CSW recently brought a group of civil servants together to discuss the use of citizen data in reducing Fraud, Error and Debt (FED). Tim Gibson reports on the debate.
“Every human creature,” claimed Charles Dickens, “is constituted to be that profound secret and mystery to every other.”
Writing two centuries ago, even someone of Dickens’s imaginative powers couldn’t possibly have envisaged the kind of world in which we now live. It’s a world in which we’re constantly divulging information about ourselves – to banks, retailers, websites and, indeed, the government.
This data gives anyone who chooses to study it an insight into our lives. And in many cases, it’s information we happily share in order to make life easier in our technological age. To those with access to the right databases, many people look less like a “profound secret” these days, and more like a trail of breadcrumbs just waiting to be followed.
While many of us may shiver at the thought that our lives can so straightforwardly be scrutinised by strangers, there are benefits. Consider the ease with which it is possible to apply for a credit card, or secure finance on a car. Think of how simple it is to set up a bank account and start paying bills from it. Remind yourself how handy it is to ‘click-and-pay’ when shopping online.
All of which begs a question: what if the government could use the data it already holds about UK citizens to simplify the delivery of services? Even better, what if it analysed that data to establish entitlement to things like benefits and tax credits? That way, it could limit the number of inaccurate payments, thereby reducing unnecessary expenditure and helping shrink the budget deficit.
The FED agenda
Like all good ideas, someone has already thought of this. And it underpins the government’s Fraud, Error and Debt (FED) reduction strategy.
FED is a major cost for the government. Despite the efforts of an FED taskforce – started in 2010 by the Cabinet Office – fraud and error alone still cost public services an estimated £31 billion per year. As Francis Maude notes in a recent report by the taskforce, that’s an annual cost of £500 to every single UK citizen, and it can be avoided. In an age of austerity, he says, “it is certainly not acceptable.”
FED occurs for a variety of reasons, but it is often the result of government organisations having inaccurate information about citizens. For example, if HMRC has an incorrect estimation of earnings from a taxpayer, it may not bill them for the right amount. Similarly, DWP may overpay benefits because it does not have a clear picture of a claimant’s entitlement.
Some of these problems are beginning to be surmounted through effective use of citizen data by government departments. Indeed, a recent CSW roundtable on the topic – sponsored by data effectiveness specialist Transactis – revealed that many public servants are ahead of the curve in this area, and already use information about citizens in FED reduction.
Getting better information
At the heart of the use of citizen data to reduce FED is the gathering of more accurate, up-to-date, information about the population.
For example, RTI (Real Time Information) – under which companies report PAYE data to HMRC every month, rather than annually – is already proving useful in efforts to reduce FED. As HMRC’s Andy Farrar said, “RTI will allow us internally, for the first time, to have a glimpse of our [tax credit] claimants’ income in real time, rather than having to wait a year, during which we could have had erroneous estimates, or overpayments. So there’s something in it from the customer perspective – getting the right money at the right time – and something in it for us, in terms of reducing error and fraud.”
RTI will also help ensure accuracy in Universal Credit payments, reported Huw Jones from the Department for Work and Pensions. “We’re utilising earnings information that is fed into HMRC through RTI,” he explained “which will then be fed across to DWP for Universal Credit claimants who are employed, to save them having to report their earnings directly to government. One of the big reasons for that is to cut down the amount of error we got from individuals reporting their own earnings.”
A problem shared
As the Universal Credit example shows, sharing citizen data across departments is key to FED reduction. Mark Cheeseman, from the Legal Services Commission, stated that in his organisation internal analysis of data often goes hand-in-hand with the question of how that data might benefit other public sector bodies. “It’s interesting to see DWP and HMRC working together,” he said. “I think the interesting next step is how more government departments can get involved with that.”
There seems to be a clear desire for greater data sharing across government. For example, Lyn Harding from HM Courts and Tribunals Service (HMCTS) said he would like to see more information passing between his organisation and bodies such as the Legal Services Commission. Similarly, Sean Rigney from the Ministry of Justice stated that his department is looking into ways of sharing information with DWP and HMRC about the setting of fines for offenders, and fine enforcements.
When it comes to joined-up service delivery, in recent years local government has often led the way. By way of example, Charlotte Piper from the Department for Communities and Local Government (DCLG) pointed to Kent County Council. It has recently established a buying framework, open to all public sector organisations, for the purchase of credit reference checks. She suggested that a similar framework may be useful for central government – something that Paul Daley, from Transactis, reported is provided to some extent by the recent Data, Access, Processing & Analysis (DAPA) framework.
Using citizen data proactively
Steve Dalby, from DWP, stated that rules within his department prevent officials from using credit reference agencies unless they receive a specific allegation against an individual – so its approach is inherently reactive. Some speakers suggested that government should find ways to be more proactive in its use of such information, so as to prevent losses from happening in the first place.
This sentiment was given clear expression by Mimo Ahmed from the Medicines and Healthcare products Regulatory Agency (MHRA). The agency is only allowed to run background checks on people when they are accused of wrongdoing, and can’t require them from new applicants for licenses to sell pharmaceuticals. As Ahmed stated, the fallout from a convicted criminal selling counterfeit drugs under license would be considerable, yet the risk is intrinsic to the current rules.
In response, Mark Babington, fraud lead at the National Audit Office (NAO) and member of the FED taskforce, exhorted colleagues not to feel hamstrung by legislation. He said the challenge was to establish what existing powers enable departments to do, and to make full use of all the room for manoeuvre that government has. When that happens, he argued, data can be used preventatively – as is the case with DWP and HMRC in Universal Credit – rather than to identify FED and deal with it retrospectively.
The need for structure
There was broad consensus on the need for a structure that enables government departments to access, and share, data.
“Perhaps we need to look a bit more radically,” opined the DWP’s Huw Jones, “[and think of this as] government data. [Then] you can have it in one place, and people can pull off what they need for their departmental bit, rather than passing this data around from one box to another.”
The danger of this, a number of participants warned, is that it could look as if the government is compiling a “mega-database”. The public is likely to be highly resistant to such an idea, as the Blair administration discovered when it mooted ID cards in the 2000s – and, as Ben Durcan from DWP warned, such a system could easily be abused.
However, John Sharman from Transactis asserted that providing access to data across government does not require it to be stored centrally. Rather, a technical system can be put in place that links various sources of data so that officials can gain access to them all, and analyse their chosen data accordingly.
Indeed, said his colleague Paul Daley, most of the potential gains can be realised by making a limited set of arrangements to share key datasets: “When specific organisations have a specific requirement to share a specific set of data for fraud and error reduction, I think it works,” he said. “But I personally think you can never say: ‘Government, share data for debt: go!’ You’ve got to be very, very specific.” He cited one project for HMRC and DWP, which used data from sources including the Land Registry and credit reference agencies to achieve limited but highly productive goals.
Of course, as Holly Adams from HMRC was quick to point out, this assumes the government has skilled people in place to conduct the data analysis. She identified a shortage of analysts in government who can identify “good” data, and spot where anomalies point to potential abuses by fraudsters. “One of the modes of operation for fraudsters is to give poor information,” confirmed Sharman. “So what looks like incorrect data is an indicator of bad behaviour. We need to understand what good data looks like, and what bad data looks like, and the processes that someone has to go through to generate that data record. If you understand those three things, you get a read on what’s really going on.”
Adams’ colleague, James Armitage, explained that having identified anomalies in the data, personnel also need to be able to communicate the significance of analysed data to their managers, in order that it can be used to inform decisions.
An example of the power of good analysis, explained the DWP’s John Viggers, is his department’s use of predictive modelling, based on past patterns of overpayment, to reduce fraud. The value of analytics explains its inclusion on the DAPA Framework, he said; it may be necessary for departments to outsource this work if they lack relevant expertise in-house.
Philip Nye, from BIS, echoed Viggers’ remarks, saying that the skills shortage is likely to be especially acute in smaller organisations. Government departments need linking, he said, in order to leverage the expertise in the system for both data handling and analysis.
A joined-up future?
The effective use of citizen data to reduce FED, the group concluded, depends on civil servants’ ability to share data across organisational boundaries – something that can help departments track down debtors, support cross-checking work to squeeze out government errors, and close the gaps between services through which fraudsters can slip. Given that substantial reform of the relevant legislation is not currently on the cards, officials will have to find practicable ways of improving the exchange of data between departments. While departmental lawyers tend to take a cautious line on such exchanges, citing data protection laws, the Land Registry’s Karen Wilbourn argued that stimulating better data exchange is “complex, but not insurmountable.”
While the government has created a FED taskforce and developed an over-arching strategy, there was a clear appetite amongst the civil servants at our round table for stronger action by the centre of government to facilitate and support better data exchange. Improving the sharing and use of public sector data may involve altering some of government’s processes and procedures; it’s likely to involve testing the boundaries of data protection laws; and it will certainly require some culture change. When the alternative is continuing to pay massive bills for fraud, error and debt costs, however, such work looks like a very sensible investment.