Select Committee on Public Accounts Minutes of Evidence



APPENDIX 2

Supplementary memorandum submitted by the Department for Work and Pensions in response to additional questions from the Committee

Question 1: What is the Department's national estimate, on the basis of a) validated data; and b) the further cases which the Area Benefit Review for each year identifies as i) "highly suspected of fraud"; or ii) having "a low suspicion" of fraud; of the amounts of income support overpaid due to fraud, customer error and official error in the years: 1997-98; 1998-99; 1999-2000; and 2001-02.

INCOME SUPPORT (IS)—LOSSES FROM FRAUD AND ERROR

Type of loss

1997-98

1998-99

1999-2000

2000-01

 

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

Fraud

596m

5.10%

549m

4.70%

559m

4.40%

573m*

4.30%

Customer Error

131m

1.10%

140m

1.20%

147m

1.20%

154m*

1.10%

Additional Official Error

173m

1.50%

171m

1.30%

184m

1.50%

173m*

1.30%

Fraud and Error

900m

  7.7%

860m

7.20%

890m

7.10%

900m  

6.70%

 

  * The figures quoted above are those that appear in the National Statistics published on 22 February 2002 and are marginally different to previously quoted figures. The variation is a result of refinement of the data.

HIGH SUSPICION AND LOW SUSPICION OF FRAUD IN INCOME SUPPORT 1997-98 TO 2000-01 (AREA BENEFIT REVIEW ESTIMATES)

 

1997-98

1998-99

1999-2000

2000-01

 

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

High Suspicion
of fraud

not
available

not
available

236m

2.00%

206m

1.60%

125m

0.90%

Low Suspicion
of fraud

not
available

not
available

365m

3.10%

363m

2.90%

287m

2.10%

 

  These figures relate to cases that may be incorrect. In the absence of evidence of fraud, they are correctly assessed and do not contribute to the overall estimates of fraud and error.

Question 2: What is the Department's total estimate of the amount of Jobseeker's Allowance overpaid due to fraud, customer error and official error (using the same criteria as before)?

JOBSEEKER'S ALLOWANCE (JSA)—LOSSES FROM FRAUD AND ERROR

Type of loss

1997-98

1998-99

1999-2000

2000-01

 

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

Fraud

302m

8.40%

291m

8.20%

266m

8.20%

201m

7.00%

Customer Error

24m

0.70%

22m

0.60%

21m

0.70%

11m

0.40%

Additional Official Error

154m

4.20%

197m

5.60%

133m

4.00%

  78m

2.80%

Fraud and Error

480m

13.20%

510m

14.4%

420m

12.90%

290m

10.20%

 

HIGH SUSPICION AND LOW SUSPICION OF FRAUD IN JOBSEEKER'S ALLOWANCE 1997-98 TO 2000-01 (AREA BENEFIT REVIEW ESTIMATES)

 

1997-98

1998-99

1999-2000

2000-01

 

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

Amount

As % of
spend

High Suspicion
of fraud

not
available

not
available

111m

3.10%

  77m

2.40%

28m

1.00%

Low Suspicion
of fraud

not
available

not

available

137m

3.90%

135m

4.20%

80m

2.80%

 

  These figures relate to cases that may be incorrect. In the absence of evidence of fraud, they are correctly assessed and do not contribute to the overall estimates of fraud and error.

Question 3: What is the Department's total estimate of the amount of Housing Benefit overpaid due to fraud, customer error and official error?

HOUSING BENEFITLOSSES (ESTIMATES) FROM FRAUD AND ERROR (NATIONAL BENEFIT REVIEW)

Benefit Reviews

Type of loss

Estimated loss
(m)

% of claims

Sampling In
(Year)

Housing Benefit

Fraud

180m

1.9%

1997-98

 

Customer error

260m

6.6% (excludes
underpayments)

 
 

Official error

60m

1.5% (excludes
underpayments)

 
 

Total

500m

   

 

Question 4: What is the Department's total estimate of the amount of all benefits overpaid due to fraud, customer error and official error?

 

ESTIMATING FRAUD

Losses in Income Support and Jobseekers Allowance The Department has, since 1998, run a continuous exercise to measure levels of fraud and error in Income Support and Jobseekers Allowance. The latest validated figures cover the year 2000-01. A similar exercise is now running in respect of Housing Benefit. The first results will be available next year.

Losses in other major benefits In the early to mid 1990s losses from most major benefits were measured either by "snapshot" full measurement exercises, known as National Benefit Reviews (NBRs), or in the case of Family Credit, Incapacity Benefit and Attendance Allowance by a pilot exercise. The figures obtained are not a reliable guide to current levels of fraud and error because of changing levels of spend on the benefits themselves, changes to the benefit rules and changes to the way in which the measurement of fraud and error is conducted.

In particular, the Department has changed its approach to the recording of cases of "suspicion of fraud." Following criticism from the NAO, we now make much greater efforts to resolve cases one way or the other. Consequently, some of these cases are now classified as confirmed fraud but many cases that would have been recorded as suspicious are now cleared altogether. While, in respect of IS and JSA, we continue to collect data on cases recorded as highly suspicious for management purposes, we believe that the figures for confirmed fraud are the best guide. They are based on cases where benefit has actually been stopped or reduced as a result of a rigorous fraud investigation.

Providing an up-to-date estimate of the levels of fraud in benefits that were covered only by the older NBRs and pilots is therefore complicated and speculative. The estimates perforce rest on the assumption that the real underlying levels of fraud have not changed, which is not the case given the increased efforts by the department to tackle the problem. It is then necessary to adjust the original figures to take account of benefit uprating, benefit rule changes affecting levels of the benefit spend and changes in our measurement methodology.

Losses in other benefits Estimates of fraud levels in benefits where no measurement exercise has taken place can be made, based on the overall fraud rate detected in the measured benefits.

Total fraud losses A calculation on the basis described above yields the following figures though it will be evident that this is subject to significant margins of error:

Continuously measured benefits

   

Income Support

 

573m

Jobseeker's Allowance

 

201m

Total

 

774m

National Benefit Review in 2000 (no adjustment required)

   

Incapacity Benefit

 

19m

Other benefits—adjusted figures derived from older NBRs and AA pilot

 

(Disability Living Allowance, Attendance Allowance, Retirement Pension, Invalid Care Allowance, Child Benefit, Housing Benefit)

   

Total (rounded)

 

1,100m

Estimate for unmeasured benefits

   

Total (rounded)

 

190m

Total (rounded)

 

2,000m

ESTIMATING LEVELS OF CUSTOMER AND OFFICIAL ERROR

  It is similarly possible to provide estimates for the losses attributable to Customer Error and Official Error, though again there will be wide margins of error:

Continuously measured benefits

   

Customer Error

   

Income Support

 

154m

Jobseeker's Allowance

 

11m

Total

 

165m

Official Error

   

Income Support

 

173m

Jobseeker's Allowance

 

78m

Total

 

251m

National Benefit Review of Incapacity Benefit (no adjustment)

   

Customer Error

 

16m

Official Error

 

22m

Other benefits—adjusted figures derived from older NBRs and AA pilot

   

Customer Error Total (rounded)

 

400m

Official Error Total (rounded)

 

100m

Estimate for unmeasured benefits

   

Customer Error Total (rounded)

 

60m

Official Error Total (rounded)

 

40m

Total (rounded)

 

1,000m

ESTIMATE OF TOTAL LOSSES

  On the basis of the above, the Department estimates that the total amount of all benefits overpaid due to fraud, customer error and official error is in the region of 3 billion a year.

EARLIER ESTIMATES OF TOTAL FRAUD

  The 1998 Green Paper "Beating Fraud is Everyone's Business: Securing the Future" gave estimates of confirmed fraud (2 billion); high suspicion of fraud (3 billion); and low suspicion of fraud (2 billion); a total potential loss of 7 billion if all the suspicions of fraud were well-founded. The Green Paper drew on the results of NBRs and pilots available at the time. The estimates served to highlight the seriousness of the fraud problem but there have been significant changes subsequently and they no longer provide a useful guide. Since then:

    —  Family Credit (estimated then to have contained some 730 million fraud in all three categories) has ceased to be a benefit;

    —  There have been major changes in the way fraud measurement is conducted, as detailed above, which means that fewer cases of suspicion of fraud would now be recorded and it would be a less compelling indicator of the true level of fraud.

  The Department now bases it overall estimate of fraud levels on what is known about confirmed fraud. It is misleading to speculate on overall levels of suspected fraud.

Department for Work and Pensions

March 2002

 


 
previous page contents

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2002
Prepared 11 September 2002