Fighting fraud through data integration

A June 2013 report issued by the National Fraud Authority (NFA) estimates that fraud and error in the public sector carries an annual cost of £20.6 billion to the taxpayer, representing an increase on a previous estimate of £20.3 billion. While this can be partly attributed to improved measurement techniques, it also indicates that the government’s existing prevention methods have not been successful. Recent reports from the National Audit Office (NAO) and the Public Accounts Committee show that HMRC has not been meeting its targets for fraud reduction. In these times of budget cutbacks, such waste cannot be tolerated, as frontline services inevitably suffer.
    
On a positive note, dealing with fraud and error is not just a moral necessity – it can also be a relatively straightforward way of recovering much-needed funds for key public services, and the government is already undertaking to improve the situation. In its report Tackling Fraud and Error in Government, the government outlines an “ambitious but focused delivery programme that seeks to reduce levels of
fraud and error across government.”

Key themes
The 2012 report is wide ranging and has several key themes running throughout it, including the need for improved intelligence sharing between government departments; focusing on prevention rather than waiting for problems to arise; using analytic software to identify areas of higher risk; and the implementation of stronger sanctions to deter fraud and error. The report also describes some of the successful initiatives that are already in place, such as the DWP and HMRC’s Prevent, Detect, Correct, Punish and Deter strategy for handling welfare crime.
    
While one method alone cannot guarantee reductions in fraud, the theme that consistently crops up is a need to use information more effectively. Government departments have at their disposal vast amounts of data that can be used to identify and combat fraud, but too often fail to use this information as an asset.

Sharing information
As public services are increasingly delivered online, the amount of public data held by government departments has grown massively. However, the insular culture in government departments has meant that they have failed to coordinate with one another, and so information that could be helpful to several departments is not shared. This prevents departments from building up a comprehensive analysis of fraud behaviour. Certain behavioural patterns are known to indicate potential wrongdoing, but it is extremely difficult to recognise these patterns using information from just one department.
    
However, ‘big data’ technologies represent a great chance for government to integrate cross-departmental data. Sophisticated analytics can be quickly and easily applied to all available data, yielding much greater returns in terms of knowledge and insight. The government has recognised the need

for greater cooperation between departments, and with this in mind has pledged to improve information sharing, not only within government but with law enforcement agencies. Intelligence Sharing Architecture (ISA) is vital to the government’s new approach. ISA aims to coordinate information retrieval and analysis, and thereby give a holistic view of fraud related behaviour.
    
Obstacles to integration
While these signs of a greater willingness to share information are encouraging, there remain a number of barriers in place before the government can fully take advantage of ISA. For example, the uncoordinated legacy systems of data management that are currently in place mean that information is difficult to join up across different departmemts – and fraudsters are fully aware of this, habitually perpetrating the same scams with several departments, confident that they won’t be caught. Inaccuracies need to be removed, and regular auditing is necessary to ensure they don’t mount up again.

Changing the culture
Entrenched attitudes represent another obstacle to greater information sharing. Concerns over data security lead to departments feeling that information should be kept hidden from other parties. While this insular culture is understandable, the case needs to be made that information sharing is a key line of defence against fraud, giving disparate departments the ability to recognise suspicious patterns.
    
Not all data is equally useful in identifying and preventing fraud and error. Of the vast amounts of information concerning the public, there is much that is impractical to keep and share. Analytics will help to identify the most valuable information.
    
A shift its focus is also required from a ‘pay now, check later’ culture to one that applies effective screening methods to prevent fraud from occurring in the first place. Screening of applicants for credits, benefits and grants needs to be stepped up. The ‘rules-based’ ethos currently favoured can only detect fraud that closely matches known behaviour patterns – and also runs the risk of giving ‘false positives’, identifying wrongdoing where none has taken place. As criminals adapt to bending and breaking new rules, analytics techniques need to keep one step ahead, for example by detecting anomalies in the system, or by using social network analysis to predict fraud before it becomes serious. This approach, known as ‘hybrid analytics’ has been explored by the software company SAS and its effectiveness has been proven.

Using data innovatively
Automated services represent a proven way to take advantage of detailed customer data. With this techonology, the government could act quickly on any anomalies detected, contacting claimants to inform them that the department concerned is aware of a change in their circumstances. Low-level threats could be easily dealt with which would give analysts more time to tackle more serious wrongdoing.
    
Finally, research undertaken by SAS shows that public sector employees don’t have a strong enough understanding of fraud behaviour. Almost half of the civil servants polled were unsure whether their department had carried out investigations into fraud losses, and just over half had received no training in combating fraud over the previous year. Better training is necessary if civil servants are to implement the zero tolerance approach outlined by the government.

Recovering revenues
The SIngle Fraud Investigation Service (SFI) could be a great help to the government in recovering revenues lost through fraud. In order for this to happen, government departments need to collaborate more consistently. SFIS and other public sector fraud organisations need to pool data and prioritise cases by criteria like the value of the fraud, error or debt to be recovered; the individual’s ability to pay; and the probability of the individual re-offending.  If data analytics are used effectively, anti‑fraud efforts can be a valuable source of revenue for the government.

Further information
SAS is a worldwide organisation specialising in business analytics software and services. Its solutions  helps customers at more than 65,000 sites improve performance and deliver value by making better decisions faster. To download a copy of the full report ‘Eliminating Public Sector Fraud and Error, visit the company’s website at www.sas.com