Universal Credit: progress review inquiry

The Department for Work & Pensions has announced yet another delay to completing the roll-out of Universal Credit, which it attributes to policy changes announced over a year ago. We reported in February 2016 that the Department had again delayed the programme and that it then expected Universal Credit to be fully operational by March 2021. However, on 20 July 2016, the day we took oral evidence on this inquiry, the Department released a written Ministerial Statement setting out further delays to the roll-out of Universal Credit. The Department now plans to continue to roll out its full service to around five jobcentres per month until July 2017 rather than February 2017. This delay means that the Department now expects to complete the roll-out of the programme in March 2022, a year later than when we last reported. The Department maintains that policies introduced in the summer Budget 2015 changed the scope of the programme and required additional time, but these changes were not reflected in the outline business case submitted to HM Treasury in September 2015. The Department denies that it is attributing wider operational problems to policy changes and claims that internal reviews found that without these changes the Department had been on track for scaling up the full service from February 2017. But a flexible system should be able to cope with a degree of policy change. The Department needs to be clearer about what policy change it can incorporate without delay. Given the length of time taken to implement the programme some policy change is likely and we need a clearer picture of the impact on timetable. Recommendation: The Department should explain why its flexible approach to system development has been unable to accommodate policy changes announced in July 2015 and should set out clearly what impact these delays will have on operational costs, staff and claimants on both Universal Credit and legacy systems. This explanation should be provided to the Committee by March 2017.

The Department has not updated its assessment of the expected benefits of Universal Credit in the light of policy and operational changes. The Department has now spent £1.16 billion on implementing Universal Credit, which has a caseload of around 280,000, compared to the over 6 million claimants expected in the long term. Despite having previously estimated that a six month delay to the programme could reduce net benefits to the taxpayer by £2.3 billion, the Department now maintains that the net benefits of the programme have not changed significantly from the £20 billion quoted in its 2015 outline business case. The Department rejected the recommendation we made in February 2016 that it should explain how the business case has changed following changes in policy to Universal Credit and other working age benefits, on the grounds that revising a business case takes four months. However the Department told us that it does have ready-reckoners and is able to model the effect of changes quickly, suggesting that it should have been able to accept our recommendation without causing disproportionate extra work. Recommendation: We reiterate our previous recommendation that the Department should set out clearly the changes to the business case for Universal 6 Universal Credit and fraud and error: progress review Credit since its outline business case in 2015. It should include a brief summary of the policy changes and, using its ready-reckoners, a clear explanation of the impact on the programme’s costs and benefits.

Systems underpinning Universal Credit are still underdeveloped and there are signs of pressure on staff. We welcome the fact that the Department has changed its mind and has now accepted our recommendations and those made by the previous Committee concerning the need for better contingency planning. But the Department still has a long way to go before systems will be ready to scale up Universal Credit significantly; we heard, for example, that only 25% of claims in the new full service are paid automatically. We also received written evidence that staff are concerned about the lack of training and the pressures of work preventing adequate testing and learning within the new service. Recommendation: Before the speed at which Universal Credit is rolled out is increased, the Department should ensure that there are sufficient opportunities for staff to engage in testing and learning processes, and set out for the Committee what it has done to address staff concerns. The Department should write to the Committee to inform it of action taken by May 2017.

Universal Credit’s rigid monthly assessment period causes difficulties for claimants whose pay or rent are based on four-weekly periods. Claimants whose pay or rent cycle does not match the monthly assessment period used for Universal Credit may experience difficulties, such as a drop in payment without warning. Similar issues arise when people are paid early for Christmas. The Department’s only solution appears to be to try and persuade employers and landlords to change their pay and rent practices, rather than seeking to make its own systems more flexible. With the number of employees and landlords the former is unlikely to be feasible. Recommendation: The Department should ensure that claimants whose pay or rent cycles do not align with Universal Credit assessment periods are made aware of this issue and the potential consequences, and are informed of what support is available should this be needed. The Department should also examine what it can do to adapt its systems to cater for these circumstances or provide more information about what it is doing to secure change with employers and landlords.

Neither the Department for Work & Pensions nor HMRC has set meaningful targets for tackling fraud and error. HMRC has set a target for overpayments of tax credits due to fraud and error for 2016–17 of “no more than 5%” of expenditure, which is higher than the 4.8% it is estimated to have achieved in 2014–15. HMRC tells us it is doing everything it can to reduce fraud and error further, so it is disappointing that it is unwilling to back that up with a more stretching target. The Department for Work & Pensions has disaggregated its overall fraud and error figure to set projections in four major areas of benefit expenditure of what it expects to achieve in 2017–18 compared to 2013–14. However, it has not set a proper target for each of these areas. The two departments’ unwillingness to set demanding targets that require improvement on current performance levels seems to imply a lack of confidence in and commitment to their own efforts to improve. Universal Credit and fraud and error: progress review 7 Recommendation: The Department for Work & Pensions and HMRC should set stretching targets for fraud and error across all benefits and tax credits to secure better performance, review these targets annually, and report progress to the Committee.

The Department for Work & Pensions’ understanding of the level and causes of fraud and error in Universal Credit and some other benefits is incomplete, potentially undermining efforts to reduce losses. While the Department expects Universal Credit to reduce fraud and error overpayments by £1 billion a year when it is fully rolled out, initial estimates indicate that the level is currently higher than the Jobseeker’s Allowance that it is replacing. The Department attribute this to the difficulty of developing a suitable methodology to measure fraud and error in Universal Credit, as the new benefit is designed to support both those in work as well as those out of work, and to cases where the Department was unable to contact claimants to verify the payment made. The Department does not regularly measure fraud and error across all its other benefits; for example, fraud and error in the payment of Carer’s Allowance has not been measured for 20 years. Recommendation: The Department for Work & Pensions should: establish and agree with the National Audit Office a robust method for estimating Universal Credit fraud and error; and undertake regular risk assessments to improve its understanding of the causes of fraud and error in those benefits where it has not been measured for some time or at all.

The Department for Work & Pensions estimates that inaccuracies in its information on income and earnings resulted in almost £1 billion of losses in 2015–16. We recognise that both departments now have a better understanding of the major causes of losses and are improving their understanding of the root causes of fraud and error. Both departments have developed initiatives for income and earnings, which is the Department for Work & Pensions’ largest cause of loss, based on the use of Real-Time Information (RTI), where employers with PAYE schemes send details to HMRC every time they pay employees. The Department for Work & Pensions now has a real opportunity to fully exploit RTI data to adjust benefits before they are paid to achieve a meaningful reduction in income and earnings losses. Both departments told us that they were developing initiatives in other areas also, such as working with banks to identify undeclared capital. They also both noted that some areas of loss remained more difficult to tackle, including cohabitation (living together), abroad fraud (where claimants pretend to live in the UK to claim benefits, but are actually living overseas) and earnings not covered by RTI (such as for the self-employed). Recommendation: The Department for Work & Pensions should update the Committee, following the publication in November 2016 of the 2015–16 final fraud and error estimates, on its progress in tackling the largest areas of loss. It should include details of the impact of making full use of RTI in reducing over and underpayments due to errors in income and earnings. 8 Universal Credit and fraud and error: progress review

Too many claimants are unclear about the rules and obligations that apply when claiming benefits or tax credits, leading to high levels of claimant error. Both HMRC and the Department for Work & Pensions referred to the actions they undertake to engage with claimants to make them more aware of their obligations to report changes that may affect their benefits, and cited the opaque definition of “living together” as an example of where claimants may struggle to interpret rules and reporting obligations. We recognise the difficulties claimants face in understanding what changes to report and when to report them. Improving claimants’ understanding of the relevant rules and obligations is essential to reducing claimant error. Work to enhance accuracy of payments and so prevent overpayments from occurring in the first place will also reduce the worrying burden on claimants who later face recovery of erroneous payments. The departments did not accept two of our October 2015 recommendations: to set targets for reducing underpayments and for HMRC to review claimants’ experiences of the tax credits process. We are pleased that the departments have changed their views and now accept these recommendations. Recommendation: The Department for Work & Pensions and HMRC should: report back to the Committee by the end of the year on what has been set for their new underpayments and accuracy targets, and the rationale behind those targets; and undertake research to understand the issues that confuse claimants and take appropriate action to better inform claimants of the rules and their obligations in order to reduce the level of claimant error. Universal Credit and fraud and error: progress review 9 1 Universal Credit

Despite the timetable shift and its own calculations of lost benefits, the Department for Work & Pensions continues to expect total benefits from the introduction of the Universal Credit of around £20 billion. But the Department has not updated its business case since the version that it submitted to Treasury in September 2015, despite significant changes to the design and implementation of Universal Credit since then. The Department has never set out how these changes affect the expected benefits. The Department told us 18 Committee of Public Accounts, Universal Credit: Progress update, Nineteenth Report of Session 2015–16, HC 601, February 2016 19 HM Treasury, Treasury Minutes: Government responses on the fifteenth to twentieth reports from the Committee of Public Accounts: Session 2015–16, Cm 9237, March 2016 20 Written Ministerial Statement, 20 July 2016, HCWS96 21 HC 489, Qq18, 39 22 HC 489 Qq 28, 29, 41, 42 23 Comptroller & Auditor General, Universal Credit: Progress Update, Session 2014–25, HC 786, November 2014, HC 489 Q 30, 24 HC 489, Q 71 25 C&AG’s report, paragraph 2.17 26 HC 489, Q 46 Universal Credit and fraud and error: progress review 13 that, while it had not updated the business case, it used it as the baseline against which to assess and track progress and that it took corrective action when required. It also told us that it takes four months to update a business case, but that it does have a series of ready-reckoners which allows it to “check progress against certain values”.27 We would therefore expect that the Department should be able to quickly model changes to policy and implementation timetables and to update costs and benefits accordingly.

The Institute of Fiscal Studies has estimated that around 3.2 million people will be worse off as a result of the switch to Universal Credit, with around 2.2 million people seeing an increase in their benefit entitlement.28 When asked, the Department was not able to provide its own estimate of the ‘winners and losers’ from Universal Credit. It quoted the IFS analysis, but told us that the IFS analysis did not take account of the changes in people’s behaviour that the Department expects as a result of incentives built into the new system. The Department told us it had produced some estimates itself around 2010–11, but had not updated them since.29 We have seen before how expected behavioural changes may not materialise in practice, for example as we reported in July 2016 on the Department of Energy and Climate Change’s implementation of the Green Deal.30 Ability to scale up

The Department has a significant way to go before its systems will be ready to scale up Universal Credit significantly. The Department told us that only 20–25% of claims in the new full service are paid automatically, and that it needs to be at 60% or 70% before it can increase the volume of claims significantly.31 Combined with the problems the Department faces in incorporating new policy measures, this indicates there are still significant issues for the Department to address with its systems before it can start to increase the pace of the Universal Credit full service roll out. The National Audit Office estimated in 2014 that the Department would save £610 million a year though automated processes, but this suggests it will take some time before such savings can be achieved, with reliance on costly and inefficient manual interventions in the meantime.32

We received written evidence from the Public and Commercial Services Union (PCS) on behalf of Department for Work & Pensions staff. It described the Department’s “test and learn” approach as ‘broken’. The PCS considered that the elements of the test and learn approach, which is supposed to allow staff to test the systems and feed back their views and ideas for improvements, were no longer feasible. The PCS argued that time pressures have resulted in an inability of front-facing staff to assist in ironing out any problems with Universal Credit full service, as fewer staff are able to report what is going wrong or what could be improved though the Department’s Continuous Improvement and Learning programme. The PCS also raised concerns over inadequate training opportunities for 27 HC 489, Qq 48, 50 28 HC 489, Q 58 29 HC 489, Q 59 30 HC 489, Q 60; Committee of Public Accounts, Household energy efficiency measures, Eleventh Report of Session 2016–17, HC 125, 20 July 2016 31 HC 489, Q 23 32 C&AG’s report, paragraph 2.10 14 Universal Credit and fraud and error: progress review staff to familiarise themselves with the updates to the system.33 The Department told us that now the full service was available at 18 out of its 700 jobcentres, it was not able to run these internal learning processes ‘at volume’.34 Services for claimants

The unions USDAW and Unison have raised concerns about inflexibility in the Universal Credit payment systems which may cause hardship for some claimants.35 The Universal Credit assessment takes place monthly, but an in-work claimant’s pay may follow a different pattern (for example four-weekly pay). The Universal Credit system cannot adapt to the possibility of receiving more than one payment within the assessment period and an individual’s benefit reduces accordingly. The Department for Work & Pensions told us that it was aware of these problems and had been working with employers to encourage them to change to a monthly payment cycle.36 The Department also informed us that there were similar issues with housing payments where they occur on a four-week cycle. The Department agreed to look into its approach in this area and look at what advice was available to jobcentre staff to support people with non-monthly pay arrangements.37

The Department expects people who are in work that earn less than the equivalent of 35 hours per week at the minimum wage to look to work or earn more. This “in-work conditionality” regime is still at a very early stage of development and the Department is undertaking a national trial to see what the best ways are of intervening. Approximately 40% of the current Universal Credit caseload are in work (approximately 112,000 claimants) and are moving into the trial.38 But longer term, the majority of households likely to fall within these requirements will be the 4.4 million families currently in receipt of tax credits, who are not used to such conditions being attached to their entitlement.39 These requirements may lead to families being ‘sanctioned’, or facing a financial penalty, if they cannot demonstrate that they have been looking to increase earnings during their assessment month. The Department stressed to us that the idea of this is to encourage people to work more hours and increase their earnings, not to be a system of punishment, but the Department must be sensitive to individual families’ circumstances (for example varying shift patterns and overtime requests) if the system is to prove effective. The Work and Pensions Select Committee has looked into this area in depth and we will also continue to take an interest in this area as plans develop and in work claimant numbers increase.40 33 HC 489, Q 25; PCS Union, UCR0003, July 2016 34 HC 489, Q 27 35 HC 489, Qq 79–81 36 HC 489, Q 79 37 HC 489, Q 83 38 HC 489, Q 87 39 Comptroller and Auditor General, HM Revenue & Customs 2015–16 Accounts: Report by the Comptroller and Auditor General, Session 2015–16, July 2016 40 HC 489, Q 88; Work and Pensions Select Committee, In-Work progression in Universal Credit, Tenth report of Session 2015–16, HC 549, March 2016 Universal Credit and fraud and error: progress review 15 2 Fraud and error

In September 2015 we took evidence from the Department for Work & Pensions and HMRC on how they were tackling fraud and error in benefits and tax credits. We published our report in October 2015.41 The Comptroller and Auditor General has given qualified opinions on the Department for Work & Pensions’ accounts since 1988–89, and on HMRC’s accounts since 2003–04, because of the rates of fraud and error. In our October 2015 report we commented on the two departments’ apparent lack of ambition in tackling this significant and longstanding problem. The departments’ Treasury Minute response to our recommendations did nothing to counter this impression and we therefore recalled the departments to give further evidence in July 2016.42 Progress and targets

HMRC is responsible for administering tax credits. In 2015–16 it paid £28.2 billion supporting around 4.4 million families and 7.4 million children. The Department for Work & Pensions manages most other benefits, and the State Pension, payments of which amounted to a combined £173.4 billion to some 18 million people in 2015–16. While both departments continue to pursue initiatives to reduce fraud and error there has been no step change in the levels of over or underpayments. The latest figures available (2015–16 for the Department for Work & Pensions and 2014–15 for HMRC) show that the departments still overpaid claimants by a combined £4.5 billion and underpaid by £2 billion.43

HMRC said it continued to strive to reduce fraud and error as much as possible. But its target for overpayments of tax credits due to fraud and error is now “no more than 5%” of expenditure for 2016–17—a level above the 4.8% (£1.37 billion) already achieved in 2014–15.44 The Department for Work & Pensions reported overpayments of 1.8% (£3.1 billion) of benefit expenditure for 2015–16. The Department for Work & Pensions has disaggregated its overall overpayments due to fraud and error figure to set separate projections in four major areas of benefit expenditure for what it expects to achieve in 2017–18, compared to 2013–14. The Department told us that, whilst it had set projections for four benefit groups (Universal Credit, working-age legacy benefits (those before Universal Credit), pensioner benefits and disability/carer benefits), it did not see the projections as targets given the difficulties of projecting fraud and error for benefits groups at the same level of precision as for aggregate fraud and error across all benefits.45

We are concerned at the lack of stretch in setting targets without due regard to current achievement levels and the lack of confidence and commitment this implies in the departments’ own efforts. Agreeing meaningfully stretching targets and reviewing these annually would drive better performance in the administration of tax credits and benefit payments that HMRC and DWP are responsible for and reduce the high levels of over and underpayments. 41 Committee of Public Accounts, Fraud and Error Stocktake, Fourth Report of Session 2015–16, HC 394, 28 October 2015 42 HM Treasury, Treasury Minutes: Government responses on the fourth to the eighth reports from the Committee of Public Accounts: Session 2015–16, Cm 9190, 21 January 2016 43 Committee of Public Accounts, Oral evidence: Fraud and error stocktake: progress review, HC 490, Q56 44 HC 490, Qq 6, 47, 48 45 HC 490, Qq 40, 41 16 Universal Credit and fraud and error: progress review Understanding causes

Initial estimates show that the levels of fraud and error in the Universal Credit live service are higher than fraud and error in Jobseeker’s Allowance. Although not an exact like for like comparison, the 2015–16 estimates show 7.3% overpayments and 2.6% underpayments for Universal Credit, compared to 5.0% overpayments and 0.8% underpayments in Jobseeker’s Allowance.46 The Department for Work & Pensions explained that it faces a challenge to develop a suitable methodology to measure fraud and error in Universal Credit as it supports those in and out of work, compared to Jobseeker’s Allowance which is solely for those who are out of work. In particular the Department cited the impact of ‘loss of claimant contact’, where it had not been able to make contact with a number of Universal Credit claimants to verify payments.47 As it could not confirm entitlement to benefit was correct during its measurement process the Department told us that such ‘lost contact’ formed a key category of overpayments in the first fraud and error estimates for Universal Credit that the Department would need to undertake further work to fully understand. Loss of contact may not indicate fraud or error, individuals now in work may be harder to contact or consider their claim to have ceased.48

The Department for Work & Pensions told us that it continues to expect that under Universal Credit fraud and error overpayments will come down by around £1 billion each year, once it is fully rolled out.49 The Department told us that it would increase the number of claims subject to fraud and error measurement in line with the growth in the volume of Universal Credit cases.50

We were concerned that DWP’s understanding of the level and causes of fraud and error in Universal Credit and some other legacy benefits is incomplete, particularly where benefits have never been measured for fraud and error or have not been measured recently. As well as Universal Credit having higher than expected fraud and error the Department also does not regularly measure fraud and error across all its benefits, for example Carer’s Allowance has not been measured for 20 years and the population has changed over this long time period.51

We recognise that the departments have improved their understanding of the major causes of losses due to fraud and error. Both departments have developed initiatives for underreported income and earnings—which is the Department for Work & Pensions’ largest cause of loss resulting in overpayments of £951 million in 2015–16–based on the use of RTI, where employers send details to HMRC every time they pay employees.52 The Department told us in its written submission before our evidence session that it had made significant strides in tackling overpayments related to earnings and employment through the use of RTI,53 though this is not yet reflected in a reduction in income and earnings related fraud and error. 46 HC 490, Qq 11, 27 47 HC 490, Qq 20, 53 48 HC 490, Q 25 49 HC 490, Qq 9, 18 50 HC 490, Q 29 51 HC 490, Qq 69, 70 52 HC 490, Q 9, 11 53 HC 490, Q 30 Universal Credit and fraud and error: progress review 17

We asked the Department for Work & Pensions why fraud and error rates had increased for Pension Credit, where overpayments are now 5.6%. The Department told us that it was now able to use RTI data for cleansing Pension Credit cases and it believed that RTI was making a difference. However, the Department admitted that during the year it had changed the way it prioritised working through the data from RTI, which had led to a reduction in the amount of fraud and error identified in Pension Credit claims using RTI data. The Department also said that it had not yet optimised the use of RTI data and that it had more work to do with this.54

Both departments told us that they were also developing initiatives in other areas of loss, such as working with banks to identify undeclared capital.55 They also both noted that some areas of loss remained more difficult to tackle, including cohabitation, abroad fraud (where claimants pretend to live in the UK to claim benefits, but are actually living overseas) and earnings not covered by RTI, such as for the self-employed.56 Impact on claimants

The departments did not accept two of the recommendations in our October 2015 report—to set targets for reducing underpayments and for HMRC to review claimants’ experiences of the tax credits process.57 We are pleased that the departments have finally revised their positions and will now set targets for reducing underpayments, a previously neglected issue, and review claimants’ experiences of tax credits. The Department for Work & Pensions in its written submission to us ahead of the evidence session agreed to set a global target for underpayments which it aims to have completed by the end of September 2016.58 HMRC committed to working collaboratively with the Department for Work & Pensions where possible over the summer months and to produce options to discuss and agree with Ministers.59 We look forward to sufficiently stretching targets being set for both departments to properly galvanise work that improves accuracy of payments and prevents underpayments in benefits and tax credits from occurring, so that households are properly supported with payments to which they are entitled.

HMRC told us that it was working with the Department for Work & Pensions to interview claimants to understand their experience of moving from tax credits to Universal Credit.60 The departments have an opportunity to gather meaningful feedback from this research and use it to ensure claimants continue to receive support during migration, whatever their circumstances. Both HMRC and the Department for Work & Pensions told us of actions they undertake to engage with claimants to make them more aware of their obligations to report changes that may affect their benefits so that fraud and error levels are reduced further.61 The Department for Work & Pensions told us in its written submission that claimants usually make mistakes because they do not understand what is being asked of them or the Department makes it too hard for claimants to access the Department’s services. The Department plans to undertake work to improve the quality 54 HC 490, Q 30 55 HC 490, Qq 60, 61 56 HC 490, Qq 7–9, 14, 30 57 HM Treasury, Treasury Minutes: Government responses on the fourth to the eighth reports from the Committee of Public Accounts: Session 2015–16, Cm 9190, January 2016 58 HC 490, Qq 71, 72 59 HM Revenue & Customs, FES0004, July 2016 60 HC 490, Q 71; HM Revenue & Customs, FES0004, July 2016 61 HC 490, Qq 4, 10 18 Universal Credit and fraud and error: progress review of communications and ensure content is clear so that claimants understand the actions they have to take. HMRC also told us that it planned work so that tax credit claimants were clearer on when they needed to make a joint claim with another person as opposed to a single claim.62 Both departments cited the opaque definition of “living together” as an example of where claimants may struggle to interpret rules and reporting obligations.63 62 HM Revenue & Customs, FES0004, July 2016 63 HC 490, Qq 7, 14 Universal Credit and fraud and error: progress review 19