The British government (HMG) lost up to £58.8 billion to fraud and error in 2021 alone, excluding Covid-scheme fraud, according to a new National Audit Office (NAO) report – which suggests that government departments have failed to act as ‘one government’ and performed poorly at sharing data across departments.
HMG is only in the early stages of using data matching, intelligence and analytics to prevent fraud and corruption, NAO said. And due to slow and burdensome processes to share data between departments, often made more difficult because of dated legacy systems, there also is limited buy-in from departmental officers.
Some £10 billion of this lost government income “was due specifically to tax evasion or criminal attack” NAO said in a March 30 report – as Parliament’s Public Accounts Committee (PAC) also last month warned that “The Home Office is not doing enough to influence those who are instrumental in combatting fraud. The department is dependent on the banking, technology, telecoms and retail sectors to fight fraud, but its approach will continue to be sluggish and outmanoeuvred if it relies on purely voluntary charters with these sectors…”
Tackling fraud with data analytics
This failure to coordinate on data is an expensive one, says Peter Taylor, counter fraud specialist and cyber crime consultant, who points to proposals made to use more data analytics to spot fraud made by professional services organisations like KPMG and Gartner as early as 2016: “We have many good and qualified counter fraud teams and specialists looking for work. We should be tapping into them now. We are already three years late to the party and these people are available and ready to go now” says Taylor, adding that nearly £7 billion in COVID scheme related fraud could have been avoided if better checks and balances had been in place.
Quantexa, a software intelligence system onboarded by UK government in August 2022 as a supplier for its Big Data and Analytics procurement framework agreed that some basic toolings are just not in place.
“The British government’s bounce back scheme is a recent example of how a policy with the greatest of intentions can be cynically exploited by fraudsters. According to government figures, at least £640 million (out of £4.1 billion) worth of facilities have been retrospectively marked as ‘suspected fraud’. Lenders tasked with distributing these loans did not have the capability to assess the individuals, their counterparties and connections in order to prevent fraud,” said Ross Aubrey, Head of Fraud Solutions, EMEA at the software firm.
Aubrey suggests however that government policies such as the Economic Crime and Corporate Transparency Bill could change data sharing between the public and private sector significantly, and improve the UK’s use of data analytics. The Bill, currently in the committee stage at the House of Lords, aims to provide Companies House with the mandate to verify firms and individuals when they register within the UK as businesses.
This centralised system would join multiple data sets across government, and make it easier for both public and private sector companies to access intelligence about the firms they are dealing with, making spotting fraud and criminal intent easier. Along with such policy measures, Aubrey told The Stack that technology like machine learning could help government departments spot and prevent fraud within large data sets.
“Criminals' attempts to defraud the taxpayer range from the sophisticated to relatively simple detection methods – such as applying for different loans with slightly different addresses on each form. But even simple deception is very hard to spot by human agents, especially when we consider that nearly 1.56 million businesses were approved for finance with the Bounce Back Loan Scheme,” he points out.
Decision Intelligence (and a culture change) needed
James Bore, a cybersecurity consultant, highlights how easy it is for “anyone with a little intelligence and lack of ethics to abuse the system” with disjointed data sets making detecting fraud difficult for human agents.
He pointed to a BBC story about a flat owner in Cardiff who received tax bills for 11,000 Chinese companies after they fraudulently used his home address to register for VAT. ("You'd think there'd be a system with the technology today that would have picked it up immediately," flat owner Dylan Davies was quoted as saying.)
The use of Decision Intelligence (DI) solutions, an AI type as recommended by Aubrey, is a step to solving this issue of extremely large but fragmented data-sets. DI technology helps organisations transform their data - which may contain mistakes, inconsistencies or be deliberately misleading - into accurate and complete data.
The UK Cabinet Office has agreed to leverage this technology as of February 2023 and it could potentially prevent further loss of taxpayer money and exploitation of government department vulnerabilities.
This is part of the government’s mandate of bolstering the new Public Sector Fraud Authority (PSFA), and equipping it to both prevent and recover funds lost to anti-government fraud. However, there are not just cultural and policy challenges at play but hardware and software hurdles that government departments and the PSFA need to clear before being able to seamlessly share data and intelligence in real time, Bore warns.
“Joining up data and verifying information would make a huge difference, but with a large range of legacy siloed systems never designed to be interconnected there are significant challenges. It would only be possible with a significant effort from the government, of a level not achieved to date,” he said.
The challenges also extend to the technology experts and specialists needed to steer this data transition for the government, Bore added, saying: “Even with a joined up data picture we’d then need to make sure investigation and enforcement capabilities exist, and they simply don’t. Nor will they for some time as many of the specialists working in this area have moved to the private sector due to a lack of interest from the public sector – this leaves the public sector more vulnerable to potential fraud than it should be.”
UK government fraud: HMG says PSFA will "step up"
A Cabinet Office spokesperson said: “Since 2021, we have invested more than £900 million in taking action on fraud and, as the report acknowledges, we have made progress by establishing the Public Sector Fraud Authority which is stepping up the Government’s efforts to protect taxpayers' money. The government has recovered more than £3.1bn of fraud losses in the last two years, including within Covid-19 schemes, but we know there is more we can do. That is why we are expanding the Government’s Counter Fraud Profession, developing new technologies and boosting skills and training to further protect the public purse.”
The Public Accounts Committee remains unimpressed: “The Department [Home Office] does not yet have the data needed to properly understand the threat. Its most recent estimate of the cost of fraud to individuals is based on 2015–16 data, and it has no reliable estimate of the cost of fraud to businesses” it says. NAO meanwhile believes breaking down data silos between departments will be critical to tackling fraud.
“The PSFA [which was launched in August 2022] can help to better coordinate government efforts and to provide services that fill some of the capacity gaps across government” NAO said in its audit.
“But many central services and functions fail because they do not achieve buy-in from departments. PSFA will need to work in partnership with departments such as HMRC and DWP, which contain most of government’s counter‑fraud capability, and win the support of departments with less capability” it concluded.
