Controlling for Error, Fraud and Corruption (EFC)

Transcription

1 Controlling for Error, Fraud and Corruption (EFC) Bangkok Meeting, World Bank March 2-4, 2010 Christian van Stolk

2 Content 1. Introduction 2. A framework for combating EFC 3. Tools and instruments used to combat EFC 4. General comments on applicability of EFC framework 2 02/05/2010

3 Evidence drawn from work on OECD countries as well as missions to MICs and LICS In OECD we looked at: international fraud and error rates examples of good practice or common approaches the cost-effectiveness of anti-fraud interventions With the WB, we undertook: missions in the Ukraine, the Kyrgyz Republic and Bangladesh 3 02/05/2010

4 Why focus on EFC? A small fraction of EFC in programmes can add up to large amounts 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) Evidence shows that no social protection program is immune from EFC (EFC can be 5-10% of expenditure in OECD context) A higher prevalence of EFC can be associated with: An inefficient use of public resources Availability of fewer funds to the poor A loss of credibility of the program Political manipulation of the funds A decrease in public support for the program 4 02/05/2010

5 What do we mean by EFC? Intentional Fraud Corruption Claimant Staff Customer error Official error Unintentional 5 02/05/2010

6 What is the difference between targeting error and EFC control? Targeting error indicates the fraction of program funds that are not reaching the poor EFC indicate the extent to which the program is compliant with its own rules A programme could be well targeted but have high EFC An income-tested program could have 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% will be affected by EFC 6 02/05/2010

7 What are some of the causes of fraud and error? Fraud: Error: Customer dishonesty (e.g. failing to report changes in material circumstances) Exploiting the system (e.g. identity fraud) Complexity of the social protection system (e.g. crossjurisdictional claims) Staff (e.g. excessive staff caseloads) System (e.g. failure of payment and IT systems) Complexity of benefits and rules confuse administrators and benefit claimants 7 02/05/2010

8 What are some of the causes of corruption? Corruption: Exploitation of design aspects of the program Collusion between staff and claimant Staff taking bribes to enrol individuals in the program Collusion between staff processing benefit claims and those paying out benefit claims 8 02/05/2010

9 What is an acceptable level of EFC for OECD? We found that fraud and error in five OECD countries the UK, Canada, Ireland, New Zealand and the USA ranged between 2% and 5% of overall expenditure means-tested social protection programs had the highest fraud and error rates (5-10%) unemployment benefit and disability pension programs (1-2%) Old-age pensions had the lowest rates (0.1-1%) These figures represent lower bounds as they come from countries and programs with high administrative capacity 9 02/05/2010

10 Content 1. Introduction 2. A framework for combating EFC 3. Tools and instruments used to combat EFC 4. General comments on applicability of EFC framework 10 02/05/2010

11 What is the intervention logic in tackling EFC in OECD countries? Inputs: Resources of process Process: Activities - Staff - Training - Results-based management - IT - Offices - Prevention - Detection - Deterrence Outputs and outcomes: Monitoring and performance management - Measurement of what processes do - Measuring the impact of processes - Measuring overall fraud and error - Measuring costeffectiveness 11 02/05/2010

12 What is the strategy used in OECD countries? Prevention - Improved eligibility checks - Information provision Deterrence - Sanctions Detection - Risk-based and random reviews - Data matching - Telephone hotlines 12 02/05/2010

13 What is needed to implement the strategy? Political will affects willingness to tackle EFC influences how EFC is reported, measured and monitored Administrative capacity refers to: staff (e.g. training and skills) material support (basic support required for staff to perform functions) results-based management (incentivisation of staff to reduce EFC) IT capacity (e.g. capacity to process and verify information) resources (e.g. dedicated budget to support EFC control) networks (e.g. joint working within government) public (e.g. public awareness on what EFC is) 13 02/05/2010

14 What do we know about effectiveness of a strategy: example of the UK? Expenditure overpaid ( billions) Official Error Customer Error Fraud / / / / / / /05/2010

15 What do we know about effectiveness more widely? Evidence from UK and Australia suggests that EFC actions can pay for themselves through the amount of fraud and error identified This is a powerful argument for other countries to invest in EFC controls We know less about marginal effects A number of interventions were shown to be cost-effective Preventative effects can be large but hard to measure (In UK possibly majority of fraud and error identified) Deterrent effect is hard to measure and is likely to add to costeffectiveness 15 02/05/2010

16 Content 1. Introduction 2. A framework for combating EFC 3. Tools and instruments used to combat EFC 4. General comments on applicability of EFC framework 16 02/05/2010

17 How do we classify policy instruments in the OECD context? As indicated, interventions fall in four main categories Prevention Detection Deterrence Monitoring 17 02/05/2010

18 What do we know about what works in prevention? Outcome Actions Results Prevention Fraud-proof program design and operational aspects of program Make payments directly to bank accounts instead of by check or cash IT systems Clear allocation of institutional responsibilities Setting up proportionate administrative procedures Guidance Staff training Resources Checks at outset of claim Making the public aware of the rules and obligations of the program and the issue of EFC in social protection can prevent EFC occurring Considering EFC control upfront in the program design stage when setting out the eligibility and administrative procedures reduces the risk of EFC to the program Making payments more directly to applicants cuts out important opportunities for EFC to occur Adequate IT systems allow for the easy cross-checking of data enabling staff to find fraud at the outset of a claim (requires unique identifier number) IT systems also reduce human error in processing Division of administrative responsibilities (monitoring, payment and processing of benefit) is good practice to reduce the risk of EFC in most OECD countries A complex or multi-tiered administration might increase the risk and opportunity of EFC occurring Most staff in OECD countries operate according to clear job descriptions and standard operating procedures to minimise error Ensuring that staff is trained to an appropriate standard is important in reducing error Ensuring staff have appropriate resources and are reimbursed for travel expenses can reduce error and indeed corruption Fraud, error, and corruption are harder and more costly to detect during the claim compared to identifying it at the outset of the claim Social control, transparency, and accountability are critical elements in any social protection program 18 02/05/2010

19 What do we know about what works in detection? Outcome Actions Results Detection Data matching Comparing data held across a variety data sources is highly cost-effective at detecting cases of suspected fraud and corruption Intelligence Gathering information from the public (tip-offs), telephone 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 the information provided to the program and of the decisions made by staff) are necessary aspects in detecting EFC Risk profiling Risk-based assessments to organise reviews (e.g. select client investigations) using scoring or profiling are particularly cost-effective in detecting fraud Joint operations with other agencies Information exchange and joint investigations are useful in detecting and investigating organised crime and high-level fraud and corruption 19 02/05/2010

20 What do we know about what works in deterrence and monitoring? Outcome Activities Result Deterrence Sanctions Prosecutions Confiscation of assets Applying appropriate sanctions consistently serves as an important deterrent for those considering engaging in fraud and corruption Measures to influence public attitudes, such as public awareness campaigns. See similar point on prevention Monitoring Regular measurement of the level of EFC Understanding the overall problem of EFC and particular risks to the program is of critical importance in organising reviews of cases and understanding the effectiveness of the entire strategy 20 02/05/2010

21 What are the main instruments to combat EFC per level of government? Independent Audit Agencies Ministry Responsible for Programme Local Level Financial audits Regular random-sample operational audits Case investigations & follow up Implementation evaluations (GAO type) Publication of beneficiary lists on transparency portal Internal & external registry cross-checks Hotlines with trained operators Case investigations & follow up Monitoring mechanisms for third party implementation (e.g., municipal governments, payments agent, NGOs) Publication of beneficiary lists on transparency portal Regular recertification of eligibility Local point of contact for complaints, appeals Follow up investigations (case-by-case) Social controls councils (citizen oversight) 21 02/05/2010

22 What are some of the factors in how OECD EFC controls have developed? Increased investment in IT and administrative capacity in EFC control Improvements in technology (e.g. the application of new technologies in terms of communicating with the public, processing of benefits, making payments to beneficiaries, and matching data) Learning over time from your actions (e.g. moving to more risk-based reviews as the agency learns what the EFC risks to its operations are through analysis and intelligence gathering) 22 02/05/2010

23 Content 1. Introduction 2. A framework for combating EFC 3. Tools and instruments used to combat EFC 4. General comments on applicability of EFC framework 23 02/05/2010

24 Are OECD interventions likely to work in mid- and low-income countries? How to overcome resource and capacity constraints? Investments in EFC seem to pay off in OECD (or can be at least cost-neutral) Capacity issues can be overcome by using capacity more efficiently Risk-based approaches Increased use of IT How relevant are EFC controls to small SS systems? Systems are likely to grow and upfront investment can prevent a bigger and growing problem later Learning is possible given evidence available 24 02/05/2010

25 How are these interventions currently used in MICS and LICS? In less developed countries, EFC control instruments seem underutilised pilot mission to the Ukraine and the Kyrgyz Republic found a very limited use of deterrence 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 However, the World Bank has also found some positive trends These include greater use of technology, at least in MICs A number of MICs are making large investment in management information system, a key tool for the development of an effective policy to combat EFC 25 02/05/2010

26 Ukraine: Development of a comprehensive strategy Strategy to reduce EF&C developed with WB support (Apr 2008) and introduce new elements to increase cost-effectiveness: Improve legislative framework: right s and obligations of beneficiaries, clarify the right of inspectors to search and collect evidence, reform sanction regime Improve the detection of application suspect of error or fraud via better intelligence gathering (data matching, hotlines) Move from random-based inspections to riskbased ones - Client profiling model developed, being piloted Change public attitudes toward benefit fraud: public relations campaign Example of TA work Currently, inspections to verify eligibility are done randomly -Too many inspections and too little detection of overpayments Replace random-based inspections with riskbased inspections, using a client profiling model - the model doubles the probability of inspecting an erroneous or fraudulent case compared to the current practice. 100% 80% 60% 40% 20% 0% Random-based inspection of beneficiaries: Error detection rate is negatively correlated to the share of application checked Beneficiaries, Guaranteed Minimum Income Program, ranked by actual and predicted income All Beneficiaries Predicted poorest (28% of beneficiaries) Not-poor Poorest 20% Predicted non-poor (72% of beneficiaries) 26 02/05/2010

27 How can the experience of OECD countries be translated to LICs and MICs? Any recommendation for improvement needs to pass the test of feasibility, suitability, and acceptability. In other words: how feasible is a recommendation given the institutional architecture and administrative capacity of the country how suitable or effective are the actions to address the particular EFC problem how acceptable are these actions likely to be to political decision-makers, administrators, the claimants and wider stakeholders This leads to a realistic assessment on how the EFC strategy can be improved in the short, medium, and longer term 27 02/05/2010

28 How does EFC control fit in wider program design initiatives? EFC mitigation can be an input to program design and management, but not a substitute for what is considered good program implementation 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 28 02/05/2010

29