Since the Covid-19 pandemic, benefit overpayments have increased to new levels we haven’t seen before.
A new record was set in 2021-22 and the rate of fraud and error hasn’t dropped much since. The Department for Work and Pensions' estimates show that 3.7% of benefit spending was overpaid last year, compared with 2.4% in 2019-20, meaning it now pays out over £9.5bn in benefits that people aren’t entitled to. At the same time, DWP estimates that underpayments, where people receive less money than they’re eligible for, have increased to £4.2 billion.
Last year, when it published its 2022-23 accounts, DWP said it expected fraud and error to drop back to pre-pandemic levels within five years. Now, thanks to what DWP terms ‘"an increasing propensity to fraud in society’, it says it can’t commit to substantially reducing the level of fraud and error.
Indeed, DWP now forecasts that over the next five years levels of fraud and error in benefit payments are going to stay roughly where they are today.
So is DWP right to be pessimistic about getting benefit fraud and error down?
At the NAO, we’ve highlighted the high levels of fraud and error in benefit payments administered by DWP every year as part of our audit of its accounts since 1989.
We think DWP can do more to understand how fraud and error gets into the benefits system, and what can be done to tackle it. Here are three important things it needs to do:
Assess the cost-effectiveness of its controls against fraud and error
DWP accepted our recommendation that it should aim to demonstrate cost-effective controls against fraud and error. But what does this mean?
Cost-effective controls would mean DWP was doing everything it reasonably could to limit fraud and error, while still being able to deliver on its policies and support people. Nobody would be happy if DWP took too long to pay benefits to vulnerable people because it was looking for fraudsters, and nor would it make sense for DWP to implement new checks that cost way more than they can save. This means it doesn’t just need controls over fraud and error – it needs the right controls over fraud and error.
DWP already has a thorough understanding of what fraud and error occurs. But to get the balance right, it needs a better of understanding of why fraud and error occurs and how each case gets past the controls it currently has in place. That’s why DWP has started to talk about "root cause analysis" much more and has committed to a review of its existing controls.
Use data and technology to identify risky payments
Better use of data could help DWP stop benefit overpayments before they happen, or identify risky payments that it has already made.
Data has long been at the heart of DWP’s approach. For example, DWP uses HMRC data on people’s employment earnings to work out how much Universal Credit they should get. It’s also using machine learning to identify potentially fraudulent Universal Credit advance claims.
There is, however, more that DWP can do. For example, in our 2023-24 report on its accounts, we pointed out that DWP overpaid about £163m last year because, when it‘s working out how much to pay somebody, it doesn’t use the data it holds about other benefits it's already paying them. Using this, and other data, could help bring down overpayments significantly.
Make sure staffing levels are right to investigate risky payments
Once it knows where risky payments are, DWP needs trained staff to investigate. And I mean a lot of staff.
In 2019, we reported on the Carer’s Allowance and found that, between 2016 and 2018, DWP was investigating just under 12% of the cases ‘flagged’ in the system as possible overpayments each month, meaning lots of overpayments continued and some carers were later asked to pay back thousands of pounds. Carer’s Allowance overpayments have reached the news again this year, and we have committed to doing more work on the topic.
Current public spending plans assume that DWP can get almost 6,000 staff to review over 8 million Universal Credit cases for over and under payments by April 2028. But delays in getting the right staff in place last year started to delay the £6.6bn of savings it hopes this work will achieve. DWP has promised to get these savings back on track.
Conclusion
The recommendations we’re making to DWP haven’t fundamentally changed since before the pandemic, and I am pleased that its most recent plans have picked up on these. Working towards cost-effective controls will be iterative, data integration and sharing is often challenging, and new resources will need to be deployed in the right way.
But the prize is not only substantial savings, but better public trust in the benefit system.
Joshua Reddaway is director of fraud and propriety at the National Audit Office (NAO). Find out more about the NAO’s work on fraud and error.