Guidance

Housing Benefit debt recoveries: background information and methodology note

Updated 7 September 2022

Applies to England, Scotland and Wales

1. Introduction

The Housing Benefit Recovery and Fraud (HBRF) data collection was introduced in 2008.

There were 2 parts to the collection:

• the first part collected information on the amount of Housing Benefit (HB) that local authorities (LAs) identified had been overpaid, and the amount of overpaid HB that they subsequently recovered
• the second part collected data on the number of fraud investigations carried out by LAs

From 2016, the Fraud and Error Service within DWP took on many of LAs’ fraud responsibilities. At this point, DWP stopped collecting data on the number of fraud investigations that LAs carried out and the collection was renamed “Housing Benefit Debt Recoveries” (HBDR).

The September 2016 HBRF publication (i.e. data covering the period April 2015 to March 2016) was the last to include data on the number of fraud investigations carried out by LAs.

For users with an interest in historic data on LAs’ fraud investigations, these continue to be available on the HBDR collection page and brief details of these statistics are provided at Annex A.

The rest of this document refers to the ongoing collection and publication of HBDR statistics.

2. Context of the statistics

2.1. What is Housing Benefit?

Housing Benefit (HB) is an income-related benefit which is provided to households in order to help them meet housing costs for rented accommodation. This is available to those who are either unemployed, on a low income or in receipt of other benefits.

The claimant may get help with all or part of their rent.

HB is administered by LAs on behalf of DWP. Claimants receive support in one of two ways:

• Rent Rebate – HB for council tenants, paid in the form of a reduction in their rent

• Rent Allowance – HB for housing association tenants or private renters in the form of a payment to the claimant or their landlord

Since 2018, most new working-age claimants have claimed the housing element of Universal Credit (UC) rather than HB, and all working-age HB claimants are scheduled to have transferred to UC by the end of 2024. The impact of the migration from HB to UC on the HBDR statistics is discussed in section 8.4.3.

2.2. Why might HB be overpaid?

A claimant might provide incorrect information on their HB claim – either:

• accidentally (“claimant error”)
• on purpose (“claimant fraud”)

Or they might fail to inform their LA of a change in circumstances – such as a change in the number of hours that they work – which leads to them no longer being paid the correct amount of benefit.

DWP publishes estimates of fraud and error in the benefit system. This publication provides fuller definitions of “fraud”, “claimant error” and “official error” and estimates the amount of HB overpaid for each of these reasons.

2.3. How do LAs identify overpaid HB?

LAs are encouraged and supported by DWP to identify and recover overpaid HB, for example, through the use of “real-time information” to compare claimants’ actual income, as reported by HMRC, with the information that they provided when they submitted their claim for HB. Where LAs suspect that overpayment is the result of fraud, a criminal investigation may be carried out.

2.4. How do LAs recover overpayment?

Where an HB claim is continuing, LAs may be able to recover HB overpaid to a claimant by deducting an amount from their monthly payments. There are limits to the amount by which payments may be reduced. As a result, it may take a long time for an LA to fully recover an overpayment.

In cases where the HB claim has stopped, LAs may be able to recover HB overpaid to a claimant either by invoicing them or, if they are in receipt of other benefits, by applying to DWP for a deduction from these benefits.

Some LAs may sell some of their HB debt to a debt collection agency and, in cases of suspected fraud, claimants may be taken to court to recover overpayment.

2.5. Why and when might an LA “write-off” overpayment?

When an LA believes that it has exhausted opportunities to recover overpaid HB, for example because the claimant is untraceable, has been declared bankrupt or died, it may decide to “write-off” the claimant’s debt. Some LAs write-off overpayment at the point that they identify that they are unable to recover it. Others process write-offs in batches. Further information on “write-off” of HB overpayment is provided in section 8.4.7.

3. Purpose of the HBDR statistics

These statistics allow DWP to monitor how effective local authorities are at recovering overpaid HB and allow LAs to benchmark themselves against other authorities.

4. Data source

LAs return data on spreadsheet templates. The returns are based on data taken from their HB systems.

4.1. Data fields and definitions

There are 4 fields on the HBDR return, as shown in the following table.

Field	Question
1	Total value of HB overpayments outstanding at the start of the quarter
2	Total value of HB overpayments identified during the quarter
3	Total value of HB overpayments recovered during the quarter
4	Total value of HB overpayments written off during the quarter

Each of these figures is broken down into “rent rebate” and “rent allowance”. LAs that are unable to provide a breakdown are allowed to just provide “total” figures.

LAs report the total value of overpayments identified, recovered and written-off during the quarter, regardless of when the original overpayment actually occurred.

In some circumstances, a review of an HB claim may identify that, although the claimant has been paid more than they were entitled to, they remain eligible for a lesser award (“underlying entitlement”). In this situation, LAs report the overpayment net of “underlying entitlement” and “underlying entitlement” is not recorded as either a “recovered” or “written-off” overpayment.

Some LAs sell some of their debt to a debt collection agency. In this situation, the sale price is recorded as “recovered” overpayment. The difference between the amount of debt sold to the debt collection agency and the sale price achieved is recorded as “written-off” overpayment.

4.2. Data reference area

Data are collected at an LA level and aggregated to a regional and national level. The HBDR data relate to the LAs and regions where the HB claims that have been overpaid are administered. Occasionally, claimants receiving these benefits may reside in a different local authority area.

4.3. Frequency of data collection

HBDR data are collected on a quarterly basis, according to a pre-agreed timetable.

Quarter	Reference period
Q1	1 April to 30 June
Q2	1 July to 30 September
Q3	1 October to 31 December
Q4	1 January to 31 March

LAs have a period of approximately 4 weeks from the end of the reference period to return their data to DWP. The data returned to DWP, however, refer to overpayments identified, recovered and written-off up to the end of the reference period, not the date on which the data are returned to DWP.

5. Quality management

5.1. Quality assurance

When LAs return their data to DWP, checks are carried out on their accuracy. “Logical” checks are carried out e.g. that the sum of fields 3 and 4 (the amount of overpayment either recovered or written-off during the quarter) is not greater than the sum of fields 1 and 2 (the total amount of overpayment either outstanding at the start of the quarter or identified during the quarter).

Comparisons are made with figures for previous quarters. Where there is an unusually large difference between figures for the current quarter and previous ones, this is queried with the LA. These checks focus, in particular, on field 1 (the total amount of overpayment outstanding at the start of the quarter). This is a larger and more “stable” measure than the other fields, which see larger quarter-to-quarter variation.

The total amount of outstanding overpayment at the start of a particular quarter is compared with the total outstanding overpayment at the end of the previous one. The total outstanding overpayment at the start of Q2, for instance, should, in theory, be the same as:

Total outstanding overpayment at the start of Q1 + overpayment identified in Q1 - overpayment recovered in Q1 -overpayment written-off in Q1

Although these figures are not always exactly the same – for reasons explained in section 8.4 – any differences should be small. Where a large difference is identified, this is queried with the LA.

5.2. Exclusions

Quality assurance checks are carried out on the data and the results are fed back to LAs, who are invited to review the data that they previously submitted and make any necessary changes.

DWP may withhold publication of any data which, following discussions with the LA, continue to appear incorrect. These instances will be clearly indicated in the final published tables.

6. Overview of statistical release strategy

6.1. Timeliness

Figures for Q1 and Q2 of the financial year are published in March. Figures for the whole of the financial year are published in September. This represents an approximately 5 months lag from the end of the reference period to the date at which the figures are published.

LAs are given approximately one month from the end of the reference date to return their data to DWP. There is a period of approximately two months, during which quality assurance checks are carried out on the data and any necessary revisions agreed with LAs. Once the data have been quality assured, there is a final period of approximately 6 to 8 weeks, during which the figures are aggregated to regional and national level and analysed, and final quality assurance checks are carried out prior to publication.

6.2. Data revision

Figures for the first two quarters of the financial year are published in March. In September, figures are published covering the whole of the financial year. Where LAs notify us of changes to their first two quarters’ data – or provide us with a late Q1 or Q2 return – these are incorporated into figures published in September.

Exceptionally, errors by DWP will be corrected via a revision in a subsequent publication or a special pre-announced release (depending on the severity of the error).

6.3. Rounding policy

Percentage figures presented in the summary document are rounded to the nearest 1%. In some cases, figures may not add up to 100% due to rounding.

In the supplementary tables, raw data provided by LAs have been added to generate regional and national totals, before being rounded to the nearest £1,000. As a result of this rounding, the sum of the LA figures shown in the tables may be slightly different to the regional and national totals. This is a different approach to that taken in previous publications, when LA figures were first rounded before being added to generate regional and national totals. The impact of this change on regional and national totals is negligible.

The sum of the amount of overpaid “Rent Rebate” and “Rent Allowance” in an LA may be different to the total amount of overpaid HB in that LA due to rounding.

7. Status of the statistics

The HBDR Statistics are designated “National Statistics”. This means that our statistics must meet the highest standards of:

• Trustworthiness
• Quality
• Public value

In order to establish and maintain trust in the statistics, we need to identify and, where possible, address any data quality issues. Where it is not possible to completely address data quality issues, we need to provide adequate information alongside the statistics to help users to understand their limitations.

To ensure the accuracy and completeness of data returns, and in line with UK Statistics Authority guidelines, we keep in regular contact with LAs. As part of this work, in July 2021, we asked LAs to complete a survey. The aim of the survey was to help us to: understand how LAs complete the HBDR return; explain observed trends in the data; address data quality issues; and solicit feedback from LAs on the statistics.

Some of the main findings of this survey have been incorporated into the next section.

8. Quality

8.1. Assessment of user needs and perceptions

LAs are regularly asked for information and feedback on the statistics. These comments have been used to improve the commentary on the statistics and add further details to the “background information” provided in support of the statistics.

LAs and other users are invited to provide feedback on the HBDR statistics, using the contact details provided at the end of this document.

8.2. Timeliness and punctuality

There are approximately 5 months between the end of the period that the HBDR data refer to and the data being published. This allows time for LAs to return their data; for DWP and LAs to quality assure the data; and for DWP to aggregate figures, analyse trends and carry out final quality assurance prior to publication.

8.3. Accuracy and reliability

In addition to the quality assurance checks that DWP carry out once LAs have returned their HBDR data, LAs carry out a range of checks on their data prior to sending their returns to DWP. DWP statisticians are working with LAs to make sure these checks are adequate and consistent in order to ensure the accuracy of the data.

8.4. Coherence and comparability

HB is a complicated benefit to administer and there have been a number of changes to the benefit system – such as the migration from HB to Universal Credit (UC) – which have had an impact on the amount of HB overpaid and recovered.

This section describes some of the issues that users of the HBDR statistics need to be aware of when comparing HBDR statistics.

8.4.1. Data completeness

Some LAs do not send data returns every quarter or cannot supply data for all fields. This produces missing data in the returns. The table at the end of the statistical summary shows the scale of missing data, each year since the HBDR collection began.

In the first year of the HBDR collection (FYE 2009) LAs that returned data accounted for less than 75% of all HB claimants. In view of the scale of missing returns in FYE 2009, data for FYE 2009 have been excluded from the charts in the statistical summary.

After the first year, response rates improved. In FYE 2010, responses were received covering approximately 96.5% of all HB claimants and from FYE 2014, over 99% of HB claimants were covered by the returns.

Up to Q2 of FYE 2017, when an LA did not send a return, its figures were estimated. These estimates were not published but were added to figures for LAs that had returned data, to produce estimates of HB overpayments in GB as a whole.

Different methods for estimating missing LA data were investigated. This analysis arrived at two conclusions. Firstly that, when it comes to HB overpayments, there is a large amount of variation between LAs, even between LAs of a similar size. However, it concluded that variation between quarters for the same LA was also large.

A decision was made to estimate overpaid HB in GB as a whole by uprating the total amount of overpaid HB reported by LAs that responded to the HBDR collection using HB caseload data.

The following formula was used.

Supplementary tables 1_5 to 1_8 show how the total amount of outstanding HB, and the amount of overpaid HB identified, recovered and written-off by LAs, have changed since FYE 2009. There are two columns of figures. The first column shows the sum of the data returned by LAs. The second shows “imputed” GB totals, based on estimates made for LAs whose data were missing.

During FYE 2017, the methodology for treating missing LA data was reviewed. It was decided that, in recent years, the amount of data missing from the HBDR return had reduced to the point that estimates were unnecessary. Therefore, in the statistical summary, while “imputed” GB totals are shown for earlier years, all of the figures from Q3 of FYE 2017 onwards are based on the data actually returned by LAs.

We continue to work with LAs to reduce the number of missing returns and will review our policy for treating missing data should the number of missing returns increase.

8.4.2. Comparing overpayment identified and overpayment recovered during a quarter or year

It is not possible to make direct comparisons between the amount of HB overpayments that LAs identify during a particular period and the amount of overpayment that they recover during that period. Firstly, because overpayments are not necessarily identified in the same period that they occurred; and secondly, because overpayments are not necessarily recovered in the same period that they are identified. Depending on the amount overpaid, it may take a long time for an LA to fully recover an overpayment.

8.4.3. Migration to UC

At the start of Q3 FYE 2015, there were just over 4.9 million HB claimants in GB. By the start of Q4 FYE 2022, this number had fallen by 46% to just over 2.6 million. Since Q3 of FYE 2019, most new working-age claimants have claimed the housing element of UC rather than HB and all working-age claimants are scheduled to have transferred from HB to the housing element of Universal Credit (UC) by the end of 2024.

This fall in the number of HB claimants has been a significant factor in the fall in both the amount of HB overpayment identified by LAs, and the amount of HB overpayment that LAs have recovered (as there are fewer new overpayments available to LAs to recover).

As well as the amount of overpaid HB that is available to recover, migration to UC has also had an impact on the methods available to LAs to recover overpaid HB. With fewer claimants receiving HB, LAs have fewer opportunities to deduct an amount from ongoing HB claims. Instead, they look to recover overpaid HB by either invoicing claimants when their HB claim ends or applying to DWP for deductions from UC or other benefits.

8.4.4. FERIS

The Fraud and Error Reduction Incentive Scheme (FERIS) was introduced in Q3 of FYE 2015 to help LAs to identify and prevent HB overpayment. Around the same time, real-time information (RTI) on claimant’s incomes became available to LAs. The income that claimants reported on their HB claims was checked against RTI on claimants’ actual income from HMRC. These checks led to undeclared income, and consequently overpaid HB, being identified by LAs.

8.4.5. COVID-19

At the start of the COVID-19 pandemic, many LAs redeployed staff away from debt recovery to frontline activities and restricted face-to-face meetings between staff and claimants.

The increase in the amount of overpaid HB identified and recovered by LAs in FYE 2022 compared with FYE 2021 should be seen in the context of a fall in the amount of debt recovery activity being undertaken during the COVID lockdowns in FYE 2021.

8.4.6. Retrospective changes in overpayments

LAs sometimes identify an amount by which a HB claimant has been overpaid, but then revise this amount at a later date. These revisions lead to a discrepancy between figures for one quarter and the next.

The total amount of HB overpayment at the start of a quarter should, in theory, equal the total amount of overpayment at the start of the previous quarter, plus the amount of overpayment newly identified during the previous quarter, minus the amount of overpayment recovered and written-off during the previous quarter. However, this is not the case in all LAs. Nationally, the total amount of outstanding HB overpayment that LAs reported at the start of Q4 of FYE 2022 was 0.37% less than the figures reported in Q3 suggested it would be.

This apparent discrepancy is due to the timing of the HBDR extracts. “Total debt outstanding at the start of the quarter” comes from a report run at the end of the quarter and changes can be made during the current quarter to the value of overpayments first identified in previous quarters and brought forward to the current one.

Overpayment identified in one quarter might retrospectively fall in a subsequent quarter. In some cases, for instance, “underlying entitlement” might be belatedly applied to an overpayment, reducing the amount that the claimant was originally thought to owe. In other cases, information might emerge that showed that, in accordance with HB regulations, an overpayment that an LA sought to recover was in fact “non-recoverable”.

Conversely, overpayment identified in one quarter might retrospectively increase in a later quarter. In some cases, a “write-off” might be reversed – this occurs when new information suggests that an overpayment that was previously written-off could, in fact, be recovered. Alternatively, an overpayment identified in one quarter could increase in a later quarter if recovery action led to court costs being added to the debt.

In over 80% of LAs, the difference between the total amount of overpayments outstanding at the end of quarter 3 of FYE2022, according to their quarter 3 returns; and the total amount of overpayments outstanding at the start of quarter 4 FYE2022, according to their quarter 4 returns, was less than 1%. We have therefore decided not to ask LAs to provide us with revised figures, when an overpayment identified in one quarter retrospectively changes in a later quarter, as to do so would place a disproportionate burden on LAs.

8.4.7. Write-off statistics

There is extremely large quarter-to-quarter variation in the amount of HB overpayment written-off by LAs. Although it is not the case in all LAs, across GB as a whole, there is a “spike” in overpayment write-off in Q4 followed by a sharp fall in Q1 the following year. On average, write-off in Q4 of a financial year is around 75% higher than in Q1.

Although some LAs write-off overpayments as soon as it becomes apparent that they won’t be unable to recover it, others write-off overpayments in batches – because, for instance, they need approval from their Cabinet to process write-offs over a certain amount.

Reasons for the “spike” in write-offs in Q4 include:

i. LAs make a “push” to “tidy up” accounts at the end of the financial year;

ii. LAs have made provision for “bad debt” in their annual budget; and

iii. LAs see an increased number of deaths in winter months.

In view of the high level of seasonality in “write-off”, the statistical summary includes a chart showing average write-off over the previous four quarters. This gives users of the statistics a clearer picture of the overall trend in “write-off”.

8.5. Accessibility and clarity

The statistical bulletin and ODS tables are released in accessible format and meet the guidelines set out within DWP’s accessibility statement.

8.6. Confidentiality

For confidentiality reasons, a small number of figures in the supplementary tables have been suppressed. In LAs where:

1. Fewer than 5 LA tenants were claiming HB at the start of the quarter; and

2. The amount of overpaid HB owed by LA tenants at the start of the quarter was not zero.

the breakdown between overpaid “Rent Allowance” (HB paid to LA tenants) and “Rent Rebate” (HB paid to housing association and private tenants) is not shown. This is to avoid the possibility of tenants who owe overpaid HB to their LA being identified.

In these LAs, only the total amount of overpaid HB is shown in the tables. In the latest statistics, between 5 and 6 LAs fall into this category each quarter.

9. Limitations of the statistics

Known issues

Not all LAs return HBDR data in all quarters. Up until Q2 of FYE 2017, estimates were made for missing LA data. Analysis of the HBDR data has found that there is a large degree of variability – both between different LAs and within the same LA in different quarters – in the amount of overpaid HB identified, recovered and written-off. Therefore, although the methodology that was used to estimate missing data was appropriate, there was a limit to how accurate these estimates could realistically be.

In view of the improved response from LAs to the HBDR collection, since FYE 2017, figures published in the statistical summary have been based on the HBDR returns actually received by LAs. Although the amount of missing data is small – in FYE 2022, data were missing from LAs that accounted for 0.8% of HB claimants in GB – these figures are presented on a different basis to those prior to Q3 of FYE 2017.

As already noted, there is a large amount of quarter-to-quarter variability, especially in the amount of HB overpayment identified, recovered and written-off by LAs. This makes it difficult to quality assure the figures as the large quarter-to-quarter differences routinely observed in these figures may occasionally mask errors.

The HBDR statistics show the amount of overpaid HB in an LA, as they understood it, at a particular point in time. If the LA made retrospective changes to an earlier overpayment – by, for instance, applying “underlying entitlement” to reduce the value of an overpayment in a later quarter – this would generally not be reflected in the statistics.

10. Data Tables

The supplementary tables published with the HBDR statistics show:

• Table 1_1: The total amount of outstanding HB overpayments at the start of the quarter

• Table 1_2 to 1_4: The total amount of HB overpayments (i) identified; (ii) recovered; and (iii) written-off during the quarter

These tables show a breakdown by LA. The following conventions are used:

• “.” – missing
• “-“ – zero or negligible
• “*” – included in total

Many LAs are unable to provide a breakdown between “rent reduction” and “rent allowance”. As a result, it is not possible to calculate regional and national totals for these measures.

There have recently been a number of local authority mergers. The way that figures are presented in the supplementary tables has changed since the last set of HBDR statistics were published in March 2022. The statistics shown are based on the current LA configuration, with figures for the “old” LAs that merged to form the new LAs no longer shown. This is consistent with HB caseload figures published on Stat-Xplore.

11. Other available DWP statistics

These statistics show the amount of HB overpayment that is identified and subsequently recovered by LAs. DWP carry out a sampling exercise to estimate overall levels of fraud and error in HB, including overpayment not identified by LAs.

According to the latest estimate, published in May 2022, there was approximately £860 million overpaid HB in FYE 2021. An estimated £540 million was overpaid due to fraud, £260 million due to claimant error and £70 million due to official error.

HB caseload data are available via Stat-Xplore.

A schedule of statistical releases over the next 12 months and a list of the most recent releases is available.

12. Contacts

Cross-Benefit and Migration Statistics
Client Statistics
Data as Statistics
DWP Digital

Email: cbm.stats@dwp.gov.uk

Annex A

Prior to 2016, data on fraud investigations carried out by LAs were collected and published alongside debt recovery statistics. The following fields were collected.

Fraud Investigations Fields

Field	Question
5	Number of full time equivalent fraud investigators at the end of the quarter
6	Number of cases referred to the LA fraud investigation section during the quarter
7a	Number of cases subject to investigation by the fraud section, that were closed during the quarter
7b	Total number of cases under investigation that related to DWP administered benefits (included in 7a) that were closed during the quarter
8	Number of cautions offered and accepted during the quarter
9a	Number of administrative penalties offered and accepted during the quarter
9b	Number of administrative penalties offered and accepted with a DWP benefit interest (included in 9a) during the quarter
10a	Number of cases accepted for prosecution during the quarter
10b	Number of cases accepted for prosecution with a DWP benefit interest (included in 10a) during the quarter
11a	Number of prosecutions resulting in guilty outcomes (includes guilty pleas and verdicts) during the quarter
11b	Number of prosecutions resulting in guilty outcomes (includes guilty pleas and verdicts) with a DWP benefits interest (included in 11a) during the quarter

Note: DWP benefits interest includes

• Income Support
• Jobseekers Allowance
• Pension Credit
• Incapacity Benefit
• Employment Support Allowance