Guidance

Changes in the fraud and error in the benefit system: financial year 2019 to 2020 estimates: statistical notice

Published 16 April 2020

Introduction

This statistical notice is about changes to the fraud and error in the benefit system publication.

The changes in this statistical note will take effect for the 2019 to 2020 publication, due for publication on 14 May 2020.

The changes are:

• applying a consistent assumption for ‘Cannot Review’ cases in line with the assumption developed for Universal Credit
• removing the Arrears Advance adjustment for Housing Benefit cases
• changing how we deal with multiple errors on Housing Benefit that are linked, to more accurately attribute the errors across type and reason
• splitting multiple whole award errors equally between error reasons
• removing the Net Programme Value adjustment
• netting off Official Error from Claimant Fraud and Claimant Error on Housing Benefit, Employment and Support Allowance, Pension Credit and Jobseekers Allowance
• standardising grossing, with all benefits now grossed on a monthly basis at a national level
• changing the expenditure data used in the estimates to match published benefit expenditure figures

We are making these changes because some of our assumptions have been in place for a number of years, and we were asked by the National Audit Office (NAO) to review certain assumptions in recommendations relating to their audit of the Department for Work and Pensions (DWP) 2018 to 2019 annual report and accounts.

The changes will lead to improved accuracy of the fraud and error statistics and better understood and fully documented methodologies.

Revised figures for 2018 to 2019 using the new methodology will be published within the fraud and error in the benefit system: 2019 to 2020 estimates (due for publication on 14 May 2020).

This will allow a like-for-like comparison with the newly published figures for 2019 to 2020. Due to the amount of resource required to apply the methodology changes to data prior to 2019 to 2020 we are unable to provide a back series beyond 2018 to 2019.

Further detail on these changes will be presented in the background information and methodology note that will be published with the fraud and error in the benefit system: 2019 to 2020 estimates.

If you have any queries on these changes send them to: caxtonhouse.femaenquiries@dwp.gsi.gov.uk

Background

DWP uses a complex methodology to measure the levels of fraud and error in the benefit system. Read more information on the methodology used.

Applying a consistent assumption for ‘Cannot Review’ cases in line with the assumption developed for Universal Credit

Cannot Review cases are those that do not engage with the Performance Measurement review of their benefit award, resulting in their benefit being suspended. We previously classified all these cases as fraud. We now look at each of these cases individually and classify them as follows:

• Benefit correct – if they come back on to benefit within 4 months with the same circumstances
• Fraud – if they come back on to benefit within four months with circumstances that have changed
• Inconclusive – if they do not return to benefit within 4 months

The rolling out of this assumption was requested by NAO after we introduced it for Universal Credit last year with their approval.

Inconclusive cases are removed from the main figures and footnoted.

Removing the Arrears Advance adjustment for Housing Benefit cases

This adjustment (which only affects Housing Benefit) was previously applied to account for cases that were paid in arrears or advance, as these cases could have more (or less) error at the time the payment relates to, and may have correctly informed the Department of changes after the benefit review.

The adjustment is extremely complex. It results in a large increase in the number of errors in our data (increasing by over ten-fold). In some cases, trends in the raw data do not carry through to the published figures, which makes understanding and explaining the changes in the publishes figures more difficult.

A Management Letter Point from NAO requested that we review this adjustment.

It has the biggest effect on Fraud – we believe it is incorrect to apply this to Fraud as, by their very nature, fraudulent cases would not report changes to us, so it would be highly unlikely they would be corrected by the time the payment relates to.

The adjustment is based on the error start date being correct and known. Performance Measurement do everything they can to make sure it is, but for certain error types this is not possible.

Other changes

We have made a number of other methodological changes, as follows.

Changing how we deal with multiple errors on Housing Benefit that are linked, to more accurately attribute the errors across type and reason.

Splitting multiple whole award errors equally between error reasons: Previously, if a case had multiple whole award errors, one would be picked at random, and the whole value given to that error reason, whilst the others would be set to zero (as the error on a case cannot be bigger than what the person was receiving). This means that we would have potentially undercounted the amount that each error reason contributes to the total. Moving to equally distributing the value of the award across the error reasons in these cases will give us a better reflection of which reasons are driving the Fraud/Error on benefits.

Removing the Net Programme Value adjustment: This adjustment was done on a small proportion of cases last year (10 out of around 15,000) that had errors relating to Living Together. The adjustment reset the value of the error to be the difference in what the DWP benefit amounts the claimant and partner were getting before the review and what they were getting after the review. The issue with this calculation relates to Tax Credits, which are HMRC benefits but going forward will be administered by the DWP as part of Universal Credit. The problem arises in the scenario where a claimant (or their partner) was on a legacy benefit before a review and was receiving Tax Credits, then moved to Universal Credit after a benefit review. In these cases, their Tax Credits amount would be not counted in the amount of DWP benefits being received beforehand (as it is not a DWP benefit) but would be afterwards. This situation would lead to an incorrect fraud underpayment being created.

Netting off Official Error from Claimant Fraud and Claimant Error on Housing Benefit, Employment and Support Allowance, Pension Credit and Jobseekers Allowance: When doing a review, Performance Measurement check for Official Error before they check for Fraud and Claimant Error (i.e. the check is in a separate period). This is why, previously, Official Error was not netted off Claimant Fraud and Claimant Error. However, after looking in more detail at last year’s errors, we were able to see that over 99% of the Official Errors carried on into the period where we reviewed the case for Fraud and Claimant Error. This means we should be able to net them off the Fraud and Claimant Error to give a more accurate reflection of the impact of each case to the department.

Standardising grossing, with all benefits now grossed on a monthly basis at a national level: Universal Credit was grossed on a monthly basis, due to it being a new benefit and the number of people (and the expenditure) on Universal Credit constantly growing month by month. Since Universal Credit is replacing most of the legacy benefit, the number of people (and the expenditure) on those benefits will be decreasing at a similar rate. We have therefore adopted the Universal Credit approach to grossing across the other benefits.

We have also changed the expenditure data used in the estimates, to match published benefit expenditure figures. Previously we used internal monthly forecasts, but we have changed to align with published expenditure figures. The 2018 to 2019 expenditure figures are now from the Spring Statement 2019 forecast. Future estimates will include expenditure figures from the annual March fiscal event.

Effect of the methodological changes

Adopting the methodological changes results in small changes in overpayments at a global (overall) level and, for some benefits, a slightly larger change at a benefit level. There is very little change in underpayments.

Issued by

Fraud and error measurement and analysis team, DWP

Telephone: 0113 251 9904
Press Office: 020 3267 5144

Statistician

Richard Goulsbra
Fraud and error measurement and analysis team
Department for Work and Pensions
Quarry House
Quarry Hill
Leeds
LS2 7UA

Telephone: 0113 251 9904
Email: caxtonhouse.femaenquiries@dwp.gsi.gov.uk