Select Committee on Social Security Second Report



16. ANALYSIS OF FINDINGS

The results of the exercise.

Introduction

16.1  This chapter presents the analysis of the FamC pilot sample. It presents the raw sample findings overall and by client group and where appropriate weighted sample results are given. The findings are limited for the pilot exercise, the small sample not supporting a high degree of detail or allowing the extrapolation of results to national level with margins of error. However, analysis has been undertaken to inform what could be identified from a main review. It must be borne in mind that these results are indicative, that is they are true of the selected sample, but may not be an accurate reflection of the whole FamC liveload. They should therefore be used with care.

16.2  The population was stratified into two groups, with two thirds of the sample selected from employed cases and one third from the self employed, totalling 300 cases. This stratification was taken into account when producing estimates of overall proportions of incorrect cases. Self employed cases made up only 13.9 per cent of the national caseload at the time of the sample, but were oversampled to produce sufficient cases for analysis purposes. This deliberate bias in the sampling means that raw sample results cannot be taken as representative of the overall picture without weighting the client groups together.

16.3  This chapter is arranged in the following way:-

1Paragraphs 16.4-16.13 summary of the main findings of the analysis
2Paragraphs 16.14-16.15 raw sample results
3Paragraph 16.16 the results in terms of the proportions of overall caseload currently receiving incorrect benefit as a result of fraud or error.
4Paragraph 16.17 the reasons for fraud, high suspicion of fraud and error.

For sections 3 and 4 the main features of the results are highlighted, the relevant results are presented in tabular form and any necessary explanation of the tables is noted.

Summary of the Main Findings

16.4  Weighted to represent the national caseload, 53 per cent (150 out of 298) of the cases have the correct level of benefit in payment. This includes 6 customer error and 3 departmental error cases which did not change the level of benefit in payment and on which no frauds or suspicions of fraud were recorded.

16.5  The rates of fraud and high suspicion were found to differ between the two client groups, for self employed claims the fraud rate was 14 per cent and high suspicion 21 per cent, compared with 9 per cent fraud and 15 per cent high suspicion for employed claims. However, the small sample does not allow us to test if this difference is statistically significant. Weighting the results to reflect the national caseload proportions of employed/self-employed gives an overall fraud rate of 10 per cent and a high suspicion rate of 16 per cent.

Correct

16.6  There were 141 cases on which no frauds or errors were found. These are classified as 'No change'; current benefit on these cases was correct. There were 9 further cases that had a department error or customer error which did not change benefit, and on which no frauds or suspicions ere recorded. This meant that in total 150 cases had the correct level of benefit in payment.

Customer Error

16.7   Weighting the results to reflect the national caseload composition, customer error was found on 10 per cent of cases with little difference between the client groups.

16.8  The main reasons for customer error across both client groups were earnings related, either earnings customer, earnings partner or earnings employer; over 80 per cent of self employed reasons and 75 per cent of employed reasons.

Departmental Error

16.9  Weighting the results to reflect the national caseload composition, department error was found on 9 per cent of cases with a small difference between the client groups; self employed: 7 per cent, employed: 9 per cent.

16.9  The main reason for department error across both client groups was 'mistake as to material fact' related to earnings; 78 per cent of errors on employed cases and 100 per cent of errors on self employed cases.

Fraud

16.9  Reasons for fraud are predominately earnings-related, though other frauds were also found. Earnings frauds made up 70 per cent of the employed group reasons for fraud and 60 per cent of the self employed group fraud reasons. Other frauds detected included claiming as a single person while living as part of a couple (LTAHAW), capital, other income, and child not being in full time education.

High Suspicion

16.12   The most common reason for high suspicion relates to earnings — over 80 per cent of self-employed cases of high suspicion, and 60 per cent of high suspicions on employed cases.

16.13  The other main reason for suspicion is LTAHAW, which appears to be more prevalent on employed cases.

Raw Sample Results

The tables for this section are Tables 1A, 1B and 1C.

16.14  The FamC review looked at two client groups, self employed cases and employed cases. The pilot analysis is based on 298 cases, 100 of which were self employed cases and 198 employed cases. ( 2 sample cases were dropped; one as the customer had left the area and the other because the customer had been selected in two batches due to the case being reviewed between scans.) Many of the cases had more than one outcome resulting in 326 outcomes of which 110 related to self employed cases and 216 to employed cases.

16.15  The following tables show the sample results by category of outcome for both client groups together and then each client group separately

Notes :

The sample selected is insufficient to draw statistically valid conclusions; the results are therefore indicative and should not be quoted without appropriate caveats.

Where more than one outcome occurred on a case, the incorrect benefit was assigned to outcomes according to the following hierarchy until all the benefit in payment had been assigned:- Fraud, Customer Error, Department Error, Suspected Fraud.

The Total WBI and average WBI in Table 1A are raw sample results. They have not been weighted to take account of the different sizes of client group.

No past period frauds were identified from the pilot review. A past period fraud is one that existed on an earlier award but which has stopped and not been continued into the current award.

Benefit Changed — the award in payment has changed as a result of the error found.

Benefit Affected — the error has the potential to change benefit but in some cases the benefit in payment may not change e.g. max. award cases.

Table 1A : Raw Sample Results — All Cases

OutcomeNumber of cases reviewedTotal Weekly Benefit Incorrect (WBI) Average Incorrect Benefit
Correct

Of which:

No Change

DE where benefit not affected and no fraud or suspicion recorded

CE where benefit not affected and no fraud or suspicion recorded

150



141

3


6

  
Department Error

Benefit Affected

Of which:

Weekly Overpayment

Weekly Underpayment

Benefit Not Affected



17



12

5

*8



189.06



205.55

16.49

0.00



11.12



17.13

3.30

0.00

Customer Error

Benefit Affected

Of which:

Weekly Overpayment

Weekly Underpayment

Benefit Not Affected



20



13

7

**8



101.85



135.64

33.79

0.00



5.09



10.43

4.83

0.00

Suspected Fraud — Low

Benefit Changed

Benefit Not Changed

Suspected Fraud — High

Benefit Changed

Benefit Not Changed



5

36



8

42



81.73

2474.63



93.11

3164.88



16.35

68.74



11.64

75.35

Fraud 32 862.25 26.95
Total Number of Outcomes 326

* Includes 2 cases for which WBI is zero due to confirmed frauds on the cases which nilled the benefit, and 3 cases where a suspected fraud was also recorded. The remaining 3 cases are included under "Correct".

** Includes 1 case for which WBI is zero due to a confirmed fraud on the case which nilled the benefit, and 1 case on which a suspected fraud was recorded. The remaining 6 cases are included the "Correct" category.

Table 1B : Raw Sample Results — Self Employed

OutcomeNumber of cases reviewedTotal Weekly Benefit Incorrect (WBI) Average Incorrect Benefit
Correct

Of which:

No Change

DE where benefit not affected and no fraud or suspicion recorded

CE where benefit not affected and no fraud or suspicion recorded

40



37

0


3

  
Department Error

Benefit Affected

Of which:

Weekly Overpayment

Weekly Underpayment

Benefit Not Affected



6



4

2

*1



22.81



33.19

10.38

0.00



3.80



8.30

5.19

0.00

Customer Error

Benefit Affected

Of which:

Weekly Overpayment

Weekly Underpayment

Benefit Not Affected



6



4

2

**3



17.11



21.96

4.85

0.00



2.85



5.49

2.43

0.00

Suspected Fraud 1 & 2

Benefit Changed

Benefit Not Changed

Suspected Fraud 3 & 4

Benefit Changed

Benefit Not Changed



3

16



3

18



64.63

1238.39



45.75

1335.68



21.54

77.40



15.25

74.20

Fraud 14 364.55 26.04
Total Number of Outcomes 110

* A suspected fraud was also recorded on this case.

** These 3 cases are included the "Correct" category.

Table 1C : Raw Sample Results — Employed

OutcomeNumber of cases reviewedTotal Weekly Benefit Incorrect (WBI) Average Incorrect Benefit
Correct

Of which:

No Change

DE where benefit not affected and no fraud or suspicion recorded

CE where benefit not affected and no fraud or suspicion recorded

110



104

3


3

  
Department Error

Benefit Affected

Of which:

Weekly Overpayment

Weekly Underpayment

Benefit Not Affected



11



8

3

*7



166.25



172.36

6.11

0.00



15.11



21.55

2.04

0.00

Customer Error

Benefit Affected

Of which:

Weekly Overpayment

Weekly Underpayment

Benefit Not Affected



14



9

5

**5



84.74



113.68

28.94

0.00



6.05



12.63

5.79

0.00

Suspected Fraud 1 & 2

Benefit Changed

Benefit Not Changed

Suspected Fraud 3 & 4

Benefit Changed

Benefit Not Changed



2

20



5

24



17.10

1236.24



47.36

1829.20



8.55

61.81



9.47

76.22

Fraud 18 497.70 27.65
Total Number of Outcomes 216

* Includes 2 cases for which WBI is zero due to confirmed frauds on the cases which nilled the benefit, and 2 cases where a suspected fraud was also recorded. The remaining 3 cases are included under "Correct".

** Includes 1 case for which WBI is zero due to a confirmed fraud on the case which nilled the benefit, and 1 case on which a suspected fraud was also recorded. The remaining 3 cases are included the "Correct" category.

Results for National Caseload

The table for this section is Table 2.

16.16  This section shows the results in terms of the proportions of overall caseload currently receiving incorrect benefit as a result of fraud or error. The following tables show the number of sample cases in each category of incorrectness for each of the client groups. Confidence limits are not given as the sample selected is insufficient to draw statistically valid conclusions. The margins of error would therefore be so wide as to be meaningless; the results are therefore indicative, that is they are true of the selected sample, but may not be an accurate reflection of the whole FamC liveload and should not be quoted without appropriate caveats.

Main Features:

53 per cent of the cases have the correct level of benefit in payment. This includes 6 cases with customer errors and 3 cases with departmental errors which did not affect the level of benefit in payment and on which no fraud or suspicion was recorded.

The level of fraud is higher amongst self employed claimants compared with employed claimants (14 per cent compared with 9 per cent). The same picture emerges when we look at high suspicion (21 per cent compared with 15 per cent) However, the small sample does not allow us to test if these differences are statistically significant.

There is roughly half as much again high suspicion as fraud — this is broadly true of both client groups.

In about 10 per cent of cases (28 out of 298) there were customer errors, although nearly one third of these did not affect the level of benefit.

In 9 per cent of cases (25 out of 298) there were departmental errors, although nearly one third of these did not affect the level of benefit. The proportion of departmental errors amongst employed cases was higher (9 per cent) than the proportion amongst self employed cases (7 per cent).

Notes:

The proportions in the 'All Client Groups' table have been weighted to account for the relative sizes of the client groups.

Table 2: Estimated proportions subject to fraud and error

Outcome type Employed Self-Employed All cases
 No. cases per centNo. cases per centNo. casesWeighted per cent
Correct cases11055.64040.015053.4
"No change" cases10452.53737.014150.4
Departmental error
Benefit changed
not changed
all cases

11
7
18

5.6
3.5
9.1

6
1
7

6.0
1.0
7.0

17
8
25

5.6
3.2
8.8
Customer error
benefit changed
not changed
all cases

14
5
19

7.1
2.5
9.6

6
3
9

6.0
3.0
9.0

20
8
28

6.9
2.6
9.5
Low Suspicion
benefit changed
not changed
all cases

2
20
22

1.0
10.1
11.1

3
16
19

3.0
16.0
19.0

5
36
41

1.3
10.9
12.2
High Suspicion
benefit changed
not changed
all cases

5
24
29

2.5
12.1
14.6

3
18
21

3.0
18.0
21.0

8
42
50

2.6
12.9
15.5
Fraud189.11414.0329.8

Reasons for Incorrectness

16.17  The following tables analyse fraud and incorrectness by type of fraud/error within each client group. As some customers were committing more than one type of fraud/error, the numbers shown in the tables will sum to more than the total number of cases with that particular outcome shown in previous tables. For example, there were 32 cases with a fraud, but these 32 cases produced 35 reasons for fraud as some cases had more than one reason.

Notes for tables in this section:

The proportions in the 'All cases' columns have been weighted to account for the relative sizes of the client groups.

The sample selected is insufficient to draw statistically valid conclusions; the results are therefore indicative and should not be quoted without appropriate caveats.

Reasons For Fraud

Main Features:

The main sources of fraud are earnings related. Earnings frauds made up 70 per cent (14 out of 20) of the employed group fraud reasons and 60 per cent (9 out of 15) of the self employed group fraud reasons.

At first glance, the figures suggest that partner earnings frauds may be a greater problem than customer earnings frauds, for self-employed cases — 6 of the 9 occurrences of earnings-related fraud on self-employed cases related to partner earnings. To some extent this reflects the fact that amongst self employed couples three times as many partners work as customers (for employed couple cases there are approximately twice as many partners working as customers).

Table 3 : Fraud reasons

ReasonSelf-EmployedEmployedAll cases
 Count per centCountper centCountper cent
Capital17210310
Child leaves full time education171525
Earnings Customer3207351033
Earnings partner6 407351336
LTAHAW17210310
Other17  11
Other Income213 1536
Total151002010035 100

Reasons for High Suspicion

Main features:

As with fraud, the most common reason for high suspicion relates to earnings — over 80 per cent of self-employed cases of high suspicion, and 60 per cent of high suspicions on employed cases.

The suggestion — from the fraud reasons section — that partner earnings is more of a problem on the self-employed, seems more pronounced for high suspicion.

The other main reason for suspicion is LTAHAW, which appears to be more prevalent on employed cases. To some extent this reflects the higher proportion of lone parents in the employed group (around half the sample) compared to the self-employed (where there were only 13 lone parent cases in the sample).

Table 4: High suspicion reasons

ReasonSelf-EmployedEmployedAll cases
 Count per centCountper centCountper cent
Child not in household  1312
Childcare Costs  1312
Earnings Customer3139261224
Earnings partner177112342939
LTAHAW41710291427
Other Income  2625
Total241003510059100

Reasons for Customer Error

Main Features:

The majority of customer errors related to earnings — 76 per cent of all errors.

On employed cases, most earnings errors related to partner earnings, whereas for self-employed this is not the case. This contrasts with the situation on fraud and especially high suspicion (see above).

Table 5: Customer error reasons

ReasonSelf-EmployedEmployedAll cases
 Countper centCountper centCountper cent
Change of Address  1514
Earnings Customer327315617
Earnings Partner32710501347
Earnings Employer327210512
LTAHAW  21029
Other Income218210411
Total111002010031100

Reasons for Departmental Error

Main features:

All Departmental errors related to earnings.

All errors were either due to a mistake as to material fact or due to transpositional faults.

Table 6: Departmental errors — sample counts

ReasonSelf-EmployedEmployedAll cases
 Countper centCountper centCountper cent
Mistake as to material fact710014782181
Transpositional  422419
Total71001810025100


 
previous page contents next page

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

© Parliamentary copyright 1999
Prepared 4 February 1999