely ranges. Monte Carlo simulation methods are then used to sample costs from these distributions, which are then combined to give an empirical distribution of total costs.

9.18. For most of the cost components the distributions are relatively symmetric. But it is worth noting that for the purposes of this modelling exercise, the uncertainty associated with the steady-state costs of IT and Digital infrastructure is greater than for the other cost components, spanning a range up to three times the central estimate. This is in line with HMT guidance on 'optimism bias' for IT procurement costs. This results in an overall estimated range of costs that are asymmetric - the uncertainty is greater on the high side that on the low side. The practical effect of this is that when average costs across the whole distribution are calculated, they are higher than the original point estimates.

9.19. This is a subtle analytical point, but from the point of view of presenting realistic, cost estimates before detailed design has taken place, we have used the average costs rather than the point estimates as the primary figures, as these capture a greater degree of the uncertainty inherent in the modelling.

9.3.2 Results

9.20. Costs for the hybrids are laid out in the tables below, as average values (Table 16) and as point estimates (Table 17).

9.21. For the average costs estimates, these range from a low of around £145m to a high of about £170m. and show that the different models for assessment lead to much greater variation in costs than the difference between the three hybrids. The increased expenditure on a more dispersed estate in non-prime locations for Hybrids Y and Z is somewhat offset by the saving made by requiring a smaller 'head office'.

Table 16 - Costs of the hybrids (average cost)

Table 16 - Costs of the hybrids (average cost)

Table 17 - Costs of the hybrids (point estimates)

Table 17 - Costs of the hybrids (point estimates)

9.22. Figure 20 captures the uncertainty in these estimates, showing graphically that the distribution ranges from a low of around £130m to a high of about £190m. Again, these are not budget figures, they simply represent our best estimates of the likely costs of different delivery models, and exist as an aid to decision-making, where the difference between the estimates is more important than their absolute level.

Figure 20 - Costs of the hybrids (point estimate from CAB model, with 5 th percentile, mean and 95 th percentile costs from Monte Carlo simulation)

Figure 20 - Costs of the hybrids (point estimate from CAB model, with 5th percentile, mean and 95th percentile costs from Monte Carlo simulation)

Table 18 - Breakdown of costs (point estimates) in the hybrid options.

Table 18 - Breakdown of costs (point estimates) in the hybrid options

Colours show the variation in costs within rows from blue (low) to white (medium) to red (high).

9.23. Table 18 above shows a breakdown of the cost components. A significant element of the costs are fixed across all hybrids and assessment models. This figure of £63m, which represents between 38% and 45% of the total costs is associated with core central and corporate functions that would be carried out in the same way (and in the same location) regardless of the Hybrid option. So a useful working assumption is that about 40% of the estimated costs are fixed across all the options. The rest, which represents pre-claims advice and support staff, assessments, the estate footprint and IT/digital costs are essentially the variable costs. As noted above digital/ IT costs are highly uncertain at this stage.

9.24. It is sensible to consider the impact of applying standard and appropriate assumptions (as in section 5.5) around digital costs. This would again be a contingency of 100% and discussion of optimism bias of 200% if applying the standard rate at this stage of the project. This analysis suggests that when taking into account the variability of the IT cost estimates, the 95% sensitivity levels reported in Figure 20 are a suitably prudent estimate. It should be noted that whilst at this stage, the additional variation on the digital costs does not have a bearing on the choice of preferred option, it will be important that careful and appropriate mitigation is put in place as digital implementation thinking is developed.

9.3.3 Costs discussion

9.25. There are three key messages from the above analysis;

9.26. The assessment option featuring the NHS pool ( MA3) is more expensive than other options for assessment. The two key reasons for MA3 being more expensive than MA1/ MA2 are:

A. The NHS staff pool includes specialists with in-depth knowledge of particular conditions - these specialists are more expensive than trained medical staff from within the agency.

B. The resource required for the administration of assessments (scheduling, liaising with assessors on reports) is assumed to be much higher in this option (double that of MA1 and MA2), on account of there being a much higher headcount of professionals in the pool than agency employees in the other options.

9.27. The cost of the assessment option is very sensitive to the exact mix of health professionals in any option, due to the very wide variety in hourly rates for different professionals. Notably, MA3 cost currently includes medical consultants but not GPs; including GPs would make MA3 even more costly (by replacing lower paid medical practitioners).

9.28. Hybrids Y and Z are more expensive than Hybrid X. A key driver of higher costs in Hybrids Y and Z is additional face-to-face advisers; while all Hybrids feature both remote support and face-to-face advice, Hybrid X features enhanced remote support while Hybrids Y and Z feature enhanced face-to-face advice, with the net effect being a greater resource on pre claims / support in Hybrids Y and Z.

9.29. Hybrid Z (32 offices) is more expensive than Hybrid Y (8 offices). This is primarily due to the additional front of house staff (manager, receptionist / post, house staff) and the expensive "retail" like space that they require. There is also some effect from whether offices happen to be in prime, secondary or tertiary rental rate locations. Under assessment option MA3, Hybrid Y is nearly as expensive as Hybrid Z. This is because the travel time of health professionals from their place of work to the 8 offices is deemed to be significant, which makes overall assessment time (including travel) higher, with associated costs for staff time

9.4 The preferred option

9.30. The analysis above is conclusive for the choice of delivery hybrid. It is worth noting that the three delivery hybrids already reflect the evidence within the strategic, commercial and management cases so the focus in moving to the preferred option is based on the MCA (or socio-economic case) results and the financial implications.

9.31. As such, hybrid X stands clear from the other two options. It is the best overall in terms of the MCA analysis and it has the lowest cost. As such hybrid X should be considered the preferred option for the delivery structure of the agency. That is to say the basic form of the structure should be:

Hybrid X - a central agency with enhanced phone and online support, which incorporates face-to-face pre-claims and support services locally in existing public sector locations

9.32. Given the complex nature of the delivery landscape across Scotland, the choice of hybrid X as the preferred option for the basic form of the structure should not preclude adaptation to specific local needs. For instance, existing public sector locations may be limited in some areas so the agency could explore alternative options such as a mobile workforce or mobile offices or consider the use of a limited amount of its own offices if there would be strong local value.

9.33. The situation with assessments is much more complicated. Each of the three hybrids have advantages and disadvantages. Although the use of NHS professionals in hybrid MA3 is more expensive and is harder to implement when faced with the current situation it has advantages that could be developed in the context of wider change.

9.34. For assessment, on current scoring, a model based on Agency staff carrying out assessments in Agency or other public sector locations appears most viable. This is very similar to how assessments are carried out at the moment, but with the crucial difference that private sector contractors would no longer be involved. With the choice of hybrid X, this would clearly involve the use of other public sector locations. However, a model which uses the existing expertise and workforce of the NHS is still being explored. Many of the potential drawbacks of this model which have been identified through this analysis can now be examined and addressed directly.

9.35. The optimum choice of assessment process will be dependent on decisions that are yet to be made around policy and practice. As such it is too early at this OBC level to reach a preferred option for assessments. All three assessments models examined as part of this section and indeed other variants which could include contracting with an external organisation could be implemented with any of the three hybrid options and with the preferred hybrid x in particular. As such this refines the preferred option overall, at this stage, to:

Preferred option - a central agency with enhanced phone and online support, which incorporates face-to-face pre-claims and support services locally in existing public sector locations and with assessments undertaken in a manner that is appropriate for policy choices that will be made as the final business case is progressed.

9.36. The job requirements are subject to considerable variation based on detailed decisions yet to be made but it is reasonable to suggest that the central agency will be directly responsible for at least 1500 jobs (Full Time Equivalents) with a number of additional staff in other locations.


Email: Andy Park

Phone: 0300 244 4000 – Central Enquiry Unit

The Scottish Government
St Andrew's House
Regent Road

Back to top