Fraud and Error Stocktake Contents

1 Targeting fraud and error

1. On the basis of a report by the Comptroller and Auditor General, we took evidence from the Department for Work & Pensions (DWP) and HM Revenue & Customs (HMRC) on how they were tackling fraud and error in benefits and tax credits.1 Fraud and error in these systems is a significant and long-standing problem. The Comptroller and Auditor General has given qualified opinions on DWP’s accounts since 1988–89, and on HMRC’s accounts since 2003–04, because of the rates of fraud and error. The Comptroller and Auditor General explained that he had exercised discretion where possible, and been prepared to look at fraud and error in individual benefits. He had added commentary to his reports on accounts to recognise positive developments. He stressed, however, that he was ultimately limited by statute, and that he had to qualify accounts with material irregularities. He said that he would consider lifting the qualification if he was satisfied that departments had reduced fraud and error down to the lowest level feasible, as opposed to just reducing it to what the departments found convenient. However, each department doubted it would be able to reach that point.2

2. HMRC manages tax credits and paid out £29 billion to 4.7 million claimants in 2013-14. DWP manages most remaining benefits and the State Pension, paying out £164 billion in 2013–14 to 18 million people. Since 2010, both departments have made some progress in reducing headline rates of fraud and error. However, in 2013–14, DWP and HMRC still overpaid claimants by a combined total of £4.6 billion. Benefits and tax credits fraud and error costs every household in the UK around £200 each year, and reduces the public resources available for other purposes.3

3. HMRC has reduced fraud and error from 8.1% of expenditure in 2010–11 to 4.4% in 2013–14, which is the lowest level since tax credits were introduced.4 It acknowledged, though, that there had been scope for improvement because historically the rate of fraud and error in tax credits had been high. HMRC explained that it had achieved the reduction through focusing its interventions on high-risk claimants, more than doubling returns from these checks, and increasing the number of interventions from around 100,000 to more than 2 million a year. HMRC has also introduced new reporting requirements for claimants, such as regular updates from those with high childcare costs.5

4. HMRC told us that it was trying to understand fully the reasons for recent reductions in tax credits fraud and error, and that it still had more to learn. For example, in June 2015 HMRC revised its 2012–13 estimates of fraud and error from 7.0% to 5.3%, a reduction of £490 million. HMRC told us that it has identified some of the underlying reasons for why these estimates had to be restated, but that it needed more analysis to really understand the detail.6

5. HMRC said it was confident that there is an underlying downward trend in tax credits fraud and error. However, the current rate remains higher than for many DWP benefits, and HMRC admitted that it had some way to go to reduce tax credit fraud and error to the 3.7% achieved in DWP working-age benefits in 2014–15.7 HMRC has not set a target for reducing fraud and error further during this Parliament, nor was it willing to state how low it could get fraud and error. HMRC claimed that it was not the right time to set future targets as it was currently working towards the spending review in the autumn. However, it admitted that the level of future reductions in fraud and error was not a specific negotiating factor in its spending review discussions with HM Treasury.8

6. In 2010, DWP set a target to reduce fraud and error from 2.1% of benefit expenditure in 2010–11 to 1.7% by 2014–15. DWP told us it had not met this target.9 Its preliminary figures for 2014–15 show that fraud and error was 1.9%, even though changes in the mix of benefits since 2010 had helped bring down overpayments. DWP explained that the headline figure is affected by both the relative size of spending in different benefits and by the level of incorrectness. For example, spending on State Pension, which has a low fraud and error rate (0.2%), has increased from 46% (£70 billion) of benefit expenditure in 2010–11 to 51% (£87 billion) in 2014–15. This change alone has reduced the overall rate of DWP fraud and error by around 0.2% between 2010–11 and 2014–15.10

7. DWP acknowledged that an aggregate measure of fraud and error can mask the effects of various underlying factors. For example, DWP has achieved reductions in some benefits, such as Jobseeker’s Allowance, where overpayments fell from 6.1% to 4.8% between 2010–11 and 2014–15, but these have been offset by increases in Housing Benefit fraud and error, which rose from 4.2% to 5.7% over the same period. Furthermore, DWP said that some of its planned initiatives had not delivered the savings expected, which along with the delays to Universal Credit have hindered it meeting its fraud and error target.11 DWP agreed with the need to monitor its progress at a more granular level and has started to develop strategies for individual benefits. It also accepted that there was merit in setting separate targets for each benefit, but has not done so yet.12

8. Both departments stressed that welfare reforms, such as the roll-out of Universal Credit and introduction of the New State Pension, should reduce complexity in the benefits system, and tackle some causes of fraud and error.13 However, significant fraud and error risks will still remain, particularly around verifying claimants’ income and living arrangements. DWP estimates that total benefits and tax credits fraud and error will rise from £4.6 billion in 2013–14 to £5.8 billion in 2020–21, but that Universal Credit will save only £0.5 billion each year in fraud and error overpayments when fully rolled-out.14

9. In 2013–14, fraud and error due to claimants misreporting their income was
£1.4 billion.15 The departments said that real-time information had already helped reduce fraud and error resulting from misstated income, and will continue to do so. However, real-time information will not eradicate all instances of claimants misreporting their income. HMRC expects that real-time information will save £410 million on tax credits over the next two years, and DWP expects it will save £356 million from using real-time information between 2015–16 and 2020–21.16 HMRC has worked hard with employers to clean up the real-time information they provide, but it told us that errors and omissions would still occur in the data.17

10. More than £800 million was overpaid in 2013–14 because of claimants not reporting that they had a partner.18 DWP said that HMRC has been more successful in tackling overpayments arising from claimants’ misreporting their living arrangements, and it hoped to learn from HMRC’s approach. For example, HMRC has used data from credit reference agencies on top of in-house risk assessments, which DWP said has had good results. However, both departments admitted that they continue to struggle with this huge challenge.19

11. In-year overpayments and underpayments are inherent in tax credits. They arise because the system is designed to better support families by using an estimated award at the beginning of the year. HMRC then finalises payments and makes any necessary adjustments at year-end, once actual household income and circumstances are known. Where finalisation shows the original award was over-estimated, HMRC seeks to recover the overpayment from the claimant. HMRC does not regard these overpayments as fraud and error.20

12. Universal Credit will eliminate such in-year overpayments when it replaces tax credits, as it calculates payments based on claimants’ real-time income and circumstances. For forecasting purposes, DWP included tax credits in-year overpayments alongside fraud and error in a combined baseline it produced for the 2014 Autumn Statement. This forecast, which predates the release of HMRC’s lower-than-expected 2013–14 fraud and error statistics, shows a projected reduction in total overpayments from 4% of aggregate benefit expenditure in 2013–14 to 2.5% by the end of this Parliament.21 However, the Comptroller and Auditor General’s Report shows that the 2013–14 baseline is 2.8% when it includes only actual fraud and error overpayments, not 4%, so the forecast fall to 2.5% in 2020–21 is much smaller.22 DWP accepted that, with more effort, it could maybe take fraud and error below 2.5% by 2020–21.23

1 C&AG’s Report, Fraud and error stocktake, Session 2015-16, HC 267, 21 July 2015

2 Qq 156-162; C&AG’s Report, paragraph 1.5

3 Qq 6, 168; C&AG’s Report, paragraphs 1.1, 1.4

4 C&AG’s Report, paragraphs 2.6 and 2.10

6 Q 2; C&AG’s Report, paragraph 2.10

9 Q 18; C&AG’s Report, paragraphs 2.4, 2.7

10 Q 10; C&AG’s Report, paragraphs 2.13-2.15, 2.17 and Appendix Three

11 Qq 10, 12, 20; C&AG’s Report, paragraphs 2.7, 2.12

15 C&AG’s Report, Figure 10

18 C&AG’s Report, Figure 10

20 Q 37; C&AG’s Report, paragraph 3.7

21 Qq 13, 24; C&AG’s Report, paragraphs 2.9-2.10, Figure 8

22 C&AG’s Report, paragraph 3.6

© Parliamentary copyright 2015

Prepared 26 October 2015