News feature: As the pilot begins for a city-wide programme in data analytics, it promises to extend its capabilities from detection to prevention
One of the more ambitious data sharing operations in local government formally gets underway today, with five of the capital’s boroughs taking part in the pilot for the London Counter Fraud Hub.
Ealing, Camden, Islington, Hackney and Croydon are taking part in the process to identify potential fraud around council tax, business rates and housing rents, contributing data from their own sources to the hub along with that from third parties.
It is a significant step in a process that goes back more than two years, when Ealing received a £430,000 grant from the Department of Communities and Local Government to lead the work, following which all 32 of the capital’s boroughs and the City of London Corporation signed a memorandum of understanding (MoU) to take part.
The management contract followed almost a year ago, with the Chartered Institute of Public Finance and Accountancy (CIPFA) signing on for nine years and BAE Systems engaged to provide the analytics and case management capabilities, supported by organisations including fraud prevention group Cifas, Companies House and Ordnance Survey.
Now the five boroughs are beginning a nine-month pilot with the aim of drawing lessons from the data analytics and laying the ground for every council in the capital to join within the next 18 months.
It provides an important milestone for Ian O’Donnell, Ealing’s executive director of corporate resource and chair of the board overseeing the project. He has been a prime figure from the beginning and says the MoU between the boroughs has been a big factor in making it possible.
“It made it possible for us to go the market and negotiate with suppliers, and it was attractive to them to know that all 33 councils were already on board,” he says.
This also provided the background to the BAE working with CIPFA on the algorithms to identify relationships in the available data and provide a series of fraud risk scores. The software builds links between the entities – notably people and properties – identified from the data, then constructs networks of the connections between those entities which can be enhanced to highlight risk factors.
“The key point is that it’s not just data matching, it’s about understanding those relationships in a much more sophisticated way, then attaching a score that allows us to assess risk,” O’Donnell says. “We’re still looking at how that score will be presented and what it will look like in terms of the outputs from the system.”
He says there will be flexibility around the outputs that will enable BAE to work with the councils on how they are presented and used. As an example, in the case of evidence that someone has been wrongly claiming single person discount on council tax, the output can be fed into the borough’s system to generate a letter telling the person another resident has been identified, that their discount has been suspended and they have 28 days to appeal.
The possibilities will be tested during the pilot, with a plan to add more data after six months then assess after nine how to take it all forward.
“I’m not expecting to detect any fraud in the next six months,” O’Donnell says. “We’re going to be refining the system and understanding the consequences of doing this analysis. Then we can begin industrialising it, which is when I think we will start finding fraud, using live data.”
He points out that theoretically the project could be dropped at this stage, but is optimistic about the prospects of expansion, and says there is already a plan to bring in the rest of London’s boroughs in batches of 10 at three-month intervals.
“After that we should be able to begin looking at the other types of fraud and expand the operations by adding more analytics and datasets,” he says. “I think we’ll start finding then that we’re getting a lot of cross-boundary fraud and will be learning a lot as it rolls out.”
There are also possibilities around the contractor providing investigative capabilities in the latter stages of the pilot. O’Donnell expects an early spike in demand for this as councils identify possible fraud and may not have the capacity to follow up themselves, and says the commercial model is underpinning the prospects of further development.
“We’ve put in place the commercial model that means the contractor only gets paid if there is a recovery from us detecting fraud. This means they are incentivised to go on developing the hub and innovating, so we detect more fraud and they get paid more.
“Also, we think that over time there will be a shift from detection to prevention. As the hub gets better at detecting fraud we can move to trying to prevent fraudsters getting into the system in the first place.
“That means another aspect of the hub comes into play in the form of the enquiries facility. This draws on the analytics and we will design a piece intended to score people at the point where they apply for a council service. “
This would require the algorithms to be designed for specific enquiries – such as whether an applicant for a council house is legitimate – but O’Donnell says this should not be a particularly daunting task. It would need an analysis of which pieces of information are needed and how they could be validated, and that would provide the basis of the enquiry’s design.
“It will establish identity and entitlement using a range of datasets and by understanding the relationships in a more sophisticated way. We think that will prevent people from making false applications for services and benefits.
“As the hub shifts into that prevention mode the contractor will be paid per enquiry rather than by recovery. It’s something we will build into our front end customer services so it’s triggered when somebody applies online or when speaking with a customer services representative, and will be validated in real time at point of first contact.”
While the test will come over the next nine months, O’Donnell says there has been interest from other parts of the country in setting up similar hubs, and that CIPFA has been talking with other authorities.
“We might end up with a number of regional hubs, or alternatively this hub may grow and other people joint it.”
There have also been discussions with the Greater London Authority about a possible involvement. O’Donnell says it could have applications for purposes such as ensuring people are not using family members’ free travel passes after they have died.
“It’s the first time London has attempted a big data sharing exercise that has not been either driven by central government, like the National Fraud Initiative, or by one of the organisations that sits over the top of councils,” he adds.
“Getting all 33 to sign up to it has so far been so good. It may well end up being a model for other areas.”
Picture by Mai-Linh Doan, CC Attribution Share-Alike 2.0 France through Wikimedia