Department of Work and Pensions LogoNational Statistics Logo Fraud and Error in the Benefit System

Latest data from DWP for Great Britain in 2020-21

Overview
This interactive visualisation provides estimates of fraud and error in social security benefits funded by the Department for Work and Pensions (DWP). Data for the visualisation is drawn from the latest published Data Tables found on the Fraud and Error statistics home page.

The data provided includes the final 2020/21 financial year estimates, covering the period from April 2020 to March 2021. These statistics have been developed in accordance with the National Statistics Code of Practice and supporting Principles.

The "Overpayments" tab contains estimates of benefit overpayments (when a claimant is paid more in benefit than they are entitled to). The "Underpayments" tab contains estimates of benefit underpayments (when a claimant is not paid enough).

The charts and summaries on the "Overpayment" and "Underpayment" tabs are responsive to the selections you make, so offer a new way to explore DWP's fraud and error data. The full report and supplementary documents are still available as in previous years and can be found on the Fraud and Error statistics home page.
How we estimate fraud and error
We measure fraud and error all the time for some benefits, randomly sampling people and checking that the information we hold is up-to-date. For benefits we don't measure all the time we use either old estimates or a proxy (based on a similar benefit or an average rate). This allows us to estimate the total rate of overpayments or underpayments for all benefits funded by DWP.

You can find more information about which benefits we measure and how we measure them in the Background and Methodology document, which can be found on the latest published Data Tables found on the Fraud and Error statistics home page.
Impact of Coronavirus on the 2020/21 statistics
Due to Coronavirus, we only measured Universal Credit for the 2020/21. For all other benefits the rates from the 2019/20 publication were rolled forward and the 2020/21 expenditure applied to them.

You can find more information about this in the annexes of the publication, which can be found on the Fraud and Error statistics home page.
Methodology changes in 2020/21
We made a change to how we assign Fraud and Error to the underlying reason and revised the Universal Credit figures from 2019/20 accordingly. This does not have any impact on the total Fraud and Error rates but does mean that comparisons on Universal Credit to pre 2019/20 should not be done, therefore they have been removed from the visualisation.

You can find more information about the changes in the Background and Methodology document, which can be found on the Fraud and Error statistics home page.
Methodology changes in 2019/20
For the 2019/20 publication, we made some methodology changes to the way we categorise the errors. This means that the estimates from before this change is not comparable to this year's data. We therefore reran the 2018/19 data with the new methodology, and so there is only 3 years of data available. This is to ensure that no misleading comparisons can be made.

You can find more information about the changes in the Background and Methodology document, which can be found on the latest published Data Tables found on the Fraud and Error statistics home page.
Rounding our estimates
The data for this visualisation is drawn from our published Data Tables. In the tables, monetary values of totals and error types (fraud, claimant error and official error) are rounded to the nearest £10m; monetary values of error reasons are rounded to the nearest £1m.

Universal Credit is the exception to this rule; because the monetary values are small, we have rounded all values to the nearest £1m.

In the visualisation further rounding of these figures is sometimes applied to give a high level view of our statistics. Rounded values for individual error types or reasons may therefore not always add up to the rounded total provided. To see all of our estimates in the most precise format available, please refer to the latest published Data Tables found on the Fraud and Error statistics home page.
Uncertainty and confidence intervals
In this visualisation we only provide point estimates of fraud and error. However, all sample-based estimates are subject to a degree of uncertainty because we do not check every single benefit claim. In the full report "Fraud and Error in the Benefit System" and its associated tables, we additionally provide confidence intervals for all our estimates to give an indication of the level of uncertainty.

The full report and supplementary documents can be found on the Fraud and Error statistics home page.