Fraud and Error Contents

---

2Measuring and reporting on fraud and error

17.The Counter Fraud Function describes fraud as a hidden crime which you must find before you can fight.41 The government’s counter-fraud functional strategy states an aim to be the most transparent government globally in dealing with public sector fraud.42 Cabinet Office told us it aims to put as much focus into the areas where there are low levels of detected fraud, and low levels of investment in counter fraud, as it does those areas where there are known to be higher levels of loss.43 It publishes an annual Fraud Landscape report every year, which reports on the detected fraud levels across government.44

18.HM Treasury’s guide to Managing Public Money states that that every department should measure and estimate the scale of fraud and error and disclose this in its Annual Report were material.45 The NAO identified the proper measurement of fraud and error as a crucial because it shows departments where controls need to be improved and counter fraud efforts directed.46 We asked HMRC when it expects to have an estimate of fraud and error in the Coronavirus Job Retention Scheme. HMRC told us it does not expect to have a statistically valid estimate until December 2021 as the planned random sampling exercise “will take time to complete” though the evaluation framework for the CJRS was published in December 2020.47 HMRC clarified in a written response to us that it does not intend to measure the rate of fraud and error in the Self-Employed Income Support Scheme and Eat Out To Help Out schemes and will instead use a “blended approach” bringing in a wide range of available evidence.48

19.We asked about DWP’s progress in developing a target for fraud and error. We have previously recommended that DWP analyse the extent to which fraud and error in the system was temporary because of the pandemic and what was due to longer term structural issues.49 DWP explained that COVID-19 has changed the traditional mix of cases within the benefits system as there are now an increased proportion of claimants who are self-employed or who have capital. DWP’s expectation is that this has changed the risk of fraud and error in the system, and as a result DWP could not tell us when it will set a target for reducing fraud. DWP explained that it wanted to wait to update baseline data before it committed a target to ensure it is meaningful but reaffirmed its commitment to providing us with this information in due course.50

20.We asked BEIS about efforts to estimate the level of fraud and error in the Bounce Back Loan Scheme. BEIS told us it has commissioned PwC to undertake a sampling exercise to come up with a specific estimate of fraud which it expects to be completed in May 2021. BEIS explained that this exercise is taking time as it pulls together a lot of data sources to identify high-risk loans and that working through GDPR consequences, for example, “takes care and effort and needs to be done right”. BEIS told us that the interim work performed to date demonstrated that the number of loans that it can pick out as being “absolutely fraud” is “really quite small” and that working through the subtleties that sit behind each application is “not an easy job”.51

21.We asked about what needed to be done to improve the consistency of fraud and error reporting across the public sector. Cabinet Office told us that the introduction of the Functional Standards, a set of minimum criteria for dealing with fraud, is an important tool to increasing consistency of approach.52 It has also agreed a common definition of fraud and a common typology with all Departments reporting against this.53 Cabinet Office told us that the COVID-19 response has left many departments more exposed to higher levels of risk and loss than previously, but we are not aware of any other departments that have made public commitments to measure the extent of fraud and error in their COVID-19 schemes.54

Data sharing

22.Cabinet Office and DWP told us that timely data sharing can be used to prevent fraud by data matching, improve detection of fraud by sharing intelligence, and enable recovery in cross-government schemes.55 The Digital Economy Act 2017 permits data-matching for the purposes of fraud and error.56 Cabinet Office told us that data analytics is a priority for the Counter Fraud Function which has worked with public bodies to deliver 38 data pilots with an impact of over £90 million to date.57 The Counter Fraud Function is also working to establish closer relationships with the Serious Fraud Office, City of London police, and the National Crime Agency. In addition, it told us that an explicit part of the Function’s strategy is to link up to the cyber and security sides of the Counter Fraud Profession to learn more about different avenues of fraud.58

23.In written evidence Cifas, the UK’s largest cross-sector fraud sharing organisation, told us there should be transparency around the businesses that have been in receipt of COVID-19 support scheme funds, and, in its submission to us, the Fraud Advisory Panel agreed that transparency around fraud and error is “imperative” to maintain public trust and confidence.59 The Counter Fraud Function has said it wants to help ensure the UK is the most transparent government globally in how it deals with public sector fraud.60 In December 2020 we recommended that HMRC list details of companies who received support through the CJRS scheme to try and reduce the opportunity for fraud to occur.61 HMRC published details of employers claiming CJRS from December 2020 onwards, but told us there are “limits” as to what details it holds and can publish. HMRC explained that the publication of this information was “really just for public transparency of how public money is being spent”.62 HMRC told us it was not reliant on the public sharing of information to manage fraud risk as it was not an integral part of the fraud management process, although it welcomed any useful information as a result of doing so.63 In response to a parliamentary question about the list of companies offered coronavirus business interruption loans, BEIS stated it was seeing if it could “publish something soon”.64 In written evidence the Fraud Advisory Panel told us publishing data on loan recipients would have been useful as it could have assisted with the early identification of potential instances of impersonation fraud and may have acted as deterrent for would-be fraudsters.65

24.In written evidence Cifas told us that it believed government’s failure to mandate industry standard fraud significantly contributed to the vast scale of fraud in COVID-19 support schemes.66 BEIS told us that it “found cross-Whitehall collaboration incredibly useful” on the Bounce Back Loan Scheme and local authority grant schemes. It used data held by HMRC to retrospectively check applications for the Bounce Back Loan Scheme which helped banks spot potentially fraudulent applications.67 DWP told us that going forward increased data sharing could enable it to identify changes in claimant’s eligibility for benefits and enable changes to be applied across all benefits before a payment is made, preventing errors from occurring and reducing the need for recovery. DWP told us this “ideal” could be achieved using PAYE system data to automatically adjust universal credit payments.68 DWP also explained that it is considering the use of additional powers, for example data sharing with banks, which would have to be done in collaboration across government.69