Dodgy data risks breaking Universal Credit

As many as one in 20 Universal Credit payments to working Brits are wrong. Claimants are at risk of destitution when they’re underpaid and accused of fraud when they’re overpaid, as the Department for Work and Pensions has been using a flawed data stream provided by HMRC to calculate Universal Credit payments.

This week The Spectator revealed how HMRC’s PAYE earnings data is error strewn and fundamentally unreliable. Now it has emerged this system, used to calculate Universal Credit, risks criminalising benefit recipients and automated computer systems make it impossible for claimants to put the record straight. Insiders warn of a scandal waiting to happen – one that officials seem unaware of.

The social contract that underpins the Conservatives’ introduction of Universal Credit is that work always pays. ‘Universal Credit will make work pay for some of the poorest people in Britain’, said Iain Duncan Smith when our new benefits system was envisaged. Few now disagree with this fundamental approach and the principle that the state will top up your earnings every month is fiscally sound. But for UC to work as designed, the earnings data of those claimants who also have jobs – albeit casual and low paid – must be accurate. It relies on the quality of PAYE earnings that HMRC shares with DWP for UC households’ monthly benefit calculations. For the more than two million in-work UC claimants, the social justice element of UC has been subcontracted to the taxman. The DWP have to trust what HMRC supplies it with.

As revealed by The Spectator on Monday HMRC doesn’t care if PAYE earnings data is incorrect because it can always be restated after the event. HMRC runs an annual reconciliation process for tax, adjusting the taxpayer’s liability with either a welcome repayment of tax or a demand for more, to correct known errors. However, for UC PAYE (Real Time Information) earnings must be right first time, every time, for monthly benefit payments to be correct. 

So, what happens when they aren’t? Households receiving UC who are also in work are condemned to receive the wrong amount of benefit. Its design calculates and pays claims, automatically, on the same claim date each month based on earnings received from HMRC, on the basis that RTI earnings are accurate. There is no mechanism in UC to correct incorrect earnings before UC is calculated and paid. And HMRC has never implemented automated controls to check if RTI earnings are correct. 

Errors with the data mean the amount of UC is either underpaid or overpaid. Underpaid, because RTI earnings not actually paid to the claimant have been overstated by employers or HMRC in the claim period – so their UC top up is lower. Or, overpaid, because fewer earnings were included in the UC award calculation than were actually paid by employers to the claimant in the claim period.

Benefits advisers have been raising RTI issues for years. Online forum messages show advisers complaining of incorrect and late earnings reports and how this affects their ‘clients’. One speaks of incorrect and late payments leading to a benefit claimant being £210 worse off a month. Others speak of repeatedly having to ask for ‘manual reviews’ to correct monthly decisions calculated based on incorrect RTI data. But these are often responded to with the ‘computer says no’ approach: because DWP hold themselves not to be responsible for the employment earnings data they use to calculate claims.

The government officially acknowledges an actual RTI error rate of less than one per cent but numbers shared with The Spectator show the effects of delays in RTI reaching DWP, indicating the true figure is much higher, at about 5 per cent, with one in 20 working UC household claims wrong every month. The government disputes this figure.

Because of UC’s system of thresholds and allowances these errors never correct month on month. And UC claimants have no realistic chance of getting errors corrected. Frequently, they won’t be aware of the problem payment until the DWP has already calculated and paid the wrong amount of benefit. But, worse, the system can ignore claimants when they challenge their awards.

Most popular

Melanie McDonagh

When will the BBC stop adapting Julia Donaldson books?

Current UK Data Protection law enshrines claimants’ rights to demand a mandatory reconsideration where an automated decision that affects them is shown to be wrong. But jobcentre staff are guided to tell claimants their UC calculation has used the information reported by their employer and so DWP have therefore not made any decision they can reconsider or revise. So, the DWP disregards claimants’ complaints even when it knows the system is wrong. Claimants are sent from pillar to post with the jobcentre telling them to contact their employer and their employer sending them back to the jobcentre.

The problem is causing so much concern that a senior source within the benefits system warns: ‘Overreliance on the presumed accuracy of a fallible system not only risks penalising claimants, but also denigrating them, by disregarding their legitimate grievances.’ 

The result: struggling UC claimants have no certainty over what they’ll be paid month by month and, as the Concentrix scandal showed, claimants can lose their entire award each month. This can force them into unmanageable debt or having to rely on friends and family for emergency cash. The UN has even warned that UC underpayments mixed with dangerous debt risks leaving the already poor ‘destitute’.

The system even has the potential to automatically criminalise claimants for not declaring UC overpayments. This is most clear cut when data from the RTI system is ‘missing’, as HMRC styles it. HMRC has acknowledged the quality of RTI is sufficiently poor that it is unable to match RTI reported by employers correctly to any taxpayer employee in nearly 1 per cent of all RTI employment earnings. With the result that DWP’s system automatically overpays in these cases because, even though paid by their employer, these RTI earnings never reach UC so are never taken into account in the monthly UC award.

But there is a much larger scale of underpayments and overpayments in UC than those impacted by the 1 per cent of ‘missing’ RTI: those caused by ‘late’ RTI. This is not just RTI reported late by employers, as HMRC would have us think. It is the delays that arise within HMRC where it has to process RTI returns to identify which taxpayer household it relates to, all before it can make the earnings available to DWP. Despite the name, RTI is not a real time system.

These delays in earnings information reaching Universal Credit’s systems for award calculations also cause claims to be overpaid in one period but underpaid the next. Additional hours worked in lower paid jobs, typically associated with UC, are almost certainly most affected (almost 60 per cent of UC claimants are paid weekly or fortnightly).

The actual position was, until recently, rather different. The DWP did originally anticipate these issues and the need to validate ‘Real Time Earnings’ data, in 2013, when RTI was launched. They used a data feed from a payments company, Vocalink, to validate the RTI information they were using by cross referencing RTI earnings against real payments via the BACS system used to pay employees. It helped but was a sticking plaster.

Despite DWP taking steps to clean up HMRC’s dodgy RTI data, the problems remained, with benefits advisers drawing the comparison with the Horizon Post Office scandal. However, in February this year the Vocalink contract was allowed to lapse – a decision taken by HMRC. There are no plans for its replacement. The level of data error can now only climb higher.

The real scandal is not the discovery of an erroneous data system. It is the state’s refusal to acknowledge its use of flawed data. The DWP’s approach, of telling claimants they are not responsible for claimant earnings data used in UC calculations, is familiar. It’s identical to HMRC’s sidestepping responsibility for the 2016 Concentrix Scandal saying its supplier was at fault for not managing the customer service storm caused by HMRC’s own misconceived use of RTI.

The simple lesson is that the design and management of Universal Credit is only as good as the data the system relies on. DWP’s use of RTI earnings data breaks the most basic principles of data usage. Their refusal to address the UC data quality issue is condemning working claimants to a digital workhouse.

The United Nations Special Rapporteur said as much in a 2019 UK report to the UN General Assembly. He praised UC’s core principle but said: ‘Many aspects of the design and ongoing roll-out of the programme raise grave concerns about the adequacy of the country’s flagship benefits programme.’ The UN pointed specifically to RTI as a cause of underpayments with the Special Rapporteur finding that ‘claimants sometimes wait weeks to be paid the proper amount, even when they have written proof that the [RTI] information is wrong’.

It’s all the more troubling that in July DWP announced a further crackdown to ‘eliminate fraud from the system’, targeting savings of some £1.3 billion. Late missing and incorrect RTI earnings, ‘LMI’ as HMRC and DWP euphemistically call it, risks claimants with unlinked records getting caught up in the crackdown, through no fault of their own.

More problems wait down the line too: as part time and casual workers on Tax Credits are due to be moved across to UC by the end of next year, the total population most at risk from these errors will increase (there are more than one million people still to be moved from legacy benefits). The ever changing and casual nature of their employment means their employment records serve as poorer data sources.

Work has to pay, for the benefits system to work in the way the Conservatives want. It has to make timely and correct payments to claimants so that they can plan and budget their financial lives. But under the current system it does not. Worse still, Whitehall is no closer to even acknowledging the problem. If the system is not fixed, if inaction and ignorance continue, the most obvious Conservative policy legacy risks collapsing less than two decades after its birth.

A government spokesperson said:

We do not recognise this figure. In the vast majority of cases using Real Time Information (RTI) supplied by employers is an efficient and accurate method of calculating Universal Credit payments – and less than 1 per cent of cases do not match.

If a claimant wishes to dispute the earnings information we have used, they can submit evidence to us, and we will look into the case and make any necessary changes.