S P D I S C U S S I O N PA P E R NO. 1002 53889 Toolkit on Tackling Error, Fraud and Corruption in Social Protection Programs Christian van Stolk Emil D. Tesliuc March 2010 Toolkit on Tackling Error, Fraud and Corruption in Social Protection Programs CHRISTIAN VAN STOLK RAND EUROPE EMIL D. TESLIUC THE WORLD BANK MARCH 2010 Abstract Because Social Protection (SP) and Social Safety Net (SSN) programs channel a large amount of public resources, it is important to make sure that these reach the intended beneficiaries. Error, fraud or corruption (EFC) reduces the economic efficiency of these interventions by decreasing the amount of money that goes to the intended beneficiaries, and erodes the political support for the program. While no program is immune to EFC, evidence from developed countries demonstrates that such leakage can be brought to negligible levels. In five OECD countries this fraction is between 2-5% for the SP sector as a whole. For SSN programs, which use more complex eligibility criteria and hence are more prone to EFC, this fraction is 10%. To achieve these results, programs have implemented a number of measures reviewed in this note. In contrast, efforts to combat or even measure EFC are quite rare in developing countries, although some programs are plagued by it. This toolkit is intended to provide a concise but thorough resource for social protection practitioners on how to minimize error, fraud and corruption in their program(s). It is organized in five sections. The first section introduces the topic, by clarifying the key concepts and spelling out the rationale for this activity. The second section provides a generic framework for combating EFC, developed around four building blocks of prevention; detection; deterrence; and measurement. The third section reviews the instruments, tools and mechanisms used to combat EFC in social protection programs, structured by strategic actions (prevention; detection; deterrence), by level of government and over time (how these tools and instruments can be developed over time). Section four includes generic terms of reference for the assessment of the mechanisms for combating error, fraud and corruption (EFC) in social protection programs. The last section summarizes the main findings for three such diagnostics from the Kyrgyz Republic and the Ukraine. JEL Classification: I38, H53, H75, H83, G28, D73 Keywords: social protection, social safety nets, governance, error, fraud, corruption ii Contents Table of Figures and Tables .......................................................................................................... v CHAPTER 1 Introduction .................................................................................................... 1 1.1 What is the structure of the toolkit?....................................................................................1 1.2 Why focus on error, fraud, and corruption in social protection programs? ..........................2 1.3 How are error, fraud, and corruption defined?....................................................................3 1.4 What is the difference between EFC and targeting error? ...................................................4 1.5 What is an acceptable level of EFC in social protection programs? ......................................4 1.6 What are the main causes of EFC?......................................................................................6 1.7 What is the scope of this toolkit? ........................................................................................7 1.8 How should this toolkit be applied? ...................................................................................8 1.9 For which social protection programs is this toolkit relevant? .............................................8 CHAPTER 2 A framework for combating EFC ................................................................... 10 2.1 What are the main elements of a strategy to combat EFC?................................................10 2.2 What is needed to implement the strategy? .......................................................................11 2.3 How can the strategy be adapted to country or program circumstances? ...........................13 CHAPTER 3 Tools and instruments used to combat EFC .................................................. 15 3.1 What are the main instruments to combat EFC per aspect of the strategy? .......................15 3.2 Are there specific instruments for tackling EFC respectively? ............................................20 3.3 How do OECD instruments aimed at mitigating EFC develop over time? .......................20 3.4 How are the OECD instruments to mitigate EFC currently applied in less developed countries? .........................................................................................................................22 3.5 What are the main instruments to combat EFC per level of government? .........................24 3.6 How does this approach on EFC control inform other diagnostic initiatives? ...................24 CHAPTER 4 Terms of references for a diagnostic assessment in client countries ................ 27 4.1 What are the key questions to be asked? ...........................................................................27 4.2 What type of data collection takes place? ..........................................................................27 4.3 Who are the key informants in this diagnosis? ..................................................................28 4.4 How are recommendations arrived at? ..............................................................................28 4.5 What should the composition of the team be? ..................................................................28 4.6 What are the costs of the diagnostic mission?....................................................................29 CHAPTER 5 Examples from missions so far ....................................................................... 31 iii 5.1 Ukraine ............................................................................................................................31 5.1.1 What is the problem of EFC? ..............................................................................31 5.1.2 What type of EFC mitigation is in existence?.......................................................31 5.1.3 What can be done to improve EFC mitigation? ...................................................32 5.1.4 What has been the impact of the diagnosis? .........................................................33 5.2 The Kyrgyz Republic ........................................................................................................34 5.2.1 What is the problem of EFC? ..............................................................................34 5.2.2 What type of EFC mitigation is in existence?.......................................................34 5.2.3 What can be done to improve EFC mitigation? ...................................................35 5.2.4 What has been the impact of the diagnosis? .........................................................36 5.3 Bangladesh .......................................................................................................................36 5.3.1 What is the problem of EFC in the Employment Generation Program? ..............36 5.3.2 What type of EFC mitigation is in existence?.......................................................37 5.3.3 What can be done to improve EFC mitigation? ...................................................37 5.3.4 What has been the impact of the diagnosis? .........................................................38 5.4 EFC in Conditional Cash Transfer Programs in Latin America and the Caribbean...........38 Resources....................................................................................................................................41 5.5 United Kingdom ..............................................................................................................41 5.5.1 Australia ..............................................................................................................41 5.5.2 Canada ................................................................................................................42 5.5.3 France .................................................................................................................43 5.5.4 Ireland .................................................................................................................43 5.5.5 The Netherlands..................................................................................................44 5.5.6 New Zealand .......................................................................................................45 5.5.7 Sweden ................................................................................................................46 5.5.8 USA ....................................................................................................................46 APPENDICES .......................................................................................................................... 49 Appendix A: Template for case study ..........................................................................................51 Appendix B: TOR and work program for the World Bank mission in the Kyrgyz Republic ........55 Appendix C: Program design and management with EFC lenses ................................................59 Appendix D: Approach to measuring fraud and error in the United Kingdom ............................61 Appendix E: Job descriptions of officials in the UK ....................................................................62 Appendix F: Definitions of fraud and error in the UK ................................................................65 iv Table of Figures and Tables Figure 1: Typology of fraud and error in benefit systems....................................................................3 Figure 2: A comparison of fraud and error rates in 5 OECD Countries .............................................5 Figure 3: A strategic framework to control EFC ...............................................................................11 Figure 4: EFC oversight mechanisms per level of government ..........................................................24 Table 1: Main instruments per aspect of the strategy........................................................................18 Table 2: A view on the development of EFC strategies in OECD countries over time......................21 Table 3: Good practice of oversight and control mechanisms from the Latin America region ...........23 Table 4: Assessing control and accountability mechanisms in a CCT ...............................................26 Table 5: EFC lenses on critical phases of the program ......................................................................59 v CHAPTER 1 Introduction 1.1 What is the structure of the toolkit? The report has five main substantive sections based on the terms of references drown up. These are: (i) Introduction to the topic: ˇ Why it is important to focus on reducing and monitoring EFC ˇ Definitions (error, fraud, corruption, other specialized terms) ˇ What is the difference EFC and targeting error ˇ What is a reasonable level of EFC ˇ Scope of the work (what this diagnostic tool includes and excludes) ˇ Type of programs and countries where this toolkit is applicable (ii) Generic framework for combating EFC ˇ Main elements of the strategy to combat EFC (prevention; detection; deterrence; measurement) ˇ Set of instruments, tools, mechanism to implement the strategy ˇ Adapting the strategy to country/program circumstances (iii) Instruments, tools and mechanisms used to combat EFC in social protection programs ˇ What are the tools structured by strategic action (prevention; detection; deterrence) ˇ Are there specific tools for EFC respectively ˇ How have they developed over time and been used in less developed countries ˇ Structured by level of government (see Annex 1) (iv) TORs for the Diagnostic toolkit for the assessment of the mechanisms for combating error, fraud and corruption (EFC) in social protection programs. The TORs will include: ˇ Key questions to be addressed ˇ Suggested composition of the team (key qualification of the consultants); ˇ Data collection techniques ˇ Key informants ˇ Estimated time & costs, based on the work in the Kyrgyz Republic and the Ukraine (v) Examples: 1 ˇ Thumbnail sketch of the country reports for the Kyrgyz Republic and the Ukraine 1.2 Why focus on error, fraud, and corruption in social protection programs? Social protection programs channel a large amount of public resources to targeted beneficiaries. Even a small fraction of misappropriated benefits may add up to large sums of money with high opportunity costs. On average, social protection spending represents 15.7% of GDP in developed countries, 7.4% in middle-income countries (MICs) and 3.8% in low-income countries (LICs). 1 Each social protection program aims to provide the right amount of benefit to the right recipient on time. However, the evidence shows that no social protection program is immune from error, fraud, and corruption (EFC). A 2006 international benchmark study on fraud and error in social security systems written by RAND Europe for the UK National Audit Office (NAO) puts the range of fraud and error in social protection systems where data was available between two to five percent of overall government expenditure on social security. 2 Thus, losses to the social protection system can be considerable. For instance, in the UK over Ŗ2 billion (around US$3.2 billion) of government expenditure is lost on an annual basis due to fraud and error. In all countries, error, fraud and corruption (EFC) also present significant challenges to the integrity of social protection systems. A higher prevalence of EFC can be associated with: ˇ An inefficient use of public resources; ˇ Availability of fewer funds to the poor; ˇ A loss of credibility of the program; ˇ Political manipulation of the funds; and ˇ A decrease in public support for the program. EFC is likely to be more prevalent in social protection programs of less developed countries as a proportion of overall spending compared to OECD countries due in part to the limited administrative capacity, absence of adequate monitoring of the overall problem and clear evidence- based strategies to combat EFC. 1 World Bank 2007 FEC SSN Primer note. 2 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006) 2 1.3 How are error, fraud, and corruption defined? This toolkit looks at misappropriations under the rules of a social protection program. These misappropriations are due to fraud, error or corruption (see Figure 1): 3 Figure 1: Typology of fraud and error in benefit systems Source: RAND Europe Fraud refers to intentional behaviour on the part of the benefit claimant to defraud the benefit system. The main causes of fraud are benefit claimants being deliberately dishonest on benefit forms, claimants deliberately exploiting the system by providing a false identity and the complexity of the benefit system, which offers opportunities and incentives for claimants to defraud the system. Error refers to unintentional mistakes on behalf of benefit claimants or staff in the benefit office. Some causes of error are the absence of staff training, the failure of systems to process or pay benefits applications, and again the complexity of the benefit system, which means that customers and staff can make mistakes in dealing with complex or complicated processes. An intentional attempt by staff to exploit the social protection system is called corruption. In many OECD countries, corruption is seen as a low risk due to the integrity measures built into the system for processing and paying out benefits (for instance separation of functions and IT systems) as well as factors such as proper staff training, results-based management and eligibility requirements for benefit claimants. This risk is likely higher in the middle to low income countries, due to weaker systems for audit, oversight and control, and the relatively low pay of the social protection workers. For a definition of error, fraud and corruption used in the UK, see Appendix G. 3 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006) 3 1.4 What is the difference between EFC and targeting error? The notion of error (and fraud and corruption) used in the toolkit is distinct from the notion of targeting error (as typically used, for example, in the World Bank). The two concepts refer to different aspects of program design and implementation. Targeting error is often used to indicate the fraction of program funds or beneficiaries of a social protection program that are not reaching the poor. Concerns about targeting errors are important for a subset of social protection programs, i.e. safety net program with anti-poverty objectives, such as last-resort programs, workfare programs or conditional cash transfer programs. The extent of targeting errors indicates whether the program achieves its (poverty alleviation) objective or not. These errors may be due to program design (when, for administrative feasibility, the program uses imperfect poverty proxies to identify poor beneficiaries) and to program implementation (when eligibility decisions diverge from the program rules). The later errors are, in turn, due to error, fraud or corruption. For other types of social protection programs that have other primary objectives than poverty alleviation (for example, pensions, unemployment insurance or other social insurance programs), targeting errors are less relevant. In contrast, concerns about error, fraud and corruption apply to all social protection programs, and indicate the extent to which the program is compliant with its own rules. Consequently, one can find a social safety net program which is well targeted although it has some level of EFC. This is the case of an income-tested program with a lower eligibility threshold (say, $10) than the poverty line (say, 20$). If half of its beneficiaries would have incomes below $10, and the other half between $10 and $20, the program will have perfect targeting (all beneficiaries are poor) but 50% of the caseload will be affected by error, fraud and corruption. More often, one can find SSN programs with high targeting errors but low levels of EFC. For example, an universal child allowance program may be put in place with the objective of alleviating poverty in a country where there is a strong correlation between the number of children and poverty status, and the Government lacks the administrative capacity to test who is income poor or not. For example, one third of the population may be poor, but two thirds of the children may be living in poor (and larger) families. In this case, a third of the program caseload is targeting error. If the government can test, however, whether those who receive the program are in the correct age brackets, then the level of error, fraud and corruption would be zero. 1.5 What is an acceptable level of EFC in social protection programs? The evidence in OECD countries shows that some level of fraud and error in the social protection system is unavoidable. A 2006 UK NAO study found that fraud and error rates in five OECD countries ­ the UK, Canada, Ireland, New Zealand and the USA ­ranged between 2% and 5% of overall expenditure on social protection programs (Figure 2). Within the social protection system, means-tested social protection programs had the highest fraud and error rates (5-10%), followed by 4 unemployment benefit and disability pension programs (1-2%). Old-age pensions had the lowest rates (0.1-1%). These figures should be interpreted as the lower bounds for the amount fraud and error rates, as they come from a small sample of countries and programs with high administrative capacity, with adequate procedures to minimise EFC, including in many cases procedures for estimating the level of fraud and error with reasonable accuracy. Figure 2: A comparison of fraud and error rates in 5 OECD Countries UK Canada Ireland New Zealand USA As % of expenditure/payments (2004-2005) (1994, 2003) (2004) (2001) (2003) Total Fraud and Error 2.3% 3-5% ('94) - 2.7% - Fraud and Error in Housing Benefit 5.2% - - - 3.6% Fraud and Error in Income Support 5.3% 3.5% ('03) - - 5.07% Fraud and Error Old Age, Disability 4.9%, 0.1%, 1.9% 2 - 7% - 0.53% Source: NAO (2006) Notes: 1 Comparisons are difficult as benefit types differ between countries. This table shows rates for the following benefits. The Income Support benefit for Canada and the US is Unemployment Insurance. In the category, Old Age and Disability, the benefit in Ireland is Disability Allowance and in the US Old Age, Survivors and Disability. Also measurements vary. Rates for the UK refer to fraud and error as % of total expenditure. Rates for Canada (`03) refer to the total savings identified as a percentage of total payments. Ireland measures fraud and error as a percentage of expenditure for specific benefit types. New Zealand estimates this percentage as the number of cases sampled that contained errors, which led to benefits being paid incorrectly. The USA measures the total value of improper payments per benefit type. 2 These rates are for Pension Credit, Retirement Pension, and Disability Living Allowance respectively. The information from developing countries is scarcer, as only few programs and countries have taken measures to combat and/or measure the incidence of EFC. However, it is plausible to expect the share of funds affected by EFC be larger compared to developed countries. For example, in India a Government audit in 2003 found that 31.6% of IAY (housing programs) and 53.5% of SGSY (credit program) funds did not reach the intended beneficiaries. In the case of the SGRY, an employment program, administrative data report that 31.3 million man-days of work were generated in 2001-02, while the National Sample Survey data for 2002-03 estimate this number as less than 3 million man- days, only 10% of the official figure. Three points can be made about EFC in developed and developing countries: (1) Across SP programs, the share of funds affected by EFC is larger in SSN programs, due to their inherent design features. Eligibility for these programs is often based on the economic position of the household, which is more difficult to assess and harder to verify than, for example, for an old-age pension where all you need is proof of past contribution and age. Because SSN programs are more narrowly targeted to poorer households and poverty status changes over time, eligibility also changes over time. The responsibilities for implementing SSN programs are often shared across different departments, agencies, and levels of government. All these factors increase the risk that benefits do not reach intended beneficiaries in time or in the correct amount. Having 5 the right instruments and tools to minimize EFC is thus more important for SSN programs, although there are economies of scale when the efforts to combat EFC cover the whole set of SP programs. (2) The level of EFC is likely correlated with the institutional capacity of the country and implementing agency. One would expect to find more EFC in LICs and MICs than developed countries. However, evidence is scarce as very few programs or countries measure the level of EFC using comprehensive, unbiased measurement tools. (3) In developed countries there is a much greater concern about error and fraud, while developing countries worry more about corruption. For example, none of the 9 OECD countries reviewed in NAO (2006) explicitly reported issues around corruption of departmental staff. It seems a small problem for administrators in these countries. Reasons could include a strict separation of functions (e.g. between payments and assessments, between decision-making and investigations and reviews), staff training and management, effectiveness of internal audit and the integrity of systems processing payments. In contrast, corruption is often the lead concern in developing countries. 1.6 What are the main causes of EFC? A wide range of causes for fraud are identified in and some of the pilot work. This section highlights some of the main causes. They are presented by main area. Fraud ˇ Customer dishonesty: This includes undeclared income and failing to report changes in material circumstances; ˇ Exploiting the system: This includes multiple program claims and misrepresentation of material circumstances and identity fraud; and ˇ Complexity of the social protection system: This includes cross-jurisdictional claims and complexity of rules and regulations. Error ˇ Staff: This includes excessive staff caseloads, inadequate support and training of case managers and team coaches, breakdown or override of internal control; ˇ System: This includes the failure of payment system, failure of IT systems, problematic information management, and inadequate monitoring or reporting procedures; and ˇ Complexity: Complexity of benefits and rules confuse administrators and benefit claimants. Corruption Corruption is not reported as a particular problem in OECD countries. The pilot missions in the Kyrgyz Republic, Ukraine, and Bangladesh showed problems around: 6 ˇ Exploitation of design aspects of the program (e.g. lax eligibility rules and absence of independent monitoring) by staff administering the social protection program; ˇ Collusion between staff administering social protection programs and the claimant; ˇ Staff taking bribes or favours to enrol individuals in the program; and ˇ Collusion between staff processing benefit claims and those paying out benefit claims. 1.7 What is the scope of this toolkit? The main policy objective in EFC reduction in social protection programs is compliance and asks whether the program follows its own rules. In this sense, EFC mitigation can be an input to program design and management, but not a substitute for what is considered good program implementation. In most OECD countries, teams in social protection departments involved in EFC mitigation provide advice on program design and management in order to minimise the occurrence of EFC. As we will see in Chapters 2 and 3, there are overlaps between EFC mitigation interventions and what would be more widely regarded as good program implementation. In essence, EFC control is an element that can strengthen program design and implementation. However, EFC mitigation or the reduction of wasteful spending in a social protection program is only one objective potentially competing with others in program design and implementation. This toolkit has a twofold purpose in diagnosing the EFC problem in the social protection programs. The toolkit should enable government officials and World Bank staff to: ˇ assess where data allows the error, fraud and corruption rate of social protection programs in client countries; and ˇ evaluate the effectiveness of actions taken by administrations to tackle and monitor EFC. The former provides a picture of the extent of the overall EFC problem in the country. It also allows for an assessment of how fraud and error rates and/or targeting in social protection programs compare to more similar countries and OECD good practice. This gives a potential scope for improvement for EFC control in social protection programs. In making comparisons, it is important to compare fraud and error rates and/or targeting of similar social protection programs (e.g. means-tested unemployment benefit, categorical, etc.), as fraud and error and/or targeting benchmarks can be significantly different between types of social protection programs. The latter looks at the effectiveness of actions taken by client countries to mitigate the problem of EFC in social protection programs. An assessment of effectiveness is made easier by comparing the initiatives of client countries to the evidence base on the effectiveness of initiatives in other countries, particularly in OECD countries. 7 1.8 How should this toolkit be applied? In formulating recommendations on how initiatives to mitigate EFC can be improved, it is important to keep the country context in mind. Any recommendation for improvement needs to pass the test of feasibility, suitability, and acceptability. In other words: ˇ how feasible is a recommendation given the institutional architecture and administrative capacity of the country; ˇ how suitable or effective are the actions to address the particular EFC problem; and ˇ how acceptable are these actions likely to be to political decision-makers, administrators, the claimants and wider stakeholders? These three filters are important as `good practice' is only relevant if it can be applied and effective in tackling the EFC problem. This then leads to a realistic assessment on how the EFC strategy can be improved in the short, medium, and longer term. Ideally, evaluators would form a strong set of assessment criteria in order to understand what is feasible, suitable, and acceptable. A close dialogue with policy makers also informs this assessment. 1.9 For which social protection programs is this toolkit relevant? This toolkit is designed to look at all social protection programs; however it focuses on those programs with higher risk of EFC. Programs with complex eligibility criteria (like last-resort social assistance programs, or disability programs) tend to have higher rates of EFC than pensions, other social insurance programs, or categorical SSN programs. In general, these types of programs are at a lower risk of EFC, because eligibility is easier to establish and verify in most cases. The 2006 NAO study found that old-age pensions were at the lowest risk of fraud and error at 0.1- 1% (see Figure 2). 4 However, the evidence shows that means-tested social protection programs in some OECD countries are at a higher risk of fraud, with a rate of fraud and error to overall expenditure between 5 -10% of total program expenditure, higher than the range of 2-5% for programs overall. There two main reasons for this higher risk: ˇ Complex eligibility requirements; and ˇ Changing eligibility over time due to changes in personal circumstances. Though we have limited evidence on the rates of fraud and error in less developed countries (MICs and LICs), it appears logical to conclude that the risk of fraud in similar social protection programs is higher in these countries compared to OECD countries for the following, to some extent interrelated, reasons: 4 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006) 8 ˇ Difficulties in verifying income (e.g. due to problems on how to account for material possessions, income derived from the grey economy, and/or remittances); ˇ Increased risk of corruption of officials; ˇ Frequent changes in eligibility requirements and administrative procedures of programs; ˇ Vagueness of eligibility requirements and standard operating procedures; ˇ Higher reliance on indirect distribution of benefit (rather than direct transfer from government to the beneficiary); and ˇ Limited administrative capacity to verify eligibility or detect fraudulent or corrupt behaviour. These assumptions are supported by the examples reviewed in Chapter 5: means-tested social protection programs in the Ukraine and the Kyrgyz Republic aimed at poverty alleviation; and the main public works program in Bangladesh, the Employment Generation Program (EGP). 9 CHAPTER 2 A framework for combating EFC 2.1 What are the main elements of a strategy to combat EFC? To establish good practice on strategies to tackle EFC, it is useful to look at the EFC control strategies of developed countries. These strategies tend to be the most comprehensive and evidence on their effectiveness exists. Developed countries' experience shows that reducing EFC is feasible and cost- effective. For example, the US Food Stamps Program reduced its overpayment rate from 10% a decade ago to about 4%. From 1997 to 2006, the UK halved the level of fraud in the social protection sector. To obtain these results, programs invested in developing appropriate procedures, staff and IT support. The 2006 NAO study on fraud and error in international social security systems also outlined that most OECD countries use a basic suite of interventions to control fraud and error in social security systems and to ensure compliance with eligibility requirements. 5 These activities cover the areas of prevention, detection, and deterrence. In this sense, these countries take a similar systematic approach to controlling EFC focusing on keeping fraud and error out of the programs upfront, finding fraud and error in existing claims, and deterring claimants from committing fraud. There is evidence from the United Kingdom and other OECD countries that the typical strategic framework works. Evidence suggests comprehensive strategies using multiple, interdependent, instruments aimed at prevention, detection and deterrence have a positive impact in reducing EFC. 6 In addition, OECD countries are increasingly trying to measure the effectiveness of the strategy. Such measurements allow social security administrations to track the volume of fraud and error over time and assess to a degree whether their initiatives to control EFC are having a positive impact. Increasingly, social security administrations are also assessing the cost-effectiveness of specific initiatives in the areas of prevention, detection, and deterrence to establish an optimal portfolio of effective instruments to control EFC. 7 Performance assessment increases the accountability for EFC in the social security or benefit administrations, but also them to refine their initiatives to increase the impact over time. For instance, the strategy of the Department of Work and pensions consists of the following elements: 8 ˇ Preventing fraud. Actions taken include: tighter checks at outset of claim; payments direct to bank accounts instead of by cheque; staff training; and fraud-proofing new policy and operational developments. 5 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006) 6 NAO (2008), Progress in Tackling Benefit Fraud, (HC 102 Session 2007-2008). 7 NAO (2008), Progress in Tackling Benefit Fraud, (HC 102 Session 2007-2008). 8 "The United Kingdom's Experience in Reducing Error & Fraud," http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTSOCIALPROTECTION/EXTSAFETYNETSANDTRANSF ERS/0,,contentMDK:21506638~isCURL:Y~menuPK:4159815~pagePK:210058~piPK:210062~theSitePK:282761,00.html 10 ˇ Detecting fraud. Actions taken include: data matching (on benefit and other data); investigation; intelligence; risk profiling; telephone hotline for reporting fraud; joint operations with other agencies. ˇ Deterring fraud. Actions taken include: extended sanctions; measures to influence public attitudes ­ public awareness campaign `Targeting Benefit Thieves'; and confiscation of assets. ˇ End-to-end approach. Proposed action on four fronts: o Getting it right - aiming to get benefit payments correct from the start; o Keeping it right - ensuring payments are adjusted as circumstances change; o Putting it right - detecting when payments go wrong and taking prompt action to correct them with appropriate penalties to prevent a recurrence; o Making sure the strategy works - monitoring progress, evaluating the strength of preventive measures and adjusting them in the light of experience. Figure 3 gives an overview of a typical strategic framework to control EFC. Figure 3: A strategic framework to control EFC - Improved verification on eligibility Prevention - Provision of information to applicants and public Deterrence - Sanctions - Risk-based and Detection random reviews - Data matching Measurement - Telephone hotlines - Often linked to performance management Source: RAND Europe 2.2 What is needed to implement the strategy? There are two main factors that affect the implementation of the strategy. Firstly, for successful implementation of a comprehensive strategy there has to be a political will to tackle the issue of EFC in social protection programs. This political will can also be closely associated with wider political and administrative culture. For instance, in Sweden, the threshold for committing fraud is higher 11 than in other countries as the state shares much responsibility for getting the right information on benefit claimants. This emphasis has led to what other countries would consider fraud to be categorised as error. In general, Sweden is not as tough on fraud as perhaps Anglo-Saxon countries, which strongly emphasise individual responsibility in supplying the right information to the administration and therefore have a lower threshold for categorising fraud. This political will can also affect how fraud is measured within the system. Politicians and administrators might be reluctant to publicise the extent of fraud within the social protection system. Such measurement might produce unwelcome news or raise the interests of external accountability bodies such as parliament or supreme audit institutions. In some cases, politicians and administrators might be implicated in some of the EFC occurring in the social protection program. Several factors can influence the political will to address EFC: ˇ Pressure from external accountability bodies to address EFC; ˇ Public concern over levels of EFC; ˇ Enlightened political or administrative leadership to combat EFC in social protection programs; and ˇ The business case: Increasingly benefit administrations take a pragmatic approach and find that investments in anti-EFC measures pay off in terms of the amount of overpayments identified (the investment argument). 9 Secondly, administrative capacity plays an important role in whether the strategy can be implemented successfully. A number of aspects of administrative capacity seem particularly important from previous research10 and missions: ˇ Staff: Administrations need the right number of trained and qualified staff to administer benefits and control for EFC. In terms of EFC, these staff need to be trained specifically in terms of detecting EFC within the system. Understaffing was a major issue in the countries visited in the pilot missions. ˇ Results-based management: The example of Canada shows the effectiveness of incentivising line managers through performance assessment to reduce the amount of fraud and error in benefit files across the organisation. In other words, incentivising staff to do a good job and rewarding staff are important. ˇ Material support: Staff need adequate offices to work in and material support such as transportation and reimbursement of travel expenses. Lack of material support can limit the effectiveness of administration staff but also impact staff morale. ˇ Resources: Many social protection programs have moderate resources and are limited in size. Extensive EFC controls might not be proportionate to the size of the program. ˇ IT: IT capacity is important in the effective and efficient processing benefit claims. Moreover, IT is increasingly used to verify data that is submitted by claimants by cross-referencing this data 9 NAO (2008), Progress in Tackling Benefit Fraud, (HC 102 Session 2007-2008). 10 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006). 12 against other databases held by government or the private sector. Studies have showed that such `data-matching' is a particularly cost-effective way of tackling EFC in some OECD countries. 11 ˇ Networks: Building networks across government is important in fostering cooperation and information exchange between government departments and agencies. The absence of such cooperation can be a limiting factor in the prevention and detection of EFC and the application of sanctions to fraudulent claims. The pilot missions in the Ukraine and the Kyrgyz Republic showed the importance of cooperation between the benefit administration and the tax authority in sharing information on benefit claimants, as well as the cooperation between the benefit administration and judiciary in applying sanctions. ˇ Public: The public and wider community is likely to have information that could be helpful in identifying EFC in social protection programs. Having good relations between the administration and community and creating public awareness around the seriousness of EFC can be important in efforts to reduce EFC. In addition, making the public aware of the rules of the program also can create a social control mechanism over who is eligible under the program. Publicising beneficiary lists and the rules of the program are widely regarded as good practice. 2.3 How can the strategy be adapted to country or program circumstances? In essence, Section 2.1 sketched the basic aspects of an ideal type of strategy covering the main areas of prevention, detection, deterrence, and measurement. The experiences of the OECD countries can hold valuable lessons for LICs and MICs even though these experiences might not all be directly transferable to, or applicable in, their respective contexts. Nonetheless, many of these OECD countries have seen important evolutions of and changes in their anti-fraud strategies and thus offer some insights into how the components of an anti-fraud strategy develop over time. Moreover, in many cases these countries have a solid understanding of what types of interventions are likely to be effective. `Value for Money' reports as those undertaken by the UK NAO add to this understanding of effectiveness. 12 Finally, there is evidence that some of the tools to control the level of error and fraud which are used in OECD countries seem to be "underutilised" in developing countries, though the picture seems to improving in countries such as Brazil, Colombia, and the Ukraine among others. Thus, these OECD experiences can inform any adjustments and reforms of EFC controls in client countries. As stated before, it is clear that significant hurdles in terms of political will and administrative capacity need to be overcome in many LICs and MICs to implement a comprehensive strategy. Therefore, any diagnostic mission needs to work closely with experts and officials from the client country to understand which aspects of the strategy are feasible, suitable, and appropriate for specific national circumstances; and on the basis of this consultation draft short, medium, and longer term recommendations. It is clear that an introduction of every aspect of a comprehensive strategy might not be appropriate for social protection programs in LICs and MICs, which may have limited resources and offer limited coverage. Nonetheless, evidence shows that as the benefit system develops and inevitably increases in size over time, effective 11 NAO (2008), Progress in Tackling Benefit Fraud, (HC 102 Session 2007-2008). 12 NAO (2008), Progress in Tackling Benefit Fraud, (HC 102 Session 2007-2008). 13 anti-fraud controls become essential in safeguarding the integrity of the benefit system. Therefore, an assessment should focus on recommendations in the: ˇ Short-term: These can be quick-wins at little cost, which can improve the integrity and targeting of the social protection program (e.g. information campaigns on what is considered fraud and corruption in the Kyrgyz Republic, Bangladesh and the Ukraine); ˇ Medium-term: These are improvements, which can improve the integrity and targeting of the social protection program but need some resourcing and additional administrative capacity to implement (e.g. the application of existing sanction regimes against offenders in the Kyrgyz Republic); and ˇ Longer-term: These are aspirational improvements, which likely require substantial political and administrative hurdles to be overcome (e.g. introduction of electronic data-matching between government databases in the Ukraine or direct electronic transfer of benefit to the individual claimant in Bangladesh). The missions conducted so far show a difference in how recommendations were arrived at in on the one hand the Ukraine and the Kyrgyz Republic and on the other in Bangladesh (see Chapter 5). In the former countries, they had a strategy around prevention, detection, deterrence and monitoring. Thus, the recommendations could look at the effectiveness of what was being done and also at how the existing strategy could be improved and adjusted. In Bangladesh, the EFC strategy of the Employment Generation Program was mostly limited to basic monitoring. Here, recommendations focused on developing a EFC strategy with some basic steps around prevention, detection, deterrence, and performance management. It is also important to note that a longer-term recommendation in one country might be easier to implement in others. For instance, electronic data-matching would face significant hurdles in Bangladesh but has been implemented in a number of Latin American and Caribbean countries. 14 CHAPTER 3 Tools and instruments used to combat EFC 3.1 What are the main instruments to combat EFC per aspect of the strategy? As stated earlier, a strategy consists of prevention, detection, deterrence and measurement initiatives. These initiatives overlap and are often interrelated. This section focuses in more detail on administrative procedures used in several OECD countries to control EFC, in particular compliance of the program with its own rules and regulations. Prevention consists mainly of four categories of initiatives: ˇ A first focuses on whether EFC issues are considered at the program design and each program implementation stage to ensure that the risk of EFC is manageable in the administration of the program. This speaks to wider considerations of program design and program management, which are important to the World Bank (see Appendix C). ˇ A second aims to ensure that payment of the benefit to the claimant is as secure as possible, preferably by making payment directly to the claimant. ˇ A third is centred on whether appropriate administrative procedures and capacity are in place. This ranges from the provision of adequate IT systems and capacity to the training of staff, the clear delineation of administrative procedures in the processing and verification of benefit claims (e.g. having clear job description [see Appendix F for job descriptions in the UK]) and clarity in the allocation of institutional responsibilities across the different levels of government and between government and civil society). ˇ A fourth looks at the checks undertaken at the outset of the claim. Many OECD countries have tightened the checks at the outset of the claim as fraud tends to be more difficult to identify as the claim is in progress. ˇ Information campaigns are a fifth initiative. Information campaigns aim to alert claimants on the risks of committing benefit fraud (used in Australia and the United Kingdom) and explaining rights and obligations under the rules of the social protection program to benefit claimants (used in Canada and the Netherlands). There exists some innovative `prevention' practice in OECD countries. 13 ˇ Canada holds claimant information sessions (CIS) with groups at high risk of committing benefit fraud. Since 1999, Canada claims it has saved about US$600 million in overpayments in the benefit system. The Netherlands also uses contracts and campaigns to inform claimants of their rights as a preventative measure. Claimants have to sign a contract with their rights and obligations when claiming benefits. Similarly, the New Zealand government aims to inform about 70% of claimants of rights and obligations. These measures aim to inform and remind claimants of their obligations. For 13 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006) 15 error, preventative measures include staff training and introducing results-based management. In Canada, error rates have fallen by making line managers more accountable for the errors of their staff. Detection has three main components: ˇ The first is the generation of intelligence to detect cases of EFC. This generation of intelligence can consist of telephone or online tip-offs from the public about fraudulent cases (used in Australia and the UK), intelligence provided by staff of the social security administration or other government offices, and data-matching. This intelligence then leads to cases being investigated. Telephone hotlines also serve to allow the general public to file grievances and complaints. ˇ A second is the process of reviews of benefit claims. Reviews can be random, time-based, or risk- based. Most OECD countries use risk-based assessments to organise a review of benefit cases, and some use regular, random or time-based reviews of cases (for instance in Australia and New Zealand). The advantage of risk-based reviews is that it allows the benefit administration to target scarce resources at those claims with the assumed highest risk of fraud. Time-based reviews might be helpful in cases where eligibility requirements have a clear time component (e.g. limits in the duration of the benefit provided). ˇ A third initiative is joint working between government departments and agencies and different levels of government (joint working is emphasised in the Netherlands and Australia). Joint working leads to improved information exchange and also shared administrative capacity in cases. There exists some interesting `detection' practice in OECD countries. 14 ˇ Data-matching can be highly cost-effective and target ratios are set in Australia and New Zealand for overpayments identified to cost of the program. A normal ratio would be four to one (benefit payment saved to administrative cost expended). Unique identifiers for claimants are also often part of data-matching strategies. The use of identifiers across different databases and departments allows analysts to more effectively match data on claimants held in different databases. Review activities are also instrumental to detection efforts. Different strategies are used in most countries ranging from risk-based reviews (e.g. reviews of certain professions in Canada, scoring in the UK, and profiling in Canada) to random and time-based reviews. In New Zealand, claimants have to re-establish core eligibility after a certain time period. Deterrence involves sanctions. Several countries have begun to use more severe sanctions and increased levels of prosecutions as a deterrent. Examples are Ireland, the United States, and the United Kingdom. Monitoring can include four main ways of performance assessment in EFC controls: ˇ Measurement of total volume of EFC in the social protection system: This approach will aim to estimate the total volume of EFC in social protection expenditure on the basis of random reviews of cases (used in the UK); ˇ Measurement of volume of EFC in particular social protection programs: This approach is similar to the first but used in specific programs (used in the US and Ireland); ˇ Measurement of accuracy of processing and payments: This approach will focus on the accuracy of processes and will set performance targets for accuracy of payment and processing. Accuracy is 14 NAO (2006), International benchmark of fraud and error in social security administrations, (HC 1387, Session 2005-2006) 16 established through an independent review of cases (either random or spot checks) (used e.g. in New Zealand, Sweden, and Australia); and ˇ Measurement of cost-effectiveness: This approach will seek to establish the return on investment for particular initiatives by comparing the investment in an initiative with the amount of overpayments identified by the initiative. In monitoring, the measurement of EFC in the social protection system on the basis of random statistically significant sampling is increasingly seen as the gold standard as it gives a clear indication of the total problem of EFC in programs and allows for consistent monitoring over time (see Appendix D). 17 Table 1: Main instruments per aspect of the strategy Outcome Actions taken Result Remark Prevention Fraud-proof program design Considering EFC control upfront in the program design Considerations for controlling error might be different than those for fraud and and operational aspects of stage when setting out the eligibility and administrative corruption ­ for instance error might call for simpler administration while fraud program procedures reduces the risk of EFC to the program and corruption might ask for more administrative steps around verification ­ considering risks and such trade-offs upfront is important ­ this speaks to achieving a balance between placing a proportionate burden on the administration and quality assurance Make payments directly to Making payments more directly to applicants cuts out It reduces the chance of corruption as few officials are handling the money or are bank accounts instead of by important opportunities for EFC to occur involved in transferring the funds ­ it also reduces the chance of error as the check or cash payment process tends to be simple IT systems Adequate IT systems allow for the easy cross-checking of Investments in IT are seen in most OECD countries as critically important in data enabling staff to find fraud at the outset of a claim raising the effectiveness of prevention and detection initiatives. (requires unique identifier number) IT systems also reduce human error in processing Clear allocation of Division of administrative responsibilities (monitoring, There can be a trade-off between separating responsibilities and reducing institutional responsibilities payment and processing of benefit) is good practice to administrative complexity ­ this speaks to considerations around institutional reduce the risk of EFC in most OECD countries design Setting up proportionate A complex or multi-tiered administration might increase administrative procedures the risk and opportunity of EFC occurring Guidance Most staff in OECD countries operate according to clear Clear guidance gives clarity on the responsibilities of staff as well job descriptions and standard operating procedures to minimise error Staff training Ensuring that staff is trained to an appropriate standard is In Canada, staff training was seen one key components of reducing error in important in reducing error benefit administration Resources Ensuring staff have appropriate resources and are Proper resourcing in some of the missions would enable staff to their work better reimbursed for travel expenses can reduce error and indeed (e.g. detection), but also reduced the incentive to misappropriate corruption Checks at outset of claim Fraud, error, and corruption are harder and more costly to The UK administration is focusing more resources in the current economic crisis detect during the claim compared to identifying it at the to prevent fraud from coming in the system outset of the claim Information campaigns Making the public aware of the rules and obligations of the Social control, transparency, and accountability are critical elements in any social program and the issue of EFC in social protection can protection program prevent EFC occurring 18 Detection Data matching Comparing data held across a variety data sources is highly Typically, a strategy will have a variety of ways to create referrals for further cost-effective at detecting cases of suspected fraud and investigation including data-matching and wider intelligence gathering corruption Intelligence gathering and the introduction of a complaints mechanism can also Intelligence Gathering information from the public (tip-offs), telephone strengthen the accountability within a program hotline for reporting fraud, automatic flags from Management Information System (MIS), and staff intelligence are critical in producing referrals for further referral Investigation Random or time-based reviews (controls of the accuracy of Some social security agencies solely investigate referrals ­ others supplement these the information provided to the program and of the reviews by random or time based reviews decisions made by staff) are necessary aspects in detecting Some social security agencies target their scarce administrative resources by EFC profiling (professions or population groups) or by scoring the referrals that they Risk profiling Risk-based assessments to organise reviews (e.g. select client receive - this is seen as good practice 15 investigations) using scoring or profiling are particularly cost-effective in detecting fraud Joint operations with other Information exchange and joint investigations are useful in Joint working remains a challenge even for well-developed administrations, but agencies detecting and investigating organised crime and high-level cases in Australia where information exchange occurs show positive results fraud and corruption Deterrence Sanctions Applying appropriate sanctions consistently serves as an Deterrence is an important complement to prevention ­ enforcement requires important deterrent for those considering engaging in fraud clear definitions on what EFC are, a supportive legislative framework , and Prosecutions and corruption appropriate sanctions for each possible case of EFC Confiscation of assets Measures to influence public See similar point earlier on prevention attitudes, such as public awareness campaigns. Monitoring Regular measurement of the Understanding the overall problem of EFC and particular Globally, social security administrations are measuring more rather than less ­this level of EFC risks to the program is of critical importance in organising speaks to designing the appropriate management information system reviews of cases and understanding the effectiveness of the entire strategy 15 There is no commonality between countries on how and how often reviews are carried out. Some are mandatory when applying or re-applying for a benefit. Others are random on a sample of the population. Others are solely risk- based on the basis of professions or scoring of fraud and error referrals. Most countries use a mix of these approaches. 19 3.2 Are there specific instruments for tackling EFC respectively? Most OECD countries focus on tackling benefit fraud. Indeed, most of the interventions outlined in section 3.1 are applicable to reducing fraud. Nonetheless, error is a significant problem. In the UK, between 40 and 50% of overall fraud and error in benefit program is error. Some interventions in OECD countries have been shown to have positive effect on error. They include: ˇ The introduction of results-based management and improving the level of management oversight in Canada; ˇ Improving guidance and communication with benefit applicants in the United Kingdom (errors on application forms can also arise from a lack of understanding); ˇ Staff training and development in most OECD countries (many errors are caused by staff processing incorrectly); and ˇ IT improvements (many errors are systemic failures) in Ireland and the United Kingdom. Corruption is not considered an important problem in the OECD countries. There are a number of reasons for the low prevalence of corruption in OECD social protection programs: ˇ Clear and transparent eligibility criteria; ˇ Separation between assessment and payment functions; ˇ Separation between processing/decision-making and investigations or reviews functions; ˇ Integrity of systems processing payments; ˇ Staff training and management; ˇ Investigators/reviewers not assigned cases where familiarity is suspected; and ˇ Strength and independence of internal audit. 3.3 How do OECD instruments aimed at mitigating EFC develop over time? The instruments applied in selected OECD countries to tackle the problem of EFC in social protection programs developed over time. The way in which these systems generally develop can hold lessons for introducing these types of instruments in less developed countries. This brief section does not aim to provide a definitive overview of how anti-fraud strategies develop across OECD countries, but provide a general indication on how these types of strategies evolve and mature on the basis of observations of developments in OECD countries and lessons learnt in the course of the fieldwork for this toolkit. An overview is given in Table 2. This table is based on a general interpretation of maturation by the authors 20 and outlines how typically some areas of prevention, detection, deterrence, and monitoring evolve in the short-term, medium-term and longer-term. 16 The maturation of strategies is based on three dimensions: ˇ Increased investment in IT and administrative capacity (Some OECD countries [e.g the UK and Australia] show that the public resources saved through EFC reduction are far more substantial than the investments made in EFC mitigation); ˇ Improvements in technology (the application of new technologies in terms of communicating with the public, processing of benefits, making payments to beneficiaries, and matching data held on individuals change the general administration of benefits and how EFC mitigation is approached); and ˇ Learning over time from your actions (a good example is moving to more risk-based reviews as the agency learns what the EFC risks to its operations are through analysis and intelligence gathering). Table 2: A view on the development of EFC strategies in OECD countries over time Outcome Type of Short-term Medium-term Longer-term intervention Prevention Secure -Lower transaction costs in -Pay directly to beneficiary -Pay directly to beneficiary payment payment (e.g. reduce through secure channel (e.g. number of administrators bank account) involved) Improve -Introduce clear guidance -Improve staff training to -Introduce results-based administrative on administrative ensure awareness of EFC and management capacity responsibilities and job how to tackle it -Upgrade IT systems to allow descriptions -Move to IT systems to register case history to be seen and -Provide adequate and process claimants eligibility checks to occur administrative resources (e.g. cover expenses ) Information -Make claimants and staff -Introduce a rights and -Introduce information sessions campaigns aware of what EFC is and obligations charter between to inform claimants on rights eligibility requirements of claimant and social security and responsibilities program agency -Work with civil society groups to educate wider groups of benefit claimants Detection Intelligence -Empower staff and -Formalise procedures on how -Invest in IT to perform data- beneficiaries to refer cases to refer cases for investigation matching on electronic databases of EFC for investigation -Introduce telephone hotlines -Invest in analysis of evidence -Undertake paper-based for individuals to complain and collected to inform investigations data matching refer cases -Improve joint-working -Improve data-matching to be with other government 16 See for instance some discussion on fraud regimes in van Stolk, C. and Wegrich, K. (2010) "Casting light into dark corners. A cross-national comparison of approaches to performance management in fighting fraud in social security and the hidden economy", Public Management Review, forthcoming. 21 agencies real-time Investigations -Introduce thorough and -Introduce time-based -Perform risk-based independent reviews of investigations investigations on the basis of cases that look at all facets cases most likely to see EFC -Introduce spot checks at the of a case (analysis of intelligence) outset of a claim -Introduce and apply -Introduce a dedicated guidance on how to investigations service conduct investigations Deterrence Sanctions -Introduce sanctions to -Apply sanctions consistently in -Relate severity of sanctions to deter fraud and corruption all cases EFC risks faced by social protection program Monitoring -Sample selection of cases -Sample statistically significant -Sample statistically significant and monitor outcomes of sample of cases in some number of cases to gauge level of independent reviews programs to gauge level of EFC EFC in all programs -Monitor outcomes on the -Measure cost-effectiveness of basis of categories of EFC interventions used to combat EFC -Measure accuracy of processing and payments 3.4 How are the OECD instruments to mitigate EFC currently applied in less developed countries? In less developed countries, these types of administrative initiatives to control EFC seem underutilised. 17 For instance, pilot mission to the Ukraine and the Kyrgyz Republic found a very limited use of deterrence, with administrations quite reluctant or unable to apply sanctions to fraudulent claims. In the pilot mission in Bangladesh, EFC controls did not extend much beyond basic monitoring activities. Evidence from South Asia also shows important limitations in control and accountability mechanisms in conditional cash transfer programs. 18 However, the World Bank has also found some positive trends emerging. These include greater use of technology, at least in MICs; more explicit thought on transparency; and greater attention to implementation issues (see Box 3 for examples). Better practices in collecting and using program information in developing countries can provide an indication of the scope and nature of the problem, and may even prevent EFC from happening. A number of MICs are making large investment in management information system of the social protection program, a key tool for the development of an effective policy to combat EFC. Table 3 gives an overview of some good practice in MICs. 17 World Bank 2007 FEC SSN Primer note. 18 See for instance, World Bank (2009), "Control and Accountability Mechanisms in Cash Transfer and Conditional Cash Transfer Programs: A Review of Programs in South Asia". 22 Table 3: Good practice of oversight and control mechanisms from the Latin America region 19 A number of middle-income countries have made significant steps forward in implementing mechanisms to control the use of funds in social protection programs. Some of these initiatives are more comprehensive than others and include: ˇ Brazil's Bolsa Familia Program (BFP) is a large conditional cash transfer (CCT) program implemented in a federal context. At federal level, the program is managed by the Ministry of Social Development (MDS). Most of its implementation, however, is done by the 5,564 municipalities. Consequently, the main pressure points for EFC in Brazil's decentralized context include political interference, political biases, and administrative errors in the registration process. To manage, monitor and reduce the risk of EFT, the Government has instituted several mechanisms, including: (a) Oversight & Control by Supreme Audit Agencies that undertakes regular random-sample operational audits (also known as quality control reviews); implementation evaluations; annual financial audits, and investigation of at-risk cases; (b) MDS' own program controls, including internal and external cross-checks of the registry; monitoring municipal implementation quality using the new system of financial incentives (performance-based administrative cost subsidies) and the "Decentralized Management Index", and; monitoring of the activities of the operating agent for registry and payments, via the new performance-based contract and associated financial penalties, and; (c) Municipal controls (social controls, local contacts). Municipalities serve as first point of contact for complaints or appeals for the beneficiaries. Formal Social Controls Councils have been established in all municipalities, and their roles include: overseeing the registration process; periodically evaluating beneficiary list; monitoring reporting on, and compliance with, conditionalities. Centralized hotlines are used to handle inquiries from general public, beneficiaries, and local program managers. ˇ The Heads of Household Program in Argentina is a large-scale emergency workfare program set up quickly during 2001-2 economic crisis. The Ministry of Labor is the responsible national agency and registration is decentralized through municipalities and civil society/political organizations. A set of complaint resolution tools has been established, including: (a) Toll free hotlines manned by call centres addressing questions on payment dates, eligibility and for reporting ineligible beneficiaries; (b) a Commission in the Ministry of Labor to handle allegations of program abuse or complaints; (c) Criminal offenses referred to Federal Prosecutor of the Social Security System; (d) monthly cross-checking of databases. ˇ Colombia's Familias en Accion is a CCT program under Presidential Executive Agency (Social Action) implemented at municipal level. Spot checks reviewing and monitoring the operational procedures for program implementation in different localities are carried out on a random basis by external firm. These spot checks serve several objectives, such as: checking if operations and procedures of program met objectives in practice; verifying how procedures were being interpreted; validating the accuracy of the data relevant for the program (i.e. verification of fulfilment of conditionalities, inscription of eligible families, payments, etc.). Source: Lindert et al (2007) and World Bank (2007) 19 Taken from World Bank 2007 FEC SSN Primer note. 23 3.5 What are the main instruments to combat EFC per level of government? The administration of social protection programs often involves different government actors, levels of government, and cooperation between government and civil society. This is especially true in lesser developed countries where administrative capacity is more limited than in the developed world. This section highlights where some of the responsibilities could lie in lesser developed countries. Gives an overview of responsibilities split between independent audit agencies, the ministry responsible, and local level actors. This split of responsibilities is particularly important in LICs and MICs as policy makers in these countries often involve multiple levels of government to ensure that the program can be delivered. In many OECD countries the administration of benefits ends to be more centralised or is delivered through executive agencies under the direct supervision of the department responsible. Figure 4: EFC oversight mechanisms per level of government ˇFinancial audits ˇRegular random-sample operational audits Independent Audit ˇCase investigations & follow up Agencies ˇImplementation evaluations (GAO type) ˇPublication of beneficiary lists on transparency portal ˇInternal & external registry cross-checks ˇHotlines with trained operators Ministry Responsible ˇCase investigations & follow up for Programme ˇMonitoring mechanisms for third party implementation (e.g., municipal governments, payments agent, NGOs) ˇPublication of beneficiary lists on transparency portal ˇRegular recertification of eligibility ˇLocal point of contact for complaints, appeals Local Level ˇFollow up investigations (case-by-case) ˇ"Social controls" councils (citizen oversight) Source: Fiszbein, Tesliuc, and Lindert, 2009, World Bank 3.6 How does this approach on EFC control inform other diagnostic initiatives? It is important to note that the approach taken in this toolkit is broadly compatible or complementary with other diagnostic initiatives undertaken by the World Bank or in countries. A value added of this toolkit is that it provides a structured way of thinking in terms of prevention, detection, deterrence and monitoring of EFC in social protection programs on the basis of the experience of OECD countries, which can inform other, perhaps broader, diagnostic initiatives in the World Bank and contribute to what the World Bank emphasises in overall program design, as exemplified in some recent reports on South Asia and the Latin America Caribbean region. 20 These reports focused strongly on effective control and accountability mechanisms to ensure that effective targeting occurs, beneficiaries receive payment, operating procedures in programs are followed, good governance and accountability are present, and 20 World Bank (2007), Control and Accountability Mechanisms in Conditional Cash Transfer Programs: A Review of Programs in Latin America and the Caribbean. Operational Innovations in Latin America and the Caribbean, Vol. 1, No. 1; World Bank (2009), "Control and Accountability Mechanisms in Cash Transfer and Conditional Cash Transfer Programs: A Review of Programs in South Asia". 24 policymakers have the information to improve the program. The EFC toolkit can either serve as an additional checklist to see if recommendations from reports follow conventional wisdom seen in OECD countries (see Table 1) or to design a diagnostic mission with a more exclusive focus on EFC. It is useful to give some concrete examples. An example of the latter on how approaches can be merged is given in Table 4 taken from a proposed diagnostic mission in a cash conditional transfer program (CCT). Similarly, Table 5 in Appendix C gives an overview of how to look at program design with EFC lenses. 25 Table 4: Assessing control and accountability mechanisms in a CCT 21 The World Bank proposes a diagnostic mission to look at good governance and EFC control of a CCT in Southeast Asia. This mission merges the approach in this toolkit with the diagnostic approach taken to look at governance in cash conditional transfer programs in the Latin America Caribbean region. The approach taken in CCT in the Latin America Caribbean region highlighted the following areas for attention: ˇ Institutional design. Define lines of authority and accountability, including administration and service delivery. Align incentives and financing, particularly for intergovernmental responsibilities. Identify risks and remedies up front. ˇ Targeting. Focus more attention on possible exclusion errors (that is, eligible households not being included) and consider the need for more frequent recertification of eligibility. ˇ Conditionalities. Balance accuracy and the burden of verification. Examine reasons for non-compliance. ˇ Payment processes. Improve cash-flow monitoring and associated control procedures. ˇ Management information systems. Strengthen frameworks to analyze and design MIS systems. Encourage crosschecking of enrolments with other databases. ˇ Internal quality control. Expand process evaluation, feedback loops, and correction mechanisms within programs, particularly through spot checks. ˇ Accountability. Ensure appropriate mechanisms to handle complaints and appeals in a timely manner, and publish program data. ˇ Financial management and auditing. Define payment accounting and reporting requirements. Focus audits on risks. Include supreme audit institutions. ˇ Procurement. Closely supervise the implementation of large service contracts. Merging this approach with the approach developed in this toolkit gives rise to five main areas of interest developed by the World Bank focusing on EFC mitigation to strengthen governance and EFC control in this CCT program. ˇ Minimising risks of corruption and resource leakage in targeting (speaking to prevention); ˇ Minimising corruption risks in program implementation (speaking to prevention); ˇ Design of design of management information system for ongoing control of program integrity (speaking to monitoring); ˇ Development of program-level control and accountability measures to ensure (speaking to detection); and ˇ Enhancement of department-wide control and accountability measures to address more systemic EFC issues (speaking to detection). 21 World Bank (March 2007), Control and Accountability Mechanisms Conditional Cash Transfer Programs: A Review of Programs in Latin America and the Caribbean 26 CHAPTER 4 Terms of references for a diagnostic assessment in client countries 4.1 What are the key questions to be asked? Appendix A provides an overview of a template used in previous missions. Some of the key questions are: ˇ What is the administrative capacity for EFC control and who is responsible for EFC control? ˇ What are the different types of social protection programs and who is involved at each in their implementation? ˇ How are EFC defined? ˇ What are the causes of EFC identified in the social protection program? ˇ What is the extent of the EFC problem in the social protection program (s)? ˇ What does the administration do in terms of prevention, detection, and deterrence initiatives? ˇ Does the administration evaluate the effectiveness of initiatives? ˇ How (if at all) are EFC measured and monitored in social protection programs? ˇ How is monitoring information used by policy makers? 4.2 What type of data collection takes place? Most of the data is likely to be provided by the ministry responsible for the social protection program. This data is likely to include: ˇ Data on social protection programs including number of claimants over time, average claim, expenditure per region, total value of expenditure and so on. ˇ Data on staff in the ministry and local levels (number per unit and salary). ˇ Data on reviews or audit activities, including number of audits, audits per region, outcome per audit (e.g. overpayments identified). 27 ˇ Data if available on the overall extent of EFC in the social protection program, including the amount of overpayments identified, overpayments per region, and average overpayments (ideally also overpayments identified as fraud, error, and corruption). ˇ Data on targeting. ˇ Data on prosecutions or sanction, including the number of sanctions issues and the average value of the sanction. 4.3 Who are the key informants in this diagnosis? Informants in previous mission have included: ˇ Head of internal audit of ministry responsible; ˇ Heads of unit dealing with the processing and payment of benefit claims; ˇ Staff in the ministry involved in audit; ˇ Staff in the ministry involved in processing and payment of benefit claims; ˇ Staff at district and local levels involved in audit; ˇ Staff at district and local levels involved in processing and payment of benefit claims; ˇ Beneficiaries; and ˇ Local experts. 4.4 How are recommendations arrived at? Recommendations arise from three questions: ˇ What is the ideal EFC control (based on OECD practice)? ˇ What type of EFC mitigation is in existence in the social protection program(s) of the client country (based on data and information collected in the mission)? ˇ What can be done to improve EFC mitigation in client countries in the short, medium and long term given the administrative context (based on the judgment of the mission team)? 4.5 What should the composition of the team be? Previous missions have had the following composition (see Appendix C for a typical work program): ˇ A senior official with knowledge of social protection programs in LICs and MICs as mission lead in charge of organising the mission; 28 ˇ A senior official with knowledge of social protection programs in the client country (a country officer or SP country lead)in charge of local facilitation of the mission; ˇ One or two consultant(s) with in depth knowledge of social protection programs of the client country. This consultant will prepare the mission by undertaking background research on the basis of the template in Appendix A. This consultant should have close contacts with the ministry responsible for administering social protection programs and a background in looking at governance, accounting, audit, or social protection. ˇ One or two consultants with in-depth knowledge of good practice in OECD countries. These have in past included heads of fraud and error strategy teams in OECD countries, senior members of fraud investigations teams in OECD countries, and an independent researcher with in-depth knowledge of EFC controls in OECD countries. Their contribution is to compare what client countries are doing to good practice in OECD countries and bring their experience of having developed EFC control over time. In the past, they have made presentations to EFC control practitioners in the client country on good practice in OECD countries and have written the diagnostic report. Appendix C gives an overview of their responsibilities during the mission. These responsibilities could also be taken on by expert World Bank staff in future missions. ˇ A translator if required. ˇ Support staff to organise logistics and scheduling. 4.6 What are the costs of the diagnostic mission? The cost of the diagnostic missions in Bangladesh, Kirgizstan and Ukraine ranged between $30,000 to $50,000. 29 CHAPTER 5 Examples from missions so far In this Chapter, we give examples from missions undertaken during 2008-2009 in the Ukraine, the Kyrgyz Republic and Bangladesh; as well as from the review of governance and accountability arrangements in CCT program in Latin America and the Caribbean 22. The missions offered a diagnosis of the issue of EFC. The diagnosis included: ˇ What is the problem of EFC in the social protection program (s) of the client country and how is the problem monitored? ˇ What type of EFC mitigation is in existence in the social protection program(s) of the client country (looking at prevention, detection, and deterrence)? ˇ What can be done to improve EFC mitigation in client countries in the short, medium and long term given the administrative context? (this is derived from the judgment of the mission team on the basis of what we know of good practice in OECD countries). On the basis of the diagnosis, a dialogue was opened with key policy makers to see how improvements in EFC mitigation could be made. These improvements can be considered wider improvements in program implementation. Below, we give an overview of the findings in the Ukraine, Kyrgyzstan and Bangladesh. 5.1 Ukraine 5.1.1 What is the problem of EFC? The Ukraine does not systematically measure or estimate the overall extent of fraud and error in the system. Rather, it bases its assessment of the level of fraud and error in its social programs on the results of the inspections undertaken by its social inspectors in specific benefit types. The Ukraine does not monitor the overall rate of fraud and error. The Ministry of Labour and Social Policy requires the local offices in the rayons to report to the regions or oblasts, which send their reports to the Ministry of Labour and Social Policy and reports on the effectiveness of its initiatives in different ways. 5.1.2 What type of EFC mitigation is in existence? Prevention Most of the emphasis in prevention of benefit fraud and error in the Ukraine is placed on extensive controls and document checks at the time of the benefit applications. In many cases, the application is filled out jointly by the benefit applicant and a member of staff in the benefit office. Each application has to go through a series of document checks. 22 World Bank (March 2007), Control and Accountability Mechanisms Conditional Cash Transfer Programs: A Review of Programs in Latin America and the Caribbean 31 Moreover, there are two ways in which the rights and obligations of the benefit claimant are explained. Firstly, the application admission officer informs an applicant on his rights, the obligation to submit correct information and on the right of social inspector to check his/her material circumstances. Secondly, regulations require the benefit authority to notify each claimant that his/her signature on a submitted application is an acknowledgement that all information in the document is complete and accurate. Detection Roles of the social inspector The main detection activities in the Ukraine centre on the work of the social inspectors. The social inspectors have three main roles. Firstly, they perform verification checks (home visits) and subsequent checks on changes in material circumstances for a number of means-tested benefits. Secondly, the social inspector can undertake random checks of benefit recipients. Finally, social inspectors also do information and awareness work. Effectiveness of social inspections The main observation on the effectiveness of social inspectors is that the number of detections per number of inspections is low. Though the rate of denial of applicants on the basis of home visits is around 15.8 percent of cases inspected and about 3.5 percent of total benefit applicants, the number of subsequent detections through random checks is much lower between 1.2 and 2.3 percent of total inspections. Deterrence The sanctions used in the Ukraine against those who obtain benefits fraudulently are limited. In all cases of overpayments identified, the benefit offices will seek to recover the overpayments identified through the courts without an additional penalty or fine. There are some other factors, which have an impact on deterrence. Firstly, the absence of clear legal definitions on fraud in the social security system as stated gives a limited legal base from which to apply sanctions. Secondly, the benefit offices have traditionally taken little action against employers who have colluded with benefit claimants. 5.1.3 What can be done to improve EFC mitigation? Prevention The Ministry of Labour and Social Policy needs to play a more active role in changing the general population's attitude towards benefit fraud. Short-term A quick win for benefit authorities is to spell out clearly and visibly on application forms or on the website that benefit fraud is a crime. Medium to long term The Ukraine could start developing a mass media campaign (radio, television, posters, and advertising in printed media) on the topic of benefit fraud. The Ukraine could also pilot the introduction of a telephone 'hotline' or website, where citizens can report instances of suspected fraud. 32 Detection The overall rate of social inspections is high, but the results in terms of detections are low There are three ways in which inspections can be more targeted. Firstly, they can be done on the basis of a risk assessment, which is based on building up profiles of clients who are at risk of committing benefit fraud. Secondly, they can be done on the basis of the hybrid means-test proposed by the World Bank. Those who do not meet these new requirements of means-testing or whose personal characteristics do not match up with the means-test could be inspected. Finally, inspections could occur on the basis of systematic intelligence gathered through data-matching. Data-matching is already in use in the Ukraine to inform random inspections. Its use could be expanded and made available through information technology systems. The social inspectors appear to have inadequate guidance on asset-testing, which leads to subjective assessments of claimants' assets Short term The Ministry of Labour and Social Policy could update guidance and develop a better and more comprehensive methodology for asset-testing in means-tested benefits. The absence of incentives affects the effectiveness of social inspections Short to medium term We would suggest exploring ways in order to steer the performance of social inspectors through process targets (e.g. number of inspections) and outcome of social inspections (e.g. overpayments identified). Deterrence The absence of a legal definition of benefit fraud does not contribute to effective legal sanctioning of fraud The sanctions in place appear insufficient in terms of providing an appropriate deterrent effect Short term There are a number of laws on the books in the Ukraine, which could be used to sanction benefit fraud. Medium term The legal definition of benefit fraud needs to be strengthened and aligned with internal policy at the Ministry of Labour and Social Policy. 5.1.4 What has been the impact of the diagnosis? The impact of the diagnosis has been positive. The World Bank has been working with the Ukrainian government to implement a number of recommendations. The Ministry of Labour and Social Policy has in principle agreed to start piloting the use of the hybrid means-test in its social protection programs in order to limit the number of ineligible claims. 33 5.2 The Kyrgyz Republic 5.2.1 What is the problem of EFC? The Kyrgyz Republic does not systematically measure or estimate the overall extent of fraud and error in the system. Rather, it bases its assessment of the level of fraud and error in its social programs on the results of the inspections undertaken by auditors in specific benefit types. 23 The Kyrgyz government does not explicitly track the effectiveness of its anti-fraud strategy by monitoring the amount of spending on social inspections against what it identifies in overpayments and ultimately recovers. However, the Ministry of Labour and Social Development (MLSD) collects information and keeps records based on results of inspections of organisations within the social assistance system carried out by the Internal Control and Audit Department. The data is submitted on a quarterly basis by rayons. 5.2.2 What type of EFC mitigation is in existence? Prevention The majority of initiatives aimed at preventing benefit fraud take place at the point of submission of the benefit claim where verification of documents and material circumstances takes place. In addition, the MLSD operates mechanism for the public to complain about potential irregularities in benefit applications and payments. Detection There are two main ways that the eligibility of benefits is verified. Firstly, most of the verification work takes place at the time of submission of the benefit claim. Secondly, subsequent detection of fraud and ineligibility takes place at two levels. At rayon level, the eligibility of applicants is verified on average once a year. These controls can detect ineligible applicants. At national level the Internal Audit and Control Department of the MLSD tries to inspect the files of each rayon once a year to detect financial infringements. Deterrence MLSD does not classify financial infringements as fraud, error and corruption. This means the audit practices of the MLSD and wider sanctions are not specifically tailored towards concepts such as fraud, error and corruption. Nonetheless, the Law of the Kyrgyz Republic `On the state benefits' contains a number of articles stipulating specific measures and sanctions in relation to persons who deliberately obtain benefits on false grounds. 23 When looking at overall fraud and error, we can calculate a proxy of the fraud and error rate for UMB, the main unemployment benefit, on the basis of financial infringements. We know that there were 451,300 recipients in UMB with an average benefit of 124 KGS per month. This amounts to about 56 million KGS a month or 672 million KGS per year. If we look at revealed financial infringements for 2007, we find a total value of about 340,000 KGS (see Figure 18) or 2.2 million KGS in 2006. This would mean if we assume that most infringements are in UMB that the amount of fraud that the MLSD identifies a fraction of a percent of total payments in 2007 and a third of percent in 2006. 34 5.2.3 What can be done to improve EFC mitigation? Prevention The Ministry of Labour and Social Development (MLSD) needs to play a more active role in changing the general population's attitude towards benefit fraud. Short term A quick win for benefit authorities is to spell out clearly and visibly on application forms or on the website that benefit fraud is a crime. Medium to long term The Kyrgyz Republic could start developing a mass media campaign (radio, television, posters, and advertising in printed media) on the topic of benefit fraud which highlights the consequences of benefit fraud and the damage it is doing to general society. The MLSD needs to provide better guidance to the social worker in assessing benefit applications Short term The MLSD needs to provide clear guidance to social workers on their roles and responsibilities (e.g. in job descriptions [see Appendix F on job descriptions in the UK]) and provide them with a standard operating manual on how to support benefit applicants and assess eligibility of applications. Medium term The MLSD could take some action on the problem of collusion between social workers and claimants. The MLSD could work with the post office to improve the way payments are handled by the post office Medium term The MLSD could work with the post office to improve how cash is handled. We would suggest a careful review of the delivery of cash to recipients and perhaps implement collection at the Post Office by claimants on production of `verified' identity. Detection The overall rate of social inspections is relatively high, but the results are low in terms of the value of infringements identified and savings made after home and employer visits There are three ways in which inspections can be more targeted. Firstly, they can be done on the basis of a risk assessment, which is based on building up profiles of clients who are at risk of committing benefit fraud. 24 Secondly, they can be done on the basis of the hybrid means-test. 25 Those who do not meet these new requirements of means-testing or whose personal characteristics do not match up with the means-test 24 In drafting the recommendations, one has to keep in mind that in specific instances inspections are a legal requirement and any change in the direction of inspections would require a change in the law. 25 A combination of income/asset testing and proxy measures. 35 could be inspected. Finally, inspections could occur on the basis of systematic intelligence gathered through data-matching. The MLSD could increase its capacity to perform inspections The lack of capacity within the MLSD to inspect benefit claimants and to analyse the reasons for fraud and main risks to the benefit system was often cited. This lack of capacity could compromise the effectiveness of anti-fraud controls. The inspectors appear to have inadequate guidance on performing inspections which can lead to variable working practice Short term The MLSD could update guidance in order to mitigate the risk that inspectors rely too much on personal judgments to inspect cases and detect fraud, error, and corruption. Deterrence The legal definition of benefit fraud does not contribute to effective legal sanctioning of fraud The sanctions in place appear insufficient in terms of providing an appropriate deterrent effect Medium term The MLSD could seek powers to levy administrative penalties and seek the recovery of overpayments outside of the courts in minor cases. This would free up the courts and ensures that there is a greater likelihood for the recovery of overpayments. The legal definition of benefit fraud needs to be strengthened. 5.2.4 What has been the impact of the diagnosis? The MLSD agreed with many of the recommendations in the report. A discussion is to be hold with the international community on how to take some of these recommendations forward. 5.3 Bangladesh 5.3.1 What is the problem of EFC in the Employment Generation Program? Very little is known about error, fraud and corruption (leakage) in the Employment Generation Program (EGP) of the Ministry of Food and Disaster Management (MoFDM) in Bangladesh. Most of information that we found is anecdotal in nature. As discussed earlier, monitoring within the program itself does not provide us with an estimates of error, fraud, and corruption in the EGP, nor does it provide us with systematic outcomes of the inspections undertaken. The monitoring procedures for the EGP are complex. There were broadly three ways of monitoring and reporting in the initial phase of the EGP: a progress report, a daily achievement report, and ad hoc reports. 36 5.3.2 What type of EFC mitigation is in existence? The EFP does not have a strategy to combat EFC in the EGP and few initiatives in terms of prevention, detection, and deterrence outside of the monitoring and reporting function. Our discussions with officials made it appear that MoFDM does not appear to have clear definitions on error, fraud and corruption. Aside from monitoring taking place in the first place, the content of the monitoring also matters. Ideally, monitoring would allow you to understand the causes of leakage within the program in order to adjust or devise strategies to mitigate error, fraud and corruption. Our discussions clearly indicated that MoFDM had insufficient management information to understand the extent of the problem and the root causes of the problem. The MoFDM does not have independent capacity to perform internal audit on the program or fraud investigations. 5.3.3 What can be done to improve EFC mitigation? A number of basic steps could be taken to enhance error, fraud and corruption control in EGP focusing on prevention, detection, deterrence, and performance management. The steps we outline follow the main principles of an anti-fraud strategy practiced in OECD countries. Prevention We see seven main steps in prevention: ˇ Pre-planning to minimise the ad hoc nature of the program and to counteract significant delays in the implementation of the program. ˇ Providing clear guidance on objectives of program, eligibility, and project selection and communicating this guidance to officials on the ground. ˇ Reducing the number of tiers of government involved in administration to minimise the delay in cash transfers to the beneficiary and mitigate the risks of corruption. ˇ Enhancing accountability by separating the functions of administration, payment, and monitoring. ˇ Providing adequate resources (e.g. travel expenses) for those involved in EGP implementation commensurate with the amount undertaken for the program, which could incentivise local officials to take more interest in EGP implementation and reduce the risk of corruption as well. ˇ Using new technologies (mobile phones, mobile ATMs, etc.) to transfer benefit directly to the recipient. ˇ Empowering local beneficiaries to counteract the influence of local officials over the EGP. Detection Current monitoring of the projects and beneficiaries enrolled in the projects appears insufficient. In essence, similar officials administering the projects are also monitoring the projects. We suggest the vetting of the list of beneficiaries in each Union by for instance independent professionals. The vetting of the list of beneficiaries is particularly important given that most opportunity for fraud, error and corruption resides at the selection of beneficiaries stage. Deterrence To enhance accountability in the EGP, a grievance procedure could be introduced. Several avenues to allow for complaints could be introduced. Complaints could be directly made to elected officials at 37 Upazila level. To limit collusion, MoFDM would monitor complaints and ensure that they were actioned at Upazila level. A second avenue, complementary to the first, could be a hotline at MoFDM to allow for complaints. Performance management At present, EGP has limited monitoring to ensure that the funds have been spent according to the rules and regulations of the program. This means that MoFDM does not have sufficient management information to understand the problem of error, fraud and corruption fully and manage the risks of error, fraud and corruption. A number of steps could be taken to improve the information received at MoFDM starting from the bottom-up moving to the Ministry. 5.3.4 What has been the impact of the diagnosis? The MoFDM is keen to explore several aspects of the recommendations with a view of improving overall program implementation. The key aspect is to see whether technical assistance can be provided to improve EFC mitigation in the EPG. In this way, program improvements need to be supported by follow-up work. 5.4 EFC in Conditional Cash Transfer Programs in Latin America and the Caribbean Conditional Cash Transfer (CCT) Programs are being widely adopted the Latin America and the Caribbean region and have been shown to be among the most effective mechanisms available to provide basic human development benefits to lower-income people. The paper "Control and Accountability Mechanisms Conditional Cash Transfer Programs: A Review of Programs in Latin America and the Caribbean" , based on seven case studies, provides an in-depth review of tools and good practices for oversight and control of error and fraud CCT programs. It also provides guidance on the options and factors to be considered selecting among different CCT mechanisms. We highlight here the key mechanisms used to reduce the level of EFC in Colombia and Mexico: Colombia. The CCT Program, Familias en Accion, incorporates several mechanisms to combat error, fraud and corruption. Since the program operates in a decentralized context, there are instruments at various levels. First, at the municipal level, there are established committees who are responsible for addressing appeals and allegations of ineligible beneficiaries participating in the Program. In addition, each municipality signs an agreement with the National Program, taking on specific obligations and responsibilities. If these responsibilities are not carried out or fraud is discovered, then the program can be suspended in those localities.There are established channels in the program (as reflected in the Operations Manual for these committees to take action, together with the required documentation). Second, at the national level, the Program has several mechanisms through which it can receive complaints and allegations, including e-mail, written correspondence, and most recently the installation of a call-center. There are procedures for handling each of these matters and the MIS records the number, location, and type of complaints or appeal. Third, at the national level, the program carried out periodically a system of spot checks to check to see how the Program's procedures are being implemented in different localities. These reviews are contracted externally and include a large randomly drawn ample and cover all aspects of the operation of the Program, including views of beneficiaries. Finally, the most recent initiative is a systems audit, which identified room for improvement in better security for the 38 system and making more procedures automatic. The Program is following up on these recommendations. Mexico. The main systems in place in Mexico to control error, fraud and corruption in Oportunidades: (1) An extensive system for reception and handling of complaints, including different points of reporting, including "mobile boxes" at payment points for beneficiaries. There are established procedures for following up on complaints and the measures taken are reflected in MIS indicators. (2) Since 2000, the Program has implemented twice a year a survey of beneficiaries and program providers called sentinel points, which checks on compliance with procedures. (3) The registration of new families, where families are told about the various mechanisms to report allegations or problems, is covered by a "citizen's report card" part of an initiative of the national government. (4) Extensive data on program, including evaluation, is published and accessible to the public, as are the operating rules. Data on program - including targeting results - available through independent surveys. (5) The Program operates a call- center which receives complaints and allegations. The borader findings of the study are: ˇ In Bank-financed CCT programs in LAC, basic control systems are in place and potential risks being handled adequately (no fiduciary "red flags" uncovered); ˇ There is a variation in performance and practices across countries; ˇ There are island of good practices in different programs in the region: ˇ Operational audits ­ Colombia, Jamaica; ˇ Spot checks ­ Colombia, Argentina; ˇ Role of Supreme Audit Agencies, municipal performance incentives and indicators ­ Brazil; ˇ Database cross checks ­ Argentina; ˇ Evaluations ­ Argentina, Brazil, Colombia, Ecuador, Jamaica. The main lessons of the study point to the need for continuous review of status and nature of risks; and the potential to help our clients on implementing instruments to control the level of EFC and develop strategies. 39 Resources The following provides resources for those interested in looking at `good practice' in OECD countries. 5.5 United Kingdom House of Commons Committee of Public Accounts (2005), "Fraud and Error in Benefit Expenditure", Fourth Report of Session 2005-06, London: NAO. http://www.publications.parliament.uk/pa/cm200506/cmselect/cmpubacc/411/411.pdf 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. http://www.nao.org.uk/publications/nao_reports/05-06/0506592.pdf NAO (2004), "Department for Work and Pensions Resource Accounts 2003-04", Report by the Comptroller and Auditor General, London: NAO. http://www.nao.org.uk/publications/nao_reports/04-05/0405170.pdf 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. http://www.nao.org.uk/publications/nao_reports/02-03/0203393.pdf 5.5.1 Australia Australian National Audit Office (2004), Fraud Control in Australian Government Agencies: Better Practice Guide, Canberra: ANAO. http://www.anao.gov.au/uploads/documents/Fraud_Control_in_Australian_Government_Agencies.pdf Attorney General's Department (2002), Commonwealth Fraud Control Guidelines 2002, Canberra: Ministry for Justice and Customs. http://www.ag.gov.au/www/agd/rwpattach.nsf/VAP/(4341200FE1255EFC59DB7A1770C1D0A5)~com monwealth-fraud+control+guidelines.pdf/$file/commonwealth-fraud+control+guidelines.pdf The Auditor-General (2004), Survey of Fraud Control Arrangement in APS Agencies, Audit Report No.14, Canberra: ANAO. http://www.anao.gov.au/uploads/documents/2003-04_Audit_Report_14.pdf Australian Institute of Health and Welfare (2005), Welfare expenditure Australia 2002-03, Health and Welfare Expenditure Series, No. 24, Canberra: Australian Institute of Health and Welfare. http://www.aihw.gov.au/publications/hwe/wea02-03/wea02-03.pdf Centrelink. (2005). Annual Report 2004/05, Canberra: Centrelink. http://www.centrelink.gov.au/internet/internet.nsf/filestores/ar0405pdf/$file/complete.pdf Commonwealth of Australia (2001). Management of Fraud and Incorrect Payment in Centrelink. http://www.anao.gov.au, Canberra: Australian Government. http://www.anao.gov.au/uploads/documents/2001-02_Audit_Report_26.pdf 41 Department of Family and Community Services (2005). Part C: Department of Family and Community Services Budget Statement. Portfolio Budget Statements 2004-05. http://www.facs.gov.au http://www.facs.gov.au/internet/facsinternet.nsf/VIA/budget2004- 05/$File/05partc_budget_statement.pdf OECD (2004), The Social Expenditure Database, Paris: OECD. http://www.oecd.org/document/2/0,2340,en_2649_34635_31612994_1_1_1_1,00.html 5.5.2 Canada CA MacDonald and Associates (1994), Reducing Fraud and Waste in Income Security Programs in Canada, Edmonton AB: CA MacDonald and Associates. http://www.camacdonald.com/fraserep.htm Canada Newswire, McGuinty Government restoring Integrity to Welfare: Plan Will Help People Move from Working for Welfare to Working for A Living, December 15. http://www.mcss.gov.on.ca/mcss/english/news/releases/041215.htm Canadian Business and Current Affairs (2005), Ontario Undoing Welfare legacy of the Harris Government, Vol. 20, No. 5, p.5, January 24. http://findarticles.com/p/articles/mi_m0LVZ/is_5_20/ai_n13726430 Edmonton Sun (2004), Cheats? What; AISH Facts Don't Support Ralph, p.4, November 13. Fernandes L (2005), Mitigating Fraud in Canadian Social Benefit Programs: Prevention through Risk Assessment, Presentation given by the Investigation and Control Branch of Service Canada to the Six- Nations Conference on Benefit Fraud in New Zealand, September 2005. Fernandes L (2004), Mitigating Risk: A Focus on Prevention at Employment Insurance, Presentation given by the Investigation and Control Branch of Service Canada to the Six-Nations Conference on Benefit Fraud in Canada, September 2004. Gabel T, Clemens J, and LeRoy S (2004) Welfare Reform in Ontario: A Report Card, Vancouver, BC: Fraser Institute. http://www.fraserinstitute.ca/admin/books/files/WelfareReformInOntario.pdf Government of Canada, Public Accounts of Canada 2005: Summary Reports and Financial Statements, Ottowa, ON: Minister of Public Works and Government Services. http://epe.lac-bac.gc.ca/100/201/301/public_accounts_can/2005/v1pa05-e.pdf Human Resources Development Canada (2000), Program Integrity: Grants and Contributions, report by Internal Audit and Risk Management Services, Ottowa, ON: Human Resources Development Canada. http://www.rhdsc.gc.ca/en/cs/sp/sdc/audit/429-98e/429_98_e.pdf Ministry of Community and Social Services Ontario (2003), Eves Government Winning the Fight against Welfare Fraud, Government of Ontario Press Releases, June 11. http://ogov.newswire.ca/ontario/GPOE/2003/06/11/c1873.html?lmatch=&lang=_e.html Social development Canada (2004), Social Security Statistics Canada and Provinces 1978-1979 to 2002- 2003, Ottowa, ON: Social Development Canada. http://www.hrsdc.gc.ca/en/cs/sp/sdc/socpol/tables/page00.shtml 42 Office of the Auditor-General Canada (OAG) (2003), 2003 Report of the Auditor General of Canada, chapter 7 on Human Resources Development Canada, Ottowa, ON: OAG. http://www.oag-bvg.gc.ca/domino/reports.nsf/html/20031107ce.html/$file/20031107ce.pdf Office of the Auditor-General Canada (OAG) (2000), 2000 Report of the Auditor General of Canada, Ottowa, ON: OAG. http://www.oag-bvg.gc.ca/domino/reports.nsf/html/00menu_e.html Office of the Auditor-General Canada (OAG) (1997), 1997 Report of the Auditor General of Canada, chapter 17 on Human Resources Development Canada, Ottowa, ON: OAG. http://www.oag-bvg.gc.ca/domino/reports.nsf/html/ch9717e.html 5.5.3 France Agence pour le Développement et la Coordination des Relations Internationales/Agency for Development and Coordination of International Relations [ADECRI] (2002), The French Social Protection System, Paris: ADECRI. http://adecri.dns2-alteaconseils.info/images/stories/pdf-adecri/the_french_social_protection_system.pdf CNAF (2004), Notre Activité en 2004: Temps Forts et Chiffres Clés. CNAF/Ministčre de la Santé et des Solidarités (2005), Convention d'Objectifs et de Gestion entre l'Etat et la CNAF 2005-2008, Paris: CNAF. http://www.caf.fr/web/WebCnaf.nsf/090ba6646193ccc8c125684f005898f3/27451290f9c9a47ac125710 e005158d0/$FILE/COG%20Int%C3%A9grale.pdf Dubois V, Buton F, Chambolle L, Dulong D (2003), Les Conditions Sociales de la Rigueur Juridique: La Politique de Contrôle et de Lutte contre la Fraude aux Prestations Sociales ā la CNAF et dans les CAF, Dossiers D'Etudes. Allocations Familiales 48. http://www.cnaf.fr/web/WebCnaf.nsf/090ba6646193ccc8c125684f005898f3/78f7691b3d4c2360c1256e 2100378b42/$FILE/DE%2048%20-%20conditions%20sociales%20rigueur%20juridique.doc Mandin C, Palier B (2002), Welfare Reform in France 1985-2002, University of Kent Working Papers July, Canterbury: University of Kent. http://www.kent.ac.uk/wramsoc/workingpapers/firstyearreports/nationalreports/francecountryreport.pdf Mercier M, Raincourt H de (2005), Plus de Droits et plus de Devoirs pour les Bénéficiaires des Minima Sociaux d'Insertion, Rapport présenté au Premier Ministre, Decembre. http://www.politiquessociales.net/IMG/pdf/0000-4.pdf Ministčre de la Santé et des Solidarités/Ministčre délégué ā la Sécurité Sociale, aux Personnes agées, aux Personnes handicappées et ā la Famille (2005), L'implication des Organismes de Sécurité Sociales dans la Lutte contre les Fraudes: Referentiel de L'évaluation, Paris: Ministčre de la Santé et des Solidarités/Ministčre délégué ā la Sécurité Sociale, aux Personnes agées, aux Personnes handicappées et ā la Famille (unpublished working paper). 5.5.4 Ireland O'Brion, Eoin (2005), Control Proofing in System and Process Design, Presentation given by the Director of Regions at the Department of Social and Family Affairs of Ireland to the Six-Nations Conference on Benefit Fraud in New Zealand, September 2005. 43 Committee of Public Accounts (2004), Hearings on Value for Money Report on Evaluation of Control Activity, Parliamentary debates, Thursday, 8 April 2004. http://debates.oireachtas.ie/DDebate.aspx?F=ACC20040408.xml&Node=H2#H2 Comptroller and Auditor-General of Ireland (2005), Annual Report 2004, Dublin: Government of Ireland. http://www.audgen.gov.ie/documents/annualreports/2004/Vol1Eng.pdf Comptroller and Auditor-General of Ireland (2003), Department of Social and Family Affairs: Evaluation of Control Activity, Dublin: Government of Ireland. http://www.audgen.gov.ie/documents/vfmreports/VFM45_Control_Evaluation.pdf Department of Social and Family Affairs (2004a), Statistical Information on Social Welfare Services, Dublin: Department of Social and Family Affairs. http://www.welfare.ie/publications/annstats/04/2004stats.pdf Department of Social and Family Affairs (2004b), Minister Coughlan Welcomes the Comptroller and Auditor-General's Value for money Report on Control Activity-Coughlan to launch Three Value for Money Surveys this Year, Press Releases (February 26), Dublin: Department of Social and Family Affairs. http://www.welfare.ie/press/pr04/pr260204.pdf Department of Social and Family Affairs (2005a), Annual report 2004, Dublin: Department of Social and Family Affairs. http://www.welfare.ie/publications/annreps/04/ar2004.pdf Department of Social and Family Affairs (2005b), 206 Million saved by Anti-Fraud Measures-Minister Reveals, Press Releases (August 18), Dublin: Department of Social and Family Affairs. http://www.welfare.ie/press/pr05/pr180805.pdf Irish Independent (2004), Welfare Fraud crackdown Nets 19,000 in a year but overpayments `still rife', Friday April 9, 2004. http://www.independent.ie/national-news/welfare-fraud-crackdown-nets-19000-in-a-year-but- overpayments-still-rife-175316.html Irish Independent (2004), Ŗ108 million Saved as Inspectors Crack Down on Welfare Fraud, Friday August 4, 2004. http://www.independent.ie/national-news/pound108m-saved-as-inspectors-crack-down-on-welfare-fraud- 369341.html 5.5.5 The Netherlands Algemene Rekenkamer (2004), Fraudebestrijding: Stand van Zaken 2004, The Hague; Tweede Kamer der Staten-Generaal. http://www.rekenkamer.nl/9282000/d/p349_tk29810_2.pdf CMC/T11 Company, Naleving van Sociale Zekerheid: Onderzoek naar Naleving van Regels in de Abw, WW en WAO, rapport in opdracht van het Ministerie van Sociale Zaken en Werkgelegenheid, Amsterdam. http://igitur-archive.library.uu.nl/fss/2007-0206- 200604/29.%20Randomized%20response%20onderzoek%20naar%20regelovertreding.pdf Directoraat-Generaal Rechtshandhaving van het Ministrie van Justitie (2003), Hoofdlijnen Kabinetsbeleid Fraudebestrijding 2003-2007en Terugblik op 2002, The Hague: Directie Handhaving. 44 Interuniversitaire Onderzoeksschool voor Psychometrie en Sociometrie (2000), Managementsamenvatting, rapport in opdracht van het Wetenschappelijk Onderzoek en Documentatiecentrum van het Ministerie van Justitie, Utrecht: Universiteit van Utrecht. Ministerie van Sociale Zaken en Werkgelegenheid (2005), Jaarverslag van het Ministerie van Sociale Zaken en Werkgelegenheid 2004, The Hague: Ministerie van Sociale Zaken en Werkgelegenheid. http://www.minfin.nl/binaries/minfin/assets/pdf/old/xv_sociale_zaken_en_werkgelegenheid.pdf Ministerie van Sociale Zaken en Werkgelegenheid (2004), Integrale Rapportage Handhaving 2004, The Hague: Ministerie van Sociale Zaken en Werkgelegenheid. http://docs.szw.nl/pdf/129/2005/129_2005_3_8044.pdf Ministerie van Sociale Zaken en Werkgelegenheid (2002), Handhavingsprogramma 2003-2006, The Hague: Ministerie van Sociale Zaken en Werkgelegenheid. http://docs.szw.nl/pdf/35/2003/35_2003_3_3288.pdf Ministerie van Financiën (1998), Handleiding ter Voorkoming en Bestrijding van Misbruik en Oneigenlijk Gebruik, The Hague: Ministerie van Financiën. http://www.minfin.nl/nl/actueel/kamerstukken_en_besluiten,1998/03/handleiding_misbruik_en_oneige nlijk_gebruik_xmxox.html OECD (2004), The Social Expenditure Database, Paris: OECD. http://www.oecd.org/document/2/0,2340,en_2649_34635_31612994_1_1_1_1,00.html Social Security Administration (2004), Social Security Programs throughout the World, Europe, Washington DC: SSA. http://www.ssa.gov/policy/docs/progdesc/ssptw/2004-2005/europe/ssptw04euro.pdf TNS-NIPO (2003), Regelovertreding in de Sociale Zekerheid: Verslag van de Tweede Meting van het Periodiek Onderzoek naar Regelovertreding in de Sociale Zekerheidsregelingen WAO, WW en Abw, rapport in opdracht van het Ministerie van Sociale Zaken en Werkgelegenheid, Amsterdam. http://igitur-archive.library.uu.nl/fss/2007-0314-200618/heijden_van_der_03_regelovertreding.pdf van der Heijden PGM, van Gils G, Bouts J and Hox JJ (2000) A Comparison of Randomized Response, Computer-Assisted Self-Interview, and Face-to-Face Direct Questioning: Eliciting Sensitive Information in the Context of Welfare and Unemployment Benefit. Sociological Methods & Research 28(4): pp 505-537. van der Heijden PGM, van Gils G, and Laudy O (2005)., Regelovertreding in de WAO, WW, en ABW/WWB in 2004 Report to the Dutch Parliament, No. 347, September 2005. http://docs.szw.nl/pdf/129/2005/129_2005_3_8047.pdf 5.5.6 New Zealand Comptroller and Auditor-General (2003) "Social Security Benefits: Accuracy of Benefit Administration", Wellington: OAG New Zealand. http://www.oag.govt.nz/2003/benefit-administration/docs/benefit-administration.pdf Ministry of Social Development (2005), Statement of Intent 2005/06, Wellington: MSD. http://www.msd.govt.nz/publications/statement-of-intent/2005/ Ministry of Social Development (2005), Annual Report 2004/05, Wellington: MSD. http://www.msd.govt.nz/documents/publications/msd/annual-report-2005-2006.pdf 45 5.5.7 Sweden Försäkringskassan (2005a), Rapport om misstänkt fusk inom aktivitetsstödet, Dnr 11504/2004: Försäkringskassan. http://www.forsakringskassan.se/press/pressmed/pm2005/pm07_05/dokument/pm07_05bilaga.pdf Försäkringskassan (2005b), The Scope and Financing of Social Insurance in Sweden 2003-2006, Stockholm: Försäkringskassan. http://www.forsakringskassan.se/filer/publikationer/pdf/ana0508-e.pdf Riksförsäkringsverket (2004a), Resultat av kontroll av misstänkt fusk med tillfällig föräldrapenning. Dnr 5601/2004 1, Stockholm: Riksförsäkringsverket. http://forsakringskassan.se/press/pressmed/pm2004/pm37_04/bilagapm37_04/pm37_04rap.pdf Riksförsäkringsverket (2004b), Slutrapport: Fusk i sjukpenningförsäkringen?, Dnr 2004/249, Stockholm: Riksförsäkringsverket. http://www.forsakringskassan.se/press/pressmed/pm2004/pm23_04/bilagapm23_04/Slutrapport_Fusk_i _sjukpenningforsakringen.pdf Riksförsäkringsverket (2005), Socialförsäkringen - Årsredovisning för budgetåret 2004, Stockholm: Riksförsäkringsverket. http://www.forsakringskassan.se/omfk/dokument/ared04_nyinkl_omslag.pdf Riksrevisionsverket (1995), Fusk - systembrister och fusk i välfärdssystemen, RRV 1995:32, Stockholm: Riksrevisionsverket. Riksrevisionsverket (1999), Kontrollen inom välfärdssystem och inkomstbeskattning RRV 1999:35A, Stockholm: Riksrevisionsverket. http://www.riksrevisionen.se/templib/pages/OpenDocument____556.aspx?documentid=3283 Riksrevisionsverket (2002), Välfärdssystemen behöver förenklas! Krångliga regler med många ingrepp gör kontrollen och regeltillämpningen i redan hårt belastade välfärdssystem dyr och svår - En uppföljning inom ramen för regeringens arbete med att stärka kontrollen i staten och minska felen och fusket i välfärdssystem RRV 2002:20, Stockholm: Riksrevisionsverket. http://www.riksrevisionen.se/templib/pages/OpenDocument____556.aspx?documentid=2998 5.5.8 USA GAO (2001) Strategies to Manage Improper Payments: Learning from Public and Private Sector Organisations, GAO Report GAO-02-69G, October 2001, Washington DC: GAO. http://www.gao.gov/new.items/d0269g.pdf United States Department of Agriculture (2005), FY2005 Performance and Accountability Report, Washington DC: USDA. http://www.ocfo.usda.gov/usdarpt/par2005/pdf/par2005.pdf US Department of Housing and Urban Development (2005), FY2005 Performance and Accountability Report, Washington DC: HUD. http://www.hud.gov/offices/cfo/reports/2005par.pdf Social Security Administration (2005), FY2005 Performance and Accountability Report, Washington DC: SSA. http://www.ssa.gov/finance/2005/FY_05_PAR.pdf 46 US Department of Labour (2005), FY2005 Performance and Accountability Report, Washington DC: USDL. http://www.dol.gov/_sec/media/reports/annual2005/annualreport.pdf US Department of Veterans Affairs (2005), FY2005 Performance and Accountability Report, Washington DC: USDVA. http://www.va.gov/budget/report/2005/VA-2005_FULL-WEB.pdf 47 APPENDICES 49 Appendix A: Template for case study The purpose of this template is to outline a set of questions to gather information on how countries are tackling the issue of fraud in their social protection systems. This template can serve as a diagnostic tool to examine the issue of fraud and error in the social protection systems of specific countries and the actions taken by countries to tackle the issue of fraud and error. It can also inform the background research undertaken by country specialists before a World Bank mission. 1. General overview of the social protection system a. What is the structure of the social protection system? This section serves to give a general overview of how the social protection system is administered and which bodies are involved in the administration of the system i. Which government body (ies) is/are responsible for setting policy on social protection and overall coordination of the social protection system? ii. Which government bodies are involved in the administration of the social protection benefits (also take into account that in some cases regional and municipal government assist in the administration of social protection)? iii. What are the responsibilities (in terms of the social protection system) of the main government bodies involved? b. What are the main social protection benefits? This section asks for an overview of the main benefits in the social protection system in a country (pensions, unemployment benefits, other social insurance benefits, active labor market programs and social safety net programs). i. What are the main benefit streams? Could you outline what type of benefits (social insurance; means-tested; universal) they are? ii. What is the expenditure on social protection (over the last 10 years) per benefit stream and in total? iii. What is overall expenditure as a percentage of GDP (over the last 10 years)? iv. What are the eligibility requirements for these benefits? c. What are the main resources within the social protection administration for dealing with EFC In this section, we are interested to know more about the units involved in tackling the issue of fraud in the social protection system. We ask for some detailed budget and staffing information. However, we realise that in some cases this might be difficult to disaggregate. We are also interested to know whether the administration has dedicated resources for tackling fraud. i. Which unit and/or minister is in charge of coordinating the efforts of the administration in tackling EFC? ­ Is there a Secretary of State with specific responsibility for the issue of EFC? 51 ii. Which units are responsible for administering the anti-fraud strategy? 1. What are their staffing levels? 2. What are their budgets? iii. Are there dedicated units within the administration for tackling the issue of EFC (e.g. specific investigation teams and coordination units)? 1. Could you outline these units and their responsibilities? 2. What are their staffing levels? 3. What are their budgets? 2. How are EFC defined? This section asks for the specific definitions used by the social protection administration in the area of fraud, differentiating between fraud and error (customer or official) a. How are EFC defined by the social protection administration? 3. What is the extent of EFC in the social protection system? In this section, we aim to establish whether the social protection administration measures rates of EFC in the social protection system and what these rates of fraud and error are. a. Does the social protection administration measure fraud and error in the social protection system? b. What is the extent of fraud and error in the social protection system? (using national estimates or those produced by the social protection administration) c. Are these estimates broken down further into specific benefit types? ­ What are the rates of fraud and error for specific benefit types? d. What are the trends in the size of fraud and error (last five years)? e. If the social protection administration does not produce any estimates on the extent of fraud and error, what are the reasons for this? (difficulty/ complexity of measurement; resource issues) ­ In the absence of estimates by the social protection administration are there other estimates or qualitative assessments of the extent of fraud and error? f. What are the causes identified by the social protection administration (and perhaps in relevant literature) for the extent of fraud and error? 4. Strategy of tax administrations This section asks where the problem of EFC features in the priorities of the social protection administration and whether the social protection administration has a specific strategy on this issue. a. What is the objective of the social protection administration in dealing with fraud and error in the social protection system? b. What priority does the tax administration give to fraud and error compared with the other risks it is tackling? (e.g. Is tackling fraud and error in the top 10 priorities? Why has it been given this particular priority compared to other risks?) 52 c. What research has the social protection administration carried out into the motivations of those committing fraud and how has it used the results? d. How does the social protection administration assess the risks to the social protection system arising from fraud and error? 5. Key initiatives of the social protection administrations to tackle fraud and error This section asks about the main initiatives used by social protection administrations. Could you describe these initiatives, realising that the list of questions might not be exhaustive and complete. What are the main initiatives used in the social protection administration in the areas of: a. Prevention: Does the social protection administration try to generate awareness among the general population on the issues of fraud and error? Does the administration try to explain the rights and responsibilities of claimants in the social protection system? Does the administration try to encourage people to stop committing fraud through simplifying the benefit system and understanding and influencing the behaviour of those committing fraud? Does the administration try to prevent fraud by informing people about the rules of the benefit system? Does the administration have internal controls in place to stop claimants committing fraud before they receive benefit? What type of systems does the administration use to check for identity? Does staff receive training to assist in preventing fraud? b. Detection: Does the administration use telephone hotlines where claimants can tip off authorities? Does the administration use data-matching to detect cases of fraud? How does the administration investigate cases of fraud? How does the administration assess risks as to identify cases and sectors to focus on? How does the administration generate fraud referrals for investigation? How regularly and systematically does the administration review its benefit claims? How is the claimant database organised and does it have in-built controls to assist in the detection of fraud? Does the administration use a unique identifier number to identify citizens and claimants? Does the administration have performance targets to incentivise staff to tackle the issue of fraud? Does staff receive specific training to assist in the detection of fraud? c. Sanctions/deterrence: What are the sanctions imposed on those committing fraud? d. Other: What other initiatives does the social protection administration use to tackle the issues of fraud and error? e. Are there examples of joint working across the public and private sectors in dealing with the hidden economy? 6. What are the results achieved by the initiatives? Could you give an overview of how the administration assesses results in the following areas? a. Measurement of impact of initiatives used in social protection administration: Does the administration use targets and produce management information relevant to fraud? b. Monitoring and evaluation: How does the social protection administration monitor and evaluate the impact of its initiatives in the area of fraud? c. Cost-effectiveness of initiatives: Do social protection administrations report on cost- effectiveness of its initiatives in the area of fraud? 53 Appendix B: TOR and work program for the World Bank mission in the Kyrgyz Republic TERMS OF REFERENCE I NTERNATIONAL C ONSULTANT (FIRM) Expertise on mechanisms to combat error, fraud and corruption (EFC) in social protection programs Institutional assessment of the mechanisms to control the level of fraud and error in the social assistance (benefit) system in two countries (Ukraine and tentatively, the Russian Federation). Participate in two one-week missions to two client-countries for a fact-finding & diagnostic mission on the mechanisms to combat error, fraud and corruption in social safety net (social assistance) programs. During the mission, you are expected to: (i) lecture during a seminar with the key counterparts, on the OECD experience in combating EFC; (ii) meetings with relevant officials, at central and local (front- office) level; (iii) prepare the outline (PPT) for a policy note with the diagnostic of the situation in the country, and recommendations; (iv) brief the central authorities on the key findings of the mission. Your work will be facilitated by a local consultant that will collect information on the key features of the SP / SSN programs; and the mechanism in place to measure, prevent, detect, and deter error, fraud or corruption. You will take a lead in organizing the work of the local consultants (produce TORs for their work; review the output produced, revise/incorporate the output in a final report). The schedule of your visits will be handled by the local office of the World Bank in the country. After the mission, in three weeks time, finalize a policy note. Deliverable: For each country, a policy note presenting (i) brief description of the programs; (ii) description of the strategies, measures and tools deployed to measure, prevent, and deter EFC; (iii) a critical assessment of the current situation; (iv) a road-map with options to step-up EFC activities. The note will include annexes illustrating the practice in other countries, which in the expert's view can be adapted to the local conditions. Value of the contract: This is a lump sum contract. For the completion of these tasks, the Consultant is expected to provide XX staff days of senior consultant(s), and cover the costs associated with travel (airfare, local taxis, communication expenses and hotel stays. The payment will be made in two tranches: 10% of the value of the contract upon signing, 90% after the receipts of all deliverables, of a quality satisfactory to the World Bank. Example of work program Reducing error & fraud in SP programs, Bishkek, June 15-21 55 The objective of the mission was to foment an exchange of experience between experts and practitioners from OECD countries and from Kyrgyzstan on the strategies, mechanisms and instruments used to reduce error and fraud (E&F), leading to an institutional assessment that will include recommendations, strategic directions to reduce error and E&F in social protection (SP) programs by adapting the mechanisms and instruments used in OECD countries to the conditions prevailing in the Ministry of Labour and Social Development of Kyrgyzstan. X led the mission, which included World Bank staff (X, HDNSP), international consultants (X; X; X) and local consultants (CASE-Kyrgyzstan). The missions worked closely with the regional SP colleagues, X and X. Our work complemented the SP activities in the country, the GFRP Grant in response to the food crisis. The main outputs of the mission were: A one-day seminar on mechanisms to reduce error and fraud in OECD and Kyrgyzstan, June 18. A descriptive report on the mechanisms to reduce error and fraud in Kyrgyzstan, prepared by CASE- Kyrgyzstan. Attached. Initial work on assessing program compliance using the 2005 HBS data. The team had a series of meetings along the service delivery chain, from the front office in charge of eligibility determination to the head office in change with internal audit, reviews or inspections. Wrap up meeting with the Minister of Labour and Social Development and her team of advisors. The team presented the preliminary findings of the mission and announced the next steps (the delivery of a policy note / institutional assessment of EFC mechanisms in few weeks time). The feedback received was very positive: interest in learning about mechanisms and instruments used to reduce EFC in OECD countries; interest in the institutional diagnostic; and interest in potential follow-up in implementing some of the recommendations. 56 Next steps: Institutional assessment. We finalised a first version of the institutional assessment in the week following the mission.. Assessment of the compliance of the UMB (last resort) program. X and X of the World Bank produced a note / assessment of the compliance of the last resort program, the Unified Monthly Benefit, by July 10, 2008. The note is an input into the institutional assessment, and into the on-going country dialogue about the cost-effectiveness of different social protection programs in mitigating the increase in food and fuel prices in Kyrgyzstan. 57 Appendix C: Program design and management with EFC lenses Table 5: EFC lenses on critical phases of the program Phase Recommendation Define lines of authority and accountability, including administration Institutional Design and service delivery. Align incentives and financing, particularly for intergovernmental responsibilities. Identify risks and remedies up front. Eligibility, Recertification Minimise inclusion and exclusion errors. Conditionality Balance accuracy and the burden of verification. Examine reasons for non-compliance. Payment processes Improve monitoring of cash flows and control procedures. Management Information Strengthen framework to analyze and design MIS systems. Promote the Systems (MIS) crosschecking of enrolments with other databases. Internal quality control Expand process evaluation, feed-back loops, and correction mechanisms within programs, particularly through spot checks. Appeals and complaints Handle complaints and appeals appropriately and in a timely manner, and publish program data. Financial management Define payment accounting and reporting requirements. Focus audits and auditing on risks. Include supreme audit institutions. Procurement Closely supervise implementation of large service contracts. Adapted from: World Bank (2007) 59 Appendix D: Approach to measuring fraud and error in the United Kingdom26 Measuring fraud and error across the benefit system in the Department of Work and Pensions (DWP) 1. Three different types of estimate contribute to the total estimates of fraud and error. These are: Continuous measurement: Just over a quarter of all benefit expenditure is covered by the main measurement exercises on Income Support, Jobseeker's Allowance, Pension Credit and Housing Benefit. Official error is measured continuously in Incapacity Benefit and Retirement Pension. There is also an annually updated estimate of `Instrument of Payment' fraud, which captures, for example, the fraudulent encashment of cheques. One-off `snapshot' measurement: Since 1995, the Department has carried out `National Benefit Reviews' (NBRs) of some benefits to estimate the level of fraud and error in a single year. These benefits cover around three fifths of total expenditure. As these NBRs are not carried out annually, we have assumed that the percentage level of fraud and error has remained constant since the last NBR. Assumptions about and modelling of unmeasured quantities: Benefits where there has not been any recent measurement activity make up around 13% of total expenditure. For these, we have based the estimates on the level of fraud and error found in similar, measured benefits. For example, Council Tax Benefit has many similarities to Housing Benefit. 2. The text below outlines the continuous measurement methodology. Further detail on the measurement methodology for Housing benefit and Income Support, Jobseeker's Allowance and Pension Credit respectively can be found at the following links: http://www.dwp.gov.uk/asd/asd2/fem/fem_tech_app_02_07.pdf http://www.dwp.gov.uk/asd/asd2/fem/fem_tec_app.pdf Sampling 3. Cases are randomly sampled on a monthly basis by the Information Directorate from DWP's administrative data sets. The sample is taken in such a way to achieve complete coverage of Great Britain across the year. 26 Provided by the Department of Work and Pensions 61 Review Process 4. For most cases, the review process involves: ˇ Preview: the collation of information from a variety of DWP or Local Authority (LA) systems to build up an initial picture of the case and to identify any discrepancies between information from different sources; ˇ Interview: the interview of the customer at their home, which follows a structured and detailed set of questions about the basis of the claim, aimed at identifying any discrepancies between the customer's current circumstances and the circumstances upon which their benefit claim was based. and if a suspicion of fraud is identified: ˇ Fraud Investigation: carried out by a trained Fraud Investigator with the aim of resolving the suspicion Appendix E: Job descriptions of officials in the UK Customer Service Agent A Customer Service Agent provides the focal point for customers/employers making enquiries to Jobcentre Plus services using the telephone. Depending on the nature of the telephone call received they: ˇ Provide a high quality and effective first contact service for customers either making a claim for Jobcentre Plus and associated benefits, or applying for a National Insurance Number (NINO). ˇ Conduct jobsearches and identify suitable vacancies for customers contacting Jobseeker Direct. ˇ Provide a high quality and effective vacancy taking service, in line with Jobcentre Plus policy, quality standards and employment legislation, for employers placing vacancies with Jobcentre Plus. ˇ Record accurate and relevant information provided by members of the public regarding allegations of potential fraud or tax evasion for possible investigation by the Fraud Investigation Service or HMRC. ˇ Establish the nature of the customer's enquiry, redirecting them if the query is not relevant to a Jobcentre Plus Benefit. ˇ Advise the customer of their rights and responsibilities in relation to their claim. ˇ Identify and initiate referrals to Fraud Investigation Services, with the support of the Team Leader. 62 Benefits Delivery Officer ˇ Complete processing action on claims as required, through to the payment screen, and set HB/CTB indicator as appropriate. ˇ Handle claims where CMS contingency is invoked. ˇ Deal with system generated reports, including appropriate Work Available Reports (WARs) and Overdue Work Reports (OWR). ˇ Liaise with other Jobcentre Plus sections, DWP sections or other Government Departments e.g. HMRC in connection with benefit claims. ˇ Record tax code and occupational pension indicator on system as appropriate. ˇ Check Child Benefit record for JSA claims from 16/17 year olds and consider eligibility. ˇ Obtain contribution details clerically or using NIRS2, as appropriate. Fraud Investigator To investigate allegations of potential and actual fraud activity; pro-actively, particularly in relation to potential areas of work, where there is evidence that employees are committing or have committed fraud and; referrals from the Generalised Matching Service that meet the criteria for investigation on behalf of the DWP. The main activities associated with this role are to: ˇ Ensure investigations are performed in line with the instructions set out in the Fraud Procedures and Instructions Manual, other user manuals and follow the processes in the Standard Operating Approach (SOA) to ensure compliance with the legal and policy requirements. ˇ Conduct Interviews Under Caution in accordance with The Police and Criminal Evidence Act 1984 (England and Wales) and Common Law (Scotland). ˇ Conduct interviews with customers to offer Administrative Penalties, Formal Cautions (England & Wales) and Administrative Cautions (Scotland). ˇ Make recommendations to the Team Fraud Investigator (TFI) on the outcome of cases and their suitability for a sanction. ˇ Record the outcome of a case on FRAIMS and to make appropriate recommendations to the Decision Maker when a case meets the criteria for a loss of benefit provision. ˇ Develop cases into prosecution cases and to attend Court as a witness when requested. 63 Appendix F: Definitions of fraud and error in the UK Customer error If a customer has failed to disclose, has concealed or misrepresented a material fact that they were required to disclose, and: ˇ this has given rise to an over (or under) payment for the selected period, and ˇ this material fact was not available to the business unit or pension centre paying benefits ˇ the error will be classed as customer error Definition of official error An official error is an error made by an officer of DWP, which no one outside of that Department has materially contributed to. In addition, an error, which is initially classed as customer error, will be classed as official error once the business unit or the pension centre is in possession, from whatever source, of the true facts. Fraud There is no exact legal definition of fraud: most of the offences usually described as fraud are actually offences under Theft Acts of 1968 and 1978 (Common Law in Scotland). But it is generally accepted that fraud involves three key elements: ˇ a deliberate act or omission ˇ with intent to deceive ˇ with a view to obtaining gain for either themselves or another. 65 Social Protection Discussion Paper Series Titles No. Title 1002 Toolkit on Tackling Error, Fraud and Corruption in Social Protection Programs by Christian van Stolk and Emil D. Tesliuc, March 2010 (online only) 1001 Labor Market Policy Research for Developing Countries: Recent Examples from the Literature - What do We Know and What should We Know? by Maria Laura Sanchez Puerta, January 2010 (online only) 0931 The Korean Case Study: Past Experience and New Trends in Training Policies by Young-Sun Ra and Kyung Woo Shim, December 2009 (online only) 0930 Migration Pressures and Immigration Policies: New Evidence on the Selection of Migrants by Johanna Avato, December 2009 (online only) 0929 Ex-Ante Methods to Assess the Impact of Social Insurance Policies on Labor Supply with an Application to Brazil by David A. Robalino, Eduardo Zylberstajn, Helio Zylberstajn and Luis Eduardo Afonso, December 2009 (online only) 0928 Rethinking Survivor Benefits by Estelle James, December 2009 (online only) 0927 How Much Do Latin American Pension Programs Promise to Pay Back? by Alvaro Forteza and Guzmán Ourens, December 2009 (online only) 0926 Work Histories and Pension Entitlements in Argentina, Chile and Uruguay by Alvaro Forteza, Ignacio Apella, Eduardo Fajnzylber, Carlos Grushka, Ianina Rossi and Graciela Sanroman, December 2009 (online only) 0925 Indexing Pensions by John Piggott and Renuka Sane, December 2009 (online only) 0924 Towards Comprehensive Training by Jean Fares and Olga Susana Puerto, November 2009 0923 Pre-Employment Skills Development Strategies in the OECD by Yoo Jeung Joy Nam, November 2009 0922 A Review of National Training Funds by Richard Johanson, November 2009 0921 Pre-Employment Vocational Education and Training in Korea by ChangKyun Chae and Jaeho Chung, November 2009 0920 Labor Laws in Eastern European and Central Asian Countries: Minimum Norms and Practices by Arvo Kuddo, November 2009 (online only) 0919 Openness and Technological Innovation in East Asia: Have They Increased the Demand for Skills? by Rita K. Almeida, October 2009 (online only) 0918 Employment Services and Active Labor Market Programs in Eastern European and Central Asian Countries by Arvo Kuddo, October 2009 (online only) 0917 Productivity Increases in SMEs: With Special Emphasis on In-Service Training of Workers in Korea by Kye Woo Lee, October 2009 (online only) 0916 Firing Cost and Firm Size: A Study of Sri Lanka's Severance Pay System by Babatunde Abidoye, Peter F. Orazem and Milan Vodopivec, September 2009 (online only) 0915 Personal Opinions about the Social Security System and Informal Employment: Evidence from Bulgaria by Valeria Perotti and Maria Laura Sánchez Puerta, September 2009 0914 Building a Targeting System for Bangladesh based on Proxy Means Testing by Iffath A. Sharif, August 2009 (online only) 0913 Savings for Unemployment in Good or Bad Times: Options for Developing Countries by David Robalino, Milan Vodopivec and András Bodor, August 2009 (online only) 0912 Social Protection for Migrants from the Pacific Islands in Australia and New Zealand by Geoff Woolford, May 2009 (online only) 0911 Human Trafficking, Modern Day Slavery, and Economic Exploitation by Johannes Koettl, May 2009 0910 Unemployment Insurance Savings Accounts in Latin America: Overview and Assessment by Ana M. Ferrer and W. Craig Riddell, June 2009 (online only) 0909 Definitions, Good Practices, and Global Estimates on the Status of Social Protection for International Migrants by Johanna Avato, Johannes Koettl, and Rachel Sabates-Wheeler, May 2009 (online only) 0908 Regional Overview of Social Protection for Non-Citizens in the Southern African Development Community (SADC) by Marius Olivier, May 2009 (online only) 0907 Introducing Unemployment Insurance to Developing Countries by Milan Vodopivec, May 2009 (online only) 0906 Social Protection for Refugees and Asylum Seekers in the Southern Africa Development Community (SADC) by Mpho Makhema, April 2009 (online only) 0905 How to Make Public Works Work: A Review of the Experiences by Carlo del Ninno, Kalanidhi Subbarao and Annamaria Milazzo, May 2009 (online only) 0904 Slavery and Human Trafficking: International Law and the Role of the World Bank by María Fernanda Perez Solla, April 2009 (online only) 0903 Pension Systems for the Informal Sector in Asia edited by Landis MacKellar, March 2009 (online only) 0902 Structural Educational Reform: Evidence from a Teacher's Displacement Program in Armenia by Arvo Kuddo, January 2009 (online only) 0901 Non-performance of the Severance Pay Program in Slovenia by Milan Vodopivec, Lilijana Madzar, Primoz Dolenc, January 2009 (online only) 0838 Investing for the Old Age: Pensions, Children and Savings by Vincenzo Galasso, Roberta Gatti and Paola Profeta, December 2008 (online only) 0837 Can the Introduction of a Minimum Wage in FYR Macedonia Decrease the Gender Wage Gap? by Diego F. Angel-Urdinola, December 2008 (online only) 0836 Highly Labor-Intensive Public Works in Madagascar: Issues and Policy Options by Nirina Haja Andrianjaka and Annamaria Milazzo, October 2008 (online only) 0835 Can Conditional Cash Transfer Programs Play a Greater Role in Reducing Child Undernutrition? by Lucy Bassett, October 2008 0834 The Performance of Social Pensions in India: The Case of Rajasthan by Puja Vasudeva Dutta, July 2008 (online only) 0833 Labor Regulations in Developing Countries: A Review of the Evidence and Directions for Future Research by Tito Boeri, Brooke Helppie, Mario Macis, October 2008 (online only) 0832 The Incentives to Invest in Job Training: Do Strict Labor Codes Influence this Decision? by Rita K. Almeida and Reyes Aterido, October 2008 0831 Reforming the Pension Reforms: The Recent Initiatives and Actions on Pensions in Argentina and Chile by Rafael Rofman, Eduardo Fajnzylber and German Herrera, May 2008 (online only) 0830 Community-based Risk Management Arrangements: An Overview and Implications for Social Fund Programs by Ruchira Bhattamishra and Christopher B. Barrett, October 2008 0829 Work History and the Access to Contributory Pensions in Uruguay: Some Facts and Policy Options by Marisa Bucheli, Alvaro Forteza and Ianina Rossi, May 2008 (online only) 0828 A Theory of Contribution Density and Implications for Pension Design by Salvador Valdés-Prieto, July 2008 (online only) 0827 On the Financial Sustainability of Earnings-Related Pension Schemes with "Pay-As-You-Go" Financing by David A. Robalino and András Bodor, July 2008 (online only) 0826 An Ex-Ante Evaluation of the Impact of Social Insurance Policies on Labor Supply in Brazil: The Case for Explicit Over Implicit Redistribution by David A. Robalino, Eduardo Zylberstajn, Helio Zylberstajn and Luis Eduardo Afonso, July 2008 (online only) ­ REVISED AND PUBLISHED AS SP DISCUSSION PAPER NO. 0929 0825 The Portability of Pension Rights: General Principals and the Caribbean Case by Alvaro Forteza, May 2008 (online only) 0824 Pension Systems and Reform Conceptual Framework by Robert Holzmann, Richard Paul Hinz and Mark Dorfman, September 2008 (online only) 0823 Mandated Benefits, Employment, and Inequality in a Dual Economy by Rita Almeida and Pedro Carneiro, August 2008 (online only) 0822 The Return to Firm Investments in Human Capital by Rita Almeida and Pedro Carneiro, June 2008 (online only) 0821 Population Aging and the Labor Market: The Case of Sri Lanka by Milan Vodopivec and Nisha Arunatilake, August 2008 (online only) 0820 China: Improving Unemployment Insurance by Milan Vodopivec and Minna Hahn Tong, July 2008 (online only) 0819 Management Information Systems in Social Safety Net Programs: A Look at Accountability and Control Mechanisms by Cesar Baldeon and Maria D. Arribas-Baņos, August 2008 (online only) 0818 Guidance for Responses from the Human Development Sectors to Rising Food Prices by Margaret Grosh, Carlo del Ninno and Emil Daniel Tesliuc, June 2008 (revised and published as a non-SP Discussion Paper) 0817 Levels and Patterns of Safety Net Spending in Developing and Transition Countries by Christine Weigand and Margaret Grosh, June 2008 (online only) 0816 Labor Regulation and Employment in India's Retail Stores by Mohammad Amin, June 2008 (online only) 0815 Beyond DALYs: Developing Indicators to Assess the Impact of Public Health Interventions on the Lives of People with Disabilities by Daniel Mont and Mitchell Loeb, May 2008 0814 Enforcement of Labor Regulation and Firm Size by Rita Almeida and Pedro Carneiro, May 2008 (online only) 0813 Labor Markets Lending and Analytical Work at the World Bank: FY2002- 2007 by Milan Vodopivec, Jean Fares and Michael Justesen, May 2008 0812 Risk and Vulnerability Analysis in the World Bank Analytic Work: FY2000- 2007 by Valerie Kozel, Pierre Fallavier and Reena Badiani, May 2008 0811 Pension Lending and Analytical Work at the World Bank: FY2002-2007 by Richard Hinz, Melike Egelmelzer and Sergei Biletsky, May 2008 (online only) 0810 Social Safety Nets Lending and Analytical Work at the World Bank: FY2002-2007 by Margaret Grosh and Annamaria Milazzo, May 2008 0809 Social Funds as an Instrument of Social Protection: An Analysis of Lending Trends - FY2000-2007 by Samantha De Silva and June Wei Sum, July 2008 0808 Disability & Development in the World Bank: FY2000-2007 by Jeanine Braithwaite, Richard Carroll, and Karen Peffley, May 2008 0807 Migration, Labor Markets, and Integration of Migrants: An Overview for Europe by Rainer Münz, April 2008 (online only) 0806 Is the Window of Opportunity Closing for Brazilian Youth? Labor Market Trends and Business Cycle Effects by Michael Justesen, April 2008 0805 Disability and Poverty: A Survey of World Bank Poverty Assessments and Implications by Jeanine Braithwaite and Daniel Mont, February 2008 0804 Poverty Traps and Social Protection by Christopher B. Barrett, Michael R. Carter and Munenobu Ikegami, February 2008 0803 Live Longer, Work Longer: Making It Happen in the Labor Market by Milan Vodopivec and Primoz Dolenc, February 2008 (online only) 0802 Disability in Kazakhstan: An Evaluation of Official Data by Ai-Gul S. Seitenova and Charles M. Becker, February 2008 (online only) 0801 Disability Insurance with Pre-funding and Private Participation: The Chilean Model by Estelle James, Augusto Iglesias and Alejandra Cox Edwards, January 2008 To view Social Protection Discussion papers published prior to 2008, please visit www.worldbank.org/sp. Abstract This toolkit is intended to provide a concise but thorough resource for social protection practitioners on how to minimize error, fraud and corruption in their program(s). It is organized in five sections. The first section introduces the topic, by clarifying the key concepts and spelling out the rationale for this activity. The second section provides a generic framework for combating EFC, developed around four building blocks of prevention; detection; deterrence; and measurement. The third section reviews the instruments, tools and mechanisms used to combat EFC in social protection programs, structured by strategic actions (prevention; detection; deterrence), by level of government and over time (how these tools and instruments can be developed over time). Section four includes generic terms of reference for the assessment of the mechanisms for combating error, fraud and corruption (EFC) in social protection programs. The last section summarizes the main findings for three such diagnostics from the Kyrgyz Republic and the Ukraine. HUMAN DEVELOPMENT NETWORK About this series... Social Protection Discussion Papers are published to communicate the results of The World Bank's work to the development community with the least possible delay. The typescript manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally edited texts. The findings, interpretations, and conclusions expressed herein are those of the author(s), and do not necessarily reflect the views of the International Bank for Reconstruction and Development / The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. For more information, please contact the Social Protection Advisory Service, The World Bank, 1818 H Street, N.W., MSN G7-703, Washington, D.C. 20433 USA. Telephone: (202) 458-5267, Fax: (202) 614-0471, E-mail: socialprotection@worldbank.org or visit us online at www.worldbank.org/sp.