4Recurring themes in the reviews
Timetable, scale and automation
Delays
27.The Department has been consistently over-optimistic about the time it would take to roll out UC. According to its initial plans, the programme should have been complete by October 2017. These delays continue: following the November 2017 Budget, the digital service will be rolled out to ten Jobcentres per month between February and April 2018 compared to over 60 per month in previous plans.91
Progression of UC rollout assumptions
28.Delays are not simply an embarrassment to the DWP: they are also expensive. The digital service is expected to be “predominantly online, integrated and accurate” and result in major savings from automation and self-service.92 The Department is expected to reduce numbers of administrative staff by 11,000, more than 25% of the total.93 The NAO noted in its November 2014 report that a six month delay to the digital service reduced the net benefits of the programme by £2.3 billion.94 The NAO estimated that, in staff costs alone, the digital service would be £610 million a year cheaper than using existing live service systems.95 The live service was expected to help develop operational capacity in Jobcentres. As the DWP acknowledged, however, its use as a learning tool would be limited.96 Its largely manual operation would only give limited insights into the feasibility of delivering a national digital service on the timetable planned and with the levels of automation required.
Automation and verification
29.The UC digital service is currently operating with more staff and fewer claimants than the DWP expected.97 This makes it expensive to run. The March 2017 AAP found that costs per claim were £963 in January 2018. This was a substantial decrease on the £1,750 in April 2016, but still almost four times the target of approximately £250 by May 2019.98 The September 2016 HC cautioned that delivery of the programme was “at risk” if a significant proportion of the service did not become automated, because costs would “become prohibitive”.99
30.An important component of automation is claimant identity verification. As early as November 2013, the PAC expressed concern about when software to confirm claimant identity would be available.100 In its 2015 digital service trials, the Department intended to use Verify as an alternative to the face-to-face identification used in the live service. Verify is an identity assurance service, promoted by the Cabinet Office, which was developed for use across government.101 The September 2016 HC identified the performance of Verify, which had caused problems in a wide variety of fields,102 as a major risk to the programme. The IPA found that “assumptions based on insight work into customer journey are not at all aligning with reality”.103 Only 29% of UC claimants could complete the Verify process, and the levels of manual activities involved for the remainder were “not sustainable at scale”.104 These findings were echoed in the March 2017 AAP. Around 30% of claimants were completing the Verify process, compared to an original projection of 80% and the target of 60% for the accelerated rollout in October 2017.105
31.The September 2016 HC identified “a workable ID assurance solution” as one of four “critical success factors” necessary before the digital service should be rolled out to 29 Jobcentres in July 2017, the first major “scaling event”.106 The March 2017 AAP stated clearly that the condition had not yet been met.107 The DWP responded to concerns about Verify by developing in-house and in parallel an alternative identity assurance service called Prove your Identity.108 A July 2017 update advised the UC board that Prove your Identity could, combined with Verify, help achieve a 50% verification success rate.109 The September 2017 HC encouraged the programme to trial a third option, Verify LOA 1, a Government Digital Service product with a lower assurance standard than Verify. Online verification continued to be a significant risk to the programme.110
Industrialisation
32.The September 2017 HC concluded that, to be cost-effective at scale, the programme needed to “industrialise” its approach to “complex cases and vulnerable customers”.111 The IPA found a high level of uncertainty about the proportion of UC claimants who could be considered vulnerable, noting that estimates ranged from 25% to 50%. Given the resources required to support vulnerable claimants, if levels were towards the top end of that scale the medium-term sustainability of UC would be threatened.112
33.The IPA’s call for the “industrialisation” of UC for complex cases and vulnerable customers is an unfortunate choice of phrase. In those stark terms, however, it epitomises the challenges facing the UC programme. To make its promised efficiency gains, it must become a far more automated system. Key areas such as identity verification, however, are currently manual processes for a large proportion of claimants. Delays to the rollout and automation of the digital service reduce projected efficiency savings. In seeking to maximise savings, however, the Department must monitor closely any potential adverse effects on claimants and factor those into its decision making.
Performance criteria
34.A recurring theme of the reviews is the Department’s failure to set clear criteria for determining whether to proceed with specific stages of the rollout:113
- The January 2013 PAR expressed concern that no “formal go/no go criteria” had been set for proceeding with a planned rollout of the “Pathfinder” pilot programme.114
- The February 2014 PAR said that the Department should set a go/no go review point, commission an independent review and define more detailed success criteria before proceeding with its planned rollout of the digital service.115
- The September 2016 HC called for greater focus on the key performance indicators (KPIs), identifying the lack of qualitative and quantitative performance data to inform decisions whether to scale the programme as a risk to its success.116
- The March 2017 HC found it was unclear how data would inform a decision whether to proceed the “scaling event”.117 While a “great deal” of information was being collected, “it was not clear how this data was being turned into knowledge or insight”.118 The IPA noted that while some scaling criteria appeared “absolute”, others were “more subjective and more open to interpretation”.119 It recommended that the programme team should “maximise opportunities for robust challenge” of its readiness for scaling UC.120
- The September 2017 HC, which endorsed the accelerated rollout between October and December 2017, called for a “clear, agreed view of the criteria that could prompt a course correction in January 2018”.121 This should include the circumstances in which a pause, a slowing down of, or a stop to the rollout would be warranted, however unlikely such circumstances were.122 The IPA recommended the criteria used should include customer service performance, such as payment timeliness, alongside operational and system performance. It cautioned that the capacity and role of local authorities and housing associations should be considered.123
35.The IPA has been consistently critical of the Department’s failure to set clear criteria for proceeding to the next stage of the UC rollout. Setting clear performance standards in advance is the best way of ensuring decisions are made objectively. The publication of these would both benefit scrutiny and make it far easier for the Department to explain its decisions.
Engagement with local authorities
36.UC has important consequences for local authorities, which administer housing benefit, provide safety net payments to people in hardship and are often major landlords. The IPA reports tend to alternate between criticism and praise for the Department in its engagement with local authorities. While it tended to improve in response to criticism, the rate of improvement was not sustained.
37.The November 2011 PAR cautioned that the Department’s “one way high level communication” approach to external stakeholders could lead to a “delayed or disrupted roll out”.124 The January 2013 PAR said the exclusion for councils from joint work between DWP and HMRC on the migration of tax credit claimants was “a significant gap”.125 The February 2014 PAR found that the Department has given little thought to the operational consequences of moving claimants from housing benefit to the housing costs element of UC. This was despite “real hunger” in local government to conduct trials to understand the consequences of UC for the broader social support system. The MPA said that the programme should regard local authorities as “delivery partners”, rather than mere stakeholders, as effective relationships with them would be “critical to joined up delivery” of UC.126 The October 2015 PAR highlighted again the movement of claimants from housing benefit to UC as a “risk to the programme”. The MPA did not pursue this concern further as it was outside the scope of its review.127 The latest report, the September 2017 HC, called on the Department to work better with third parties such as local authorities, housing associations and CAB in the interests of claimants, and to be aware of the potential impact of scaling on them.128
38.Local authorities are key delivery partners for UC. The programme would have benefitted, and would continue to benefit, from better engagement with them and other landlords, including housing associations.
Achieving transformational change in government
39.As set out in Chapter 3, the early stages of the UC programme were beset by a disconnect between high-level policy objectives and operational delivery. It took several years before the programme’s approach accorded with its ambition. The October 2015 PAR found that the programme had left behind its “fixation” with in-vogue “Agile” software development methodology, which had left “the Agile tail wagging the digital dog”.129 By September 2016, the IPA was confident the UC had become a transformation programme and the September 2017 identified a “very clear delivery culture” across the UC programme.130 Nevertheless, the programme still faces cultural risks to its success. The March 2017 AAP expressed concern that frontline Jobcentre and service centre staff had “a lack of trust in new systems, with people defaulting to old behaviours”.131
40.Reflecting on his six years as the minister responsible for UC, Lord Freud said that the established division of responsibilities in government, whereby ministers set policy and officials implement it, was incompatible with achieving major change. He took a far more active role in the implementation of UC than was usual for a minister. Had he understood the challenges of achieving transformational change in government in his early days as a minister, he “would have been much more frightened of trying to do something as big”.132 UC is a valuable case study of the challenges in achieving transformational change in government which should be examined by ministers and civil servants planning major projects.
91 DWP, Transition Rollout Schedule – November 2017 to December 2018, published November 2017 and Transition Rollout Schedule – April 2017 to December 2018, published November 2016
92 NAO, Universal Credit: progress update, HC (2014–15) 786, November 2014, para 13
93 IPA, UC Health Check, September 2016, para 5.4.1. The Department expects to add 9,000 front line Work Coaches.
94 NAO, Universal Credit: progress update, HC (2014–15) 786, November 2014, para 13
95 NAO, Universal Credit: progress update, HC (2014–15) 786, November 2014, para 2.10
96 NAO, Universal Credit: progress update, HC (2014–15) 786, November 2014, para 16
97 IPA, UC Health Check, September 2017, para 6.8
98 IPA, UC Assurance of Action Plan, March 2017, para 2.10
99 IPA, UC Health Check, September 2016, para 5.5.2
100 PAC, Thirtieth Report of Session 2013–14, Universal Credit: early progress, HC 619, November 2013, p13
101 IPA, UC Project Assessment Review, October 2015, p11
102 For example, the Common Agricultural Policy Delivery Programme reverted to using drop-in centres and a telephone helpline to confirm the identity of farmers because Verify did not operate as the Department for Environment, Food and Rural Affairs expected. See NAO, Early review of the Common Agricultural Policy Delivery Programme, Session 2015–16, HC 606, December 2015, para 8
103 IPA, UC Health Check, September 2016, para 5.4.2
104 IPA, UC Health Check, September 2016, para 5.6.1.3
105 IPA, UC Assurance of Action Plan, March 2017, para 5.1, and UC Health Check, September 2017, para 6.9
106 IPA, UC Health Check, September 2016, p3
107 IPA, UC Assurance of Action Plan, March 2017, p3
108 IPA, UC Health Check, September 2017, p3
109 IPA, UC Health Check, September 2017, para 6.9
110 IPA, UC Health Check, September 2017, para 6.9
111 IPA, UC Health Check, September 2017, para 6.7
112 IPA, UC Health Check, September 2017, para 6.7
113 In its November 2014 NAO report recommended the Department “develop specific milestones for both digital and live services before each additional stage of roll-out”. See NAO, Universal Credit: progress update, HC (2014–15) 786, November 2014, para 22.
114 IPA, UC Project Assessment Review, January 2013, p12. Pathfinder was a pilot of live service UC, used for single Jobseeker’s Allowance claims in a small number of Jobcentres.
115 IPA, UC Project Assessment Review, February 2014, p15
116 IPA, UC Health Check, September 2016, para 5.5.4
117 IPA, UC Assurance of Action Plan, March 2017, p2
118 IPA, UC Assurance of Action Plan, March 2017, para 4.7
119 IPA, UC Assurance of Action Plan, March 2017, para 4.2
120 IPA, UC Assurance of Action Plan, March 2017, para 4.4
121 IPA, UC Assurance of Action Plan, March 2017, p3
122 IPA, UC Health Check, September 2017, p5
123 IPA, UC Health Check, September 2017, para 6.6
124 IPA, UC Project Assessment Review, November 2011, p23
125 IPA, UC Project Assessment Review, January 2013, p12
126 IPA, UC Project Assessment Review, February 2014, p17
127 IPA, UC Project Assessment Review, October 2015, p13
128 IPA, UC Health Check, September 2017, para 6.7
129 IPA, UC Project Assessment Review, October 2015, p8
130 IPA, UC Health Check, September 2016, para 5.2 and UC Health Check, September 2017, para 6.4
131 IPA, UC Assurance of Action Plan, March 2017, p2 and para 2.17. For example, a case management culture was not yet embedded in UC service centres.
132 Institute for Government, Ministers Reflect series: Lord Freud, January 2017
29 January 2018