The Labour Government has promised the “biggest fraud crackdown in a generation” in a bid to stop those it describes as “benefit cheats”.
Work and Pensions Secretary Liz Kendall has said some people are “taking the mickey” by claiming benefits.
A bill introduced last month will give the Department for Work and Pensions (DWP) sweeping new powers – including suspending driving licences of benefit claimants who fail to pay back debts over £1,000.
The legislation will also let the Government recover money directly from people’s bank accounts if they are involved in fraudulent welfare claims.
However, the DWP faces a major challenge in reversing the huge increase in fraud and error seen in recent years.
Benefit overpayments rose to £10.2bn in 2023-24, according to the National Audit Office (NAO)’s recent report for MPs.
It is almost double the £5.6bn lost to overpayments in 2019-2020, the year before Covid hit.
Joshua Reddaway, the NAO’s director of fraud and propriety, told The i Paper: “The vast majority of benefits are paid correctly to the right people, but there are billions of pounds that are not,” said the public spending watchdog’s fraud chief.
“There is a lot of money being lost here. But that also makes it a massive opportunity for government and the taxpayer to save billions,” he added. “Trust in the system is absolutely vital.”
Reddaway wants the DWP to show it is improving its systems for flagging benefit overpayments so they can be investigated and quickly stopped. This, he said, would help prevent so much fraud happening in the first place.
“It’s about learning why overpayments are happening, and putting in place better preventative controls to stop it,” said the public spending watchdog’s fraud chief.
“This could be checking claims against information government already holds, or asking for more information.”
He is pleased the DWP has committed to a review of its controls – but thinks the department still needs a better of understanding of why so much fraud is occurring.
“It’s not good enough to say there is this propensity across society towards increased fraud. You need to be able to say: ‘What is driving that?’ They [the DWP] haven’t really answered that question.”
“The other question they’re not very good at answering is: ‘What is it about the way they administer benefits that means fraud and error is getting through their controls?’”
DWP benefit figures show £800m was lost to officials’ own errors in 2023-24, another £1.6bn to claimants’ mistakes, while £7.4bn was lost to deliberate fraud.
The NAO fraud director thinks the growth of Artificial Intelligence (AI) tools will be “helpful” in spotting possibly fraudulent benefit claims.
The DWP is already using some AI-based “machine learning” with claims to “work out the patterns of behaviour associated with fraud so you can flag stuff that is suspicious”, said Reddaway.
“I think that the DWP needs to get significantly better at this. I think DWP knows that, which is why it is investing a lot in this area.”
However, the Government will have to be both “clever and transparent” in its use of AI, the fraud chief added. He warned that relying on AI tools comes with major risks.
“The first is that you actually stop payments that are legitimate. The second is that you end up asking for more information than you need, and certain groups feel targeted for more information in a way that’s unfair.”
Reddaway wants the DWP to produce statistics on which particular groups of benefit claimants are seeing payments delayed through investigation.
The DWP must get better at cross-checking information held on the multiple benefits claimed by some, said the NAO fraud director.
“There’s an issue around legacy computer systems, and systems around the department that don’t talk to one another,” said Reddaway.
“The department has not taken sufficient action to make sure the computer systems for each different type of benefit share information.”
He said it was “slightly worrying” that there were still legacy IT problems “because we need systems between government departments to talk to one another”.
“If DWP is still struggling within its organisation, you can imagine the challenge of getting it to happen more widely [across Whitehall],” he added.
Reddaway said the DWP does not use AI or other data analytic tools to make automated decisions on who qualifies for benefits – nor should it.
“I don’t think there’s any way at the moment to get around benefit administration requiring large numbers of people [staff],” he said.
Overpayment of carer’s allowance in recent years showed there has been a shortage of DWP staff investigating problems, said the fraud chief.
Some families were taken to court and forced to repay thousands after overpayments were allowed to build up over several years.
“With carer’s allowance, the system was quite good at flagging when people were people overpaid. The problem was they didn’t have enough staff to actually look at the cases flagged,” said Reddaway.
He said a NAO report from December showed that staffing to probe carer’s allowance problems had “not improved in the way that we had hoped”, adding: “It is fundamentally down to flagged concerns not being investigated.”
Reddaway is pleased the Government has committed to getting benefit overpayments back down to pre-pandemic levels by 2030. “But we thought it was too high pre-Covid. We would like to see it fall to pre-pandemic levels, and then drop further.”
A DWP spokesperson said: “We already use machine learning to identify and prevent potential fraud but do so within a robust governance and ethical framework and with constant human oversight.”
“New measures in our Fraud, Error, and Recovery Bill will ensure taxpayers have confidence that welfare spending goes to support those who really need it, not people who seek to exploit it, saving the taxpayer £1.5bn over the next five years – part of wider plans that will save £8.6bn by 2030.”
Comments
Leave a Comment