Executive Summary

National Audit Office Value for Money Report

1. Fraud, customer error and official error represent a major risk to the successful delivery of social benefits in the UK. Overpayments incurred due to fraud and error were estimated to amount to 2.6 billion for 2004-2005.[Footnote 1] Underpayments lead to distress for claimants, and poor administrative performance also has an adverse effect on the morale of Department for Work and Pensions (the Department) staff.[Footnote 2] The issue of fraud and error has been an important factor in the Comptroller and Auditor-Generals decision to give a qualified opinion of the Departments account for the past 16 years. Over recent years, the Department has undertaken a range of initiatives to tackle the problem. There has been progress in reducing fraud and error, although the effects of some recent initiatives remain to be assessed. However, overall losses from fraud and error have remained significant.
    
2. The Department considers fraud to occur when someone deliberately makes a false statement or representation to obtain a benefit or deliberately fails to provide relevant information.[Footnote 3] Within error, a distinction is made between customer error, where incorrect information is provided without fraudulent intentions, and official error. Official error is described as due to a mistake by an act or omission by staff, which the customer did not cause or materially contribute to, and which the customer could not, at the time they received the payment, reasonably have been expected to be aware.[Footnote 4]
    
3. The complex nature of the UKs social security system, which delivers a broad range of income-related benefits to a diverse population, is susceptible to error, and may also facilitate fraud.[Footnote 5] Such complexity, however, is not unique to the UK; rather, other countries with established social security systems face similar challenges. International benchmarking is routinely used by the NAO to provide fresh insights into Value for Money issues, and has also recently been called for by the Committee of Public Accounts. Investigating other countries experiences with fraud and error makes it possible to put the Departments performance in a comparative context, which is currently lacking. Moreover, it opens up possibilities for valuable lessons in the management of fraud and error.[Footnote 6]
    
4. The benchmark undertaken considered eight countries similar to the UK in terms of wealth and diversity of population.[Footnote 7] All of the countries acknowledge public responsibility for the provision of (some) social support, and spend very substantial amounts on benefit payments, most of which are tied to specific eligibility criteria (see Figure 1). At the same time, across the selection there is considerable variation in the organisation of government social security provision. Apart from France, Sweden, and the Netherlands, all countries considered participate in an annual conference on fraud and error (Six Nations Benefit Group), although it appears that approaches to controlling the problem of fraud and error have been developed in relative isolation. As a result, there is considerable scope for international learning. An overview of the different countries social security system and handling of fraud and error is provided in Figure 3. For readers outside the UK, relevant information relating to the Department for Work and Pensions is included both in this table, and as a country profile in Appendix A.
    
5. The delivery of social benefits involves large amounts of public expenditure, which is further increased by losses through fraud and error. Fraud and error therefore represent an auditing risk for departments responsible for benefit administration. Moreover, as a criminal activity, benefit fraud represents a serious legal and moral concern. The issue therefore is hard to ignore, and indeed all of the eight countries examined have put in place anti-fraud measures. In the area of error, the picture is less consistent. Overall, the comparative analysis of the different countries approaches produced the following insights:
   
   
   
   
   Rates of fraud at the Department appear comparable to those of other countries where comparisons can be made. 
   
   
6. Direct comparisons between the levels of fraud and error across the benchmarked countries can be difficult to make. In the case of error, data availability is poor, as a majority of the countries studied place less emphasis on this area. With regard to fraud, methods of quantifying the problem at national level differ considerably. Nevertheless, given that the Department has long been criticised for not bringing fraud below an acceptable level, it is important to make at least a tentative comparison with those countries for which relevant data is available: New Zealand, Canada, Ireland and the USA for specific benefits. The comparison shows that despite differences in structures, types of benefits and anti-fraud measures, total fraud and error rates in those countries where data is available mostly range between two per cent and five per cent of expenditure as shown in Figure 2. The occurrence of fraud and error in these countries is thus relatively similar to that measured in the UK. However, it should also be borne in mind that while the range of per centages is small, given the substantial sums spent on social support, even small differences between total or benefit-specific fraud levels can signify considerable amounts of public money lost or saved.
   
   Figure 2 ("Country comparison of fraud and error") is unavailable in this version of the executive summary.
   
   
7. Despite the observed comparability of fraud and error rates, the Department is the only responsible government department for whom fraud and error has had a severe impact on the external auditors judgement of departmental finances. This is because there is a statutory requirement for the NAO to have a view on whether expenditure in the accounts has been incurred in accordance with Parliaments intentions.[Footnote 8] Across the benchmarked countries, there exist varying responses from supreme audit institutions to the problem of fraud and error. Notably, in the Netherlands, in Australia and in France, auditors have criticised existing risk management in social security administration. Scrutiny of departmental performance in this area can thus constitute a complementary, rather than central, feature of the auditing process. Nevertheless, the comments have acted as a driver of anti-fraud initiatives in the countries concerned.
   
   Availability of data and methodologies for measuring fraud and error at national level vary considerably, but the Department is at the forefront in developing estimates of losses from fraud and error in social security expenditure.
   
   
8. The benchmark shows that the Department stands out for its attention to customer and official error, as most other countries concentrate on fraud. Its system of rolling reviews of benefit payments on the basis of large samples also measures fraud and error more comprehensively than the rest of the Countries. For example in France, fraud has not been captured systematically at national level. In Ireland, reviews are not as systematic as in the United Kingdom. In the USA, they are limited to certain benefit types. Furthermore, in many cases measuring focuses on other criteria than the total number of cases, and the total value of fraud and error. These criteria include:
    
   • Minimal accuracy targets (New Zealand, Sweden, Australia): These measurements focus on the proportion of accurate decisions to pay benefits and the proportion of accurate payments being made.
      
   • Totality of improper payments per benefit type (USA): This measurement based on performance and accountability reports estimates the total amount of improper payments per benefit type for fraud and error together. There is no system-wide aggregation of the total value of improper payments.
      
   • Savings and prosecutions achieved (Australia, Ireland, the Netherlands, and New Zealand): This measurement focuses on savings, performance and also prosecutions, based on observations and outcomes of reviewed cases. There may be targets for the number of cases that should be reviewed and even for the number of prosecutions (the Netherlands and New Zealand). In Ireland, multipliers are used to produce an aggregate number of savings made within the system. Centrelink (Australia) counts as fraud only those cases successfully prosecuted in the court of law.
      
   • Sampling in specific benefit schemes (Ireland and USA): Measurements of the baseline fraud and error rates are targeted on specific benefit schemes, which are perceived at high risk from fraudulent claims.
      
   • Benefit-specific definition of fraud (the Netherlands): Social security agencies have different definitions of fraud and error and consequently different ways of measuring fraud and error per benefit type. In Canada, there are also variations in measurement between the provincial and federal levels, making aggregation difficult.
      
9. The Departments assessment of the overall incidence of fraud appears a suitable tool for developing policy measures and monitoring their impact. This is notable given the importance of informed policy making for reducing the losses resulting from fraud and error, of which all of the benchmarked countries are aware. Some of the differences in approaches can be traced to systemic or cultural factors, such as decentralisation of administration, independent social security funds, or political climate. For example, in Canada, benefits are administered both at the federal and provincial levels; in Australia, benefit administration is outsourced to a central agency; in France, the role of the state is limited to that of supervisor of independently run social security funds; in Sweden, a strong emphasis is placed on the states, rather than the customers responsibility to ensure that payments are correct; and in the Netherlands, the autonomy of different agencies administering benefits means that a variety of definitions of fraud and error are being used. There are also some changes underway: in France, the first nationwide, cross-benefit study of fraud is currently being conducted, and its results will be scrutinised by the national supreme audit institution in 2006.
   
   The Department compares favourably in terms of awareness of fraud and error, and activities to combat the problem.
   
   
10. All of the countries examined acknowledge that benefit fraud is a problem that needs to be addressed, both because of its illegality and because of the losses incurred. In response, a wide variety of approaches to tackle the issue have been developed. However, the Department not only shows above average awareness of the specific issue of error, but is distinguished by a comparatively comprehensive range of actions both in the area of prevention and of detection. Moreover, whilst some of these initiatives are recent, overall the Departments record of launching and monitoring activities to counter fraud and error seems more established than that of its peers. The Departments comparatively high level of awareness and activity is likely to be partly linked to its experience of regular scrutiny by the NAO, and the resulting sharp parliamentary and public criticism. Nevertheless, there may be some scope for developing further initiatives based on the experiences of other social security administrations.
    
    
11. The prevailing political and economic climate is a notable influence on the general prominence of benefit fraud. Although fraud is always a legal matter, its impact on public finances is a function of the overall number of claims and support paid out. For example, in France, a sharp rise in unemployment led to widespread calls to clamp down on misuse of the social security system in the early 1990s. In Canada, by contrast, provincial bids to take a tough line on social support, for example in Ontario and British Columbia, failed to win the support of the broad population over time.
    
    The Department, with support from the National Audit Office, could lead an international exchange on the measuring and management of fraud and error.
    
    
12. Despite the unique national context and internal structures of the social security systems, issues surrounding fraud are similar. This is evident for the causes of fraud and for anti-fraud actions. Common trouble spots include illegal work, document and identity fraud, weakness of internal controls, staff burden and systemic complexity. The most common responses are risk assessment, (random) checks and sampling, improvements to information technology (IT) infrastructure, communication with the public and staff training, as well as dedicated anti-fraud groups.
    
    
13. Because there are shared concerns, the benchmark countries are likely to benefit from sharing experiences regarding their effort to identify and counter these issues. The Six Nations Benefit Group is an existing forum that could accommodate such exchange, but it does not appear to have achieved widespread mutual awareness. The Department is well placed to take the lead in encouraging an open and practice-oriented process of international learning. As the auditing body judging the Departments progress in containing and reducing fraud and error, the NAO represents a natural partner in facilitating this process. This joint approach makes it possible to approach the management of fraud and error from multiple angles, while also considering implications for national audits.
    
    Other countries initiatives to tackle the problem of fraud may be of potential interest to the Department.
    
    
14. The Department has important experiences to share, but also stands to gain from learning about novel approaches developed by other administrations. Box 1 shows an overview of some of the innovative or interesting tools to improve a variety of aspects of the management of fraud and error, ranging from prevention to detection. Particular examples of interesting practice include the use of:
     
    • random response surveys to examine the motivations for fraudulent behaviour to assist in risk profiling and the directing of control measures (the Netherlands);
       
    • specific instruments to understand the characteristics of fraudulent behaviour[Footnote 9];
       
    • unique identifying numbers (identification number) to allow for comprehensive inter-agency data-matching and tracking of claimants in a system (France and the Netherlands);
       
    • a central national database for customer records (Australia and France [planned]);
       
    • cost-benefit calculations to determine the effectiveness of control measures (New Zealand, Australia, and the Netherlands);
       
    • targets for the number of reviews and specific types of reviews (Australia and New Zealand);
       
    • a single core benefit with one set of rates, eligibility criteria and add-ons (New Zealand); and
       
    • instruments such as information sessions and contracts/declarations in benefit claim forms, which emphasise the rights and obligations of claimants (Canada and the Netherlands).
       
15. Each of these tools is worth being examined closely by the Department, both to learn what has worked in other countries, and to assess the potential for use within a UK context. In practice, the adoption of certain approaches employed successfully elsewhere may not be straightforward. For example, several of the benchmarked countries use inter-agency data-matching, which in the UK may conflict with both data protection rules and technical compatibility. Nevertheless, other countries experiences in such areas will provide important input for an informed debate of such issues. 

1.  [back from footnote 1]Department for Work and Pensions Resource Accounts 2004-2005, HC 447 2005-2006.
    
2.  [back from footnote 2]House of Commons Committee of Public Accounts (2005); NAO (2004), Department for Work and Pensions Resource Accounts 2003-04. Report by the Comptroller and Auditor General; NAO (2005), Department for Work and Pensions. Dealing with the Complexity of the Benefits System, Report by the Comptroller and Auditor General, HC 592 Session 2005-2006, 18 November 2005; NAO (2003), Department for Work and Pensions. Tackling Benefit Fraud, Report by the Comptroller and Auditor General, HC 393 Session 2002-2003, 13 February 2003.
    
3.  [back from footnote 3]The Social Security Administration Act 1992; Social Security Fraud Act 2001.
    
4.  [back from footnote 4]Housing Benefit and Council Tax Circular, HB/CTB A28/2004.
    
5.  [back from footnote 5]NAO (2005), Department for Work and Pensions. Dealing with the Complexity of the Benefits System, Report by the Comptroller and Auditor General, HC 592 Session 2005-2006, 18 November 2005, London: NAO.
    
6.  [back from footnote 6]House of Commons Committee of Public Accounts (2005), p. 3.
    
7.  [back from footnote 7]Additional information on the selection of the countries and the research approach of this report is found in Appendix C.
    
8.  [back from footnote 8]Government Resources and Accounts Act 2000.
    
9.  [back from footnote 9]Aside from random response surveys, other examples of programmes, which aim to understand the characteristics of fraudulent behaviour, include Fraud and Error Surveys in Ireland and the use of neural networking and data mining in the USA. These can contribute to a better understanding of risks in the social security system.