Reducing Fraud, Error and Corruption in Social Protection Programs

Transcription

1 Social Protection Core Courses Social Safety Nets For Protection and Promotion: The Design and Implementation of Effective Social Safety Nets Washington, D.C. Reducing Fraud, Error and Corruption in Social Protection Programs Emil Tesliuc February 6, 2009

2 2 1. Overview 2. Lessons from OECD 3. Lessons from Client Countries

3 3 Overview

4 What are we talking about? Definitions Intentional Fraud Corruption Claimant Staff Customer error Official error 4 Unintentional

5 Importance of Mechanisms for Control of Fraud, Error and Corruption (FEC) Ensure efficient use of public resources Guard against political manipulation Ensure program credibility, and thus maintain public support Africa Sub- Saharan Social Protection Spending as a % of GDP East Asia Pacific Social Assistance Eastern Europe and Central Asia Latin America and Carribean Social Insurance Middle East and Northern Africa Even a small % of spending wasted / misappropriated = $$$$$$$ South Asia OECD 23 5 Notes: Data on 75 countries taken from WB Public expenditure review s or other similar w ork. We used OECD-23 since OECD countries such as Poland and Mexico are already accounted for in the regional averages. OECD data from the OECD Social

6 Implementation Challenges Number of participants and payments is large Responsibilities shared across different departments, agencies, levels of government Fraud seems a victimless crime Typically high visibility national initiatives 6

7 Some programs have bigger problems that others, none is immune to FEC India shows high irregularities: 31.6% IAY (housing); GOI audit 2003; 53.5% for SGSY (credit) GOI audit 2003; Up to 90%? SGRY (employment program) - Administrative data suggests that 31.3 million man-days of work were generated in , while the NSS data for suggests that less the 3 million man-days were generated. Polish disability pensions (Hoopengardner 2001): no overall fraud rate, but qualitative study documents price of false sick day at PLN10, and of disability certificate for 9 months as PLN US Food Stamps runs rates of overpayments of about 4% (historically up to 10%), underpayments around 2%; No equivalent studies for LAC or ECA social assistance 7

8 What can you do? Let s Start with the Conclusions! If you want to reduce FEC, you can! No program immune from error or even some degree of fraud, even with best controls but having solid mechanisms to measure, prevent, detect and deter FEC helps Use of multiple instruments strengthens the system Particularly important with decentralized implementation Introduce incentive-based systems, with appropriate rewards and penalties 8 No ready-made solution, No magic bullet Adequate program accountability and control depend on interaction of variety of mechanisms Programs are not isolated within country and reflect character of overall national governance

9 What you can do to reduce the level of fraud, error and corruption (FEC) Use program design to minimize incentives and opportunities for FEC Set up adequate administrative procedures Align institutional responsibilities well Use transparency and communication well 9

10 Use program design to minimize incentives and opportunities for FEC Make program budgets consistent with eligibility criteria to eliminate opportunities for bribery; Simple eligibility criteria imply little discretion and opportunity for bribery, where complexity is necessary reinforce the program with quality control mechanisms Conditionalities can guard against ghost beneficiaries Eliminating intermediaries reduces potential diversion of funds move funds from treasury to bank accounts 10

11 Set up adequate administrative procedures 11 Ensure processes are defined Ensure staff and resources are adequate to carry out the functions for FEC control / reduction Establish sensible tolerances Set up adequate grievance, appeal, whistle-blower protection systems Institute quality control procedures Take action against miscreants with real penalties FYI

12 Institutional Responsibilities Delivery involves actions by and interactions among different entities Incentives must be aligned with roles Reliance on third parties (local government, health and school officials) Role of civil society organizations Definition of responsibilities 12 FYI

13 Use Transparency and Communication Well Ensure eligibility rules, benefit levels, and rules are clear to staff, to clients, to public This reduces error and fraud This reduces unwarranted appeals or complaints 13 FYI

14 Key phases influencing the level of fraud and error, transfer programs OUTREACH EXIT ELIGIBILITY DETERMINATION RECERTIFICATION 14 APPEALS BENEFICIARIES MIS M&E Audit Quality Control PAYMENTS

15 FYI Use FEC lenses on critical phases of the program 15 Phase Institutional Design Eligibility, Recertification Conditionality Payment processes Management Information Systems (MIS) Internal quality control Appeals and complaints Financial management and auditing Procurement Recommendation 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. Minimize inclusion and exclusion errors. Balance accuracy and the burden of verification. Examine reasons for non-compliance. Improve monitoring of cash flows and control procedures. Strengthen framework to analyze and design MIS systems. Promote the crosschecking of enrollments with other databases. Expand process evaluation, feed-back loops, and correction mechanisms within programs, particularly through spot checks. Handle complaints and appeals appropriately and in a timely manner, and publish program data. Define payment accounting and reporting requirements. Focus audits on risks. Include supreme audit institutions. Closely supervise implementation of large service contracts.

16 What do we take away? Some investments have synergies with other metrics of program success e.g. cross-checking data with other systems may both reduce fraudulent claims and lower transactions costs for legitimate claimants. Sometimes control of FEC comes at a cost to other goals simplifying eligibility or benefit criteria helpful to FEC agenda, but reduces the ability of the program to pay just enough to just the right people. 16

17 17 Lessons from developed countries

18 OECD Experience with Control of F&E Lessons: Virtually impossible to be error-free or fraud-free 18 With adequate mechanisms, the level of fraud and error can be substantially reduced: In the US Food Stamp Program, error rate was 8.4% ( ), with a declining trend actions taken UK reduced the level of fraud in half from 2000 to 2008

19 What is the suite of activities used in OECD countries? Prevention - Improved verification on eligibility - Provision of information to applicants and public - Sanctions Deterrence Detection - Risk-based and random reviews - Data matching - Telephone hotlines Measurement - Often linked to performance management

20 Country Example: UK Institutional framework Department for Work and Pensions (DWP) responsible for policy formulation and oversight Administered by 4 different organisations - Jobcentre plus - Disability - Pension service - Local authorities 27 different benefits: - Contributory (eg Unemployment/sickness) - Non Contributory (eg Disability) - Means tested (eg Lone parent/pensioner) 20 Number of claims, beneficiaries, and budget - 50 million claims - 17 million people receive benefits Benefits expenditure 120 billion in 2006/07-120,000 million UK pounds - 240,000 million US Dollars - 25% of public expenditure - above 10% of GDP

21 Expenditure, fraud and error in DWP benefits Expenditure Housing Benefit State Pension 58bn spend 0.2% ( 90m) OP 0.3% ( 150m) UP Disability Living Allowance 10bn spend 1.9% ( 190m) OP 2.5% ( 250m) UP 16bn spend 4.6% ( 730m) OP 1.2% ( 190m) UP Income Support 9bn spend 5.0% ( 450m) OP 1.3% ( 120m) UP Key: Size of circle represents expenditure, red sector size represents the proportion or fraud and error overpayments (OP), the blue sector represents underpayments (UP) Incapacity Benefit 7bn spend 2.1% ( 140m) OP 0.7% ( 50m) UP Carers Allowance 1bn spend 5.5%( 70m) OP 0.1% UP Council Tax benefit Fraud and error (OP) Pension Credit 7bn spend 5.3% ( 390m) OP 1.7% ( 120m) UP Benefit expenditure (bn) % Overpaid % Underpaid 4 bn spend 4.1% ( 160m) OP 1.1% ( 40m) UP Jobseeker s Allowance 2bn spend 4.2%( 90m) OP 0.3% ( 10m) UP

22 Contributory Pension What is it? A set payment to individuals or couples when they reach a certain age. Retirement Pension is paid to those who have met a level of contributions throughout their working life So low risk Risks to Fraud and Error Unreported death Unreported separation or divorce

23 Income Support (IS) What it is Benefit which provides financial help for workingage people who are not required to be available for work, whose income is below a prescribed level, and who work less than 16 hours a week.. Must be aged over 16 and not have Capital exceeding 16,000. Non-contributory. Income-related. Tax-free. So high risk Risks to Fraud and Error Falsifying Circumstances. Working and claiming. Not declaring or under declaring capital amount. Customer misunderstanding of when to tell of a change in circumstances. Complexity staff getting it wrong.

24 DWP Practices: Preventing Fraud 24 Tighter checks at outset of claim On identity On circumstances Payments direct to bank accounts instead of by cheque Staff training to ensure benefit awarded is correct Fraud-proofing new policy and operational developments

25 DWP practices: Detecting fraud 25 Data matching, Benefit and other data (eg Tax) Investigation Professionalism Powers to obtain information (eg from banks) Intelligence Improved gathering, management and use Organised crime Risk profiling To identify higher risk claims Target additional checks To identify internal fraud Telephone hotline for reporting fraud Joint operations with other agencies e.g Revenue/customs/Immigration authorities Police

26 DWP: Deterrence Sanctions Extended to include cautions/administrative penalties Extra sanction Two strikes for persistent offenders Confiscation of Assets 26 Influence public attitudes: mass-media campaign on Targeting Benefit Thieves Benefit fraud unacceptable Not victimless crime

27 Targeting benefit thieves Campaign 27

28 Poster Campaign

29 Classified Adverts for Local Press

30 Puzzles Page for targeted magazine campaign

31 Keith the benefit thief

32 % of benefit expenditure What has been the impact on Fraud and Error? 3.5% Proportion of total benefit expenditure overpaid due to fraud and error: 2000/ /08 3.0% 0.4bn 0.4bn Fraud Customer Error Official Error 2.5% 0.7bn 0.6bn 0.7bn 2.0% 1.5% 1.0% 2.2bn 2.2bn 0.6bn 1.8bn 0.8bn 0.7bn 0.9bn 0.9bn 0.9bn 1.0bn 0.9bn 1.0bn 0.9bn 0.8bn 0.5% 0.0% 1.0bn 0.8bn 0.6bn 0.8bn 0.8bn 2000/ / / / / / / /08

33 Cost-effectiveness of EFC interventions in UK 33 Referrals, National Data Matching Service, lead to detection of 24 for 1 spent (27 for local authorities) Referrals, National Benefit Fraud Hotline, lead to detection of 16 for 1 spent Customer compliance sessions helped identify 0.62 for 1 spent, and reduced overpayment by Advertising campaign increased calls to the Hotline identified an estimated 900,000 Indirect effects, deterrence or prevention effects not included, as hard to quantify

34 34 Client Countries: Trends and Cases

35 Oversight and Controls in the BFP Supreme Audit Agencies Federal CGU* (Executive Branch) Regular random sample QCRs Annual financial audits Investigation of specific cases Publication of beneficiary lists Federal TCU* (Legislative Branch) Implementation evaluations, with supervision and follow up for the Cadastro Único; the BFP Ministerio Público* (Independent) Investigation of specific cases MDS Program Controls Internal and external cross-checks of the registry Performance-based management incentives administrative subsidies 35 Municipalities Municipalities Municipalities Municipalities Municipalities Municipalities Municipalities Social control councils hotlines

36 Spot Checks in Colombia CCT Mechanism of supervision in the field of application of operational procedures for Program implementation. Why use Spot Checks? Internal process evaluation, feedback loops and correction mechanisms critical: a practical way of finding out how Program operates on the ground and whether as intended Promote attention to building capacity to investigate and correct problems Conducted in real time Complement externally contracted financial and other audits 36

37 Method used in Colombia Contracted by National Coordination Unit Supervised by Planning and Supervision Department Carried out every 6 months Initially same topics covered in each round. More recently the Program narrowed focus, choosing particular themes (Adequacy of allocation of administrative resources) 20 municipalities in 4 regions, chosen randomly, as were others to be interviewed 4 rounds of spot checks contracted externally with firm: team of 47 (13 central level, 4 regional coordinators, 8 supervisors, 8 assistants, 14 interviewers) Total Cost of US$460,000 Structured Interviews with key actors using survey (ranging from questions each) Program administrators, local governments, participants 37

38 Examples of Indicators (400) Knowledge Mother s knowledge of specific themes Familiarity of clinic and schools staff of their Program guide Utilization of materials Do mother leaders use materials? Do school and clinic staff consult guides? Compliance with operational procedures Do mother leaders report on complaints? How many days do the Banks allow for payments, what is the waiting time? Infrastructure Do regional offices have necessary equipment? Do mothers stand in line at the bank inside or outside? Organization % of schools that require written excuse for absence Procedures % of complaints resolved at the Regional Coordination Unit % of municipal enlaces who have lists of schools, clinics, and completed stickers 38

39 Selected Findings and Results 39 Municipal Representatives do not completely understand Program and not dedicating enough time (3 rd round improvement, 4 th round more updated documentation) Low Commitment of schools and health centers causes problems with verification of conditions, time of mothers Minor amount of $ not claimed by mothers, UCN errors Payment processes often modified, for the better Low Participation of Mothers in Education Sessions Expansion in role of Mother Leader Program asset (3 rd round helped in verification of conditions) Insufficient budget for Regional Coordination Units to fulfill role procedures falling into disuse Program made changes based on the results of the spot checks (negotiated with Banks to expand services, discussed salaries of municipal reps. with mayors, expanded training)

40 Argentina Jefes de Hogar: Control and Accountability Hotlines Multiple purposes Complaint Resolution Provide Information (is a beneficiary on the payment list, payment information, am I eligible?) Social Accountability Initial implementation problematic, subsequent improvements in IT, staffing, staff training, answer tree, coordination with CODEM 40

41 Call center use in Jefes 41 Where it fits in structure: Toll free hot line manned by call center CODEM Commission in Ministry of Labor to handle allegations of program abuse or complaints Criminal offenses referred to Federal Prosecutor of the Social Security System Coordination between Call Center and CODEM Call center receives 84% of allegations of abuse and 74% of complaints Agree on information to be collected Developed standard formats and answers to common questions Made referrals from call center to CODEM more efficient (on-line, IT system). Simplified process and reduced administrative delays. Enabled local administrators to monitor progress on cases. Publish monthly report on CODEM intake and results

42 Lessons on call center 42 Important in Heads of Household a very large program, quickly implemented, surrounded by confusion and controversy, Beneficiaries need it for information. Preferred information source National Program decentralized admin, no resources provided to municipalities Complements CODEM, Publicity, Program Changes (Debit Cards) Large effort which requires sufficient initial investment and resources. Needs to be done well (intake and followup) otherwise harms credibility. When? Large programs, nation-wide, economies of scale, sufficient IT support and telecommunications capacity

43 Ukraine: Improving the efficiency of verification procedures (I) SSN spending equivalent to 2% of GDP 17 programs for families with children, disabled, and the poor (income-tested) The information on household circumstances related to eligibility is verified extensively by a Social Inspection under the Audit Office of the Ministry of Labor and Social Policy social inspectors in 2008 The overall rate of inspections is high (43% of active cases), but detection rate is low (0.5%), and negatively correlated, because: Intelligence about high-risk clients or programs is absent or insufficient Inspections are done at random, not targeted to high-risk cases

44 Ukraine: Improving the efficiency of verification procedures (II) The error detection rate is negatively correlated to the share of application checked

45 Ukraine: Improving the efficiency of verification procedures (III) Proposal under consideration: target inspection to high-risk cases. Identify high-risk cases by a combination of intelligence gathering tools: There are three ways in which inspections can be more targeted: building up profiles of clients who are at risk of committing benefit fraud from the results of past audits and inspections. gather systematic intelligence through data-matching and tip-off from the public or the social protection staff. Identify high-risk clients by collecting addl information on household characteristics associated with income, use this information to predict the income, and target inspections to cases with the largest difference between reported and predicted per capita income. This model will double the probability of identifying an erroneous or fraudulent case compared to the current practice: Beneficiaries of the Last Resort Program ranked by their actual and predicted per capita income Q1 Bar Q2-Q5 Bar Total Q Q2-Q Total

46 Russia, Income-tested programs Eligibility is based on the reported (per capita) income of applicants however Applicants do not report some income sources (those harder to verify, like agricultural or entrepreneurial income) Other incomes are under-reported Verification is absent Sanctions are mild and ineffective Result: High inclusion error

47 Housing allowances Child allowances Russia, Income-tested programs 47 We compared the income sources reported by program beneficiaries in a survey with the those reported to the Social protection offices (quality control review) And found out that most hard-to-verify income goes under-reported; some of the verifiable income as well. Wages Remittances Scholarships Farming, hunting, fishing Self-employment Wages Remittances Scholarships Farming, hunting, fishing Self-employment % families reporting a particular type of income Quality control review, 2006 Difference in the NOBUS, 2003

48 Trends from client countries greater complexity in eligibility and payment structure greater use of technology, at least in MICs more explicit thought on transparency 48 greater attention to implementation issues

49 49 Extra slides

50 Fraud and Error Context in UK DWP Fraud and Error Strategy Division Jobcentre Plus Disability & Carers Service The Pension Service Local Authorities Debt Management Lawyers Benefits Working Age Disabled Pensioners (over 60) Housing Costs Error Strategy Identify & Correct Identify & Correct Identify & Correct Identify & Correct Analysis Fraud Strategy Investigation/Sanction (All DWP Benefits and internal Fraud) Investigation Legal Powers Prosecution Authority Overpayments Refer to Debt Management Refer to Debt Management Refer to Debt Management Recovery Responsibilities 50

51 51 UK: Network of counter-fraud activities,

52 UK: Measuring the level of error and fraud in the benefit system 52 To measure the level of fraud and error, the UK re-examines a sample of beneficiaries each year through a rolling programme of Benefit Reviews. The Benefit Review Process: 55,000 cases are extracted as a random sample from the total number of active cases (17.2 million), to be re-reviewed (controlled) Each agency gathers records for each of the sampled cases Benefit Review Teams check electronic and clerical records on each case for inconsistencies and errors, and with the beneficiary. Where appropriate, additional information is gathered via home visits (41,000 per year) If suspected of fraud, the case is referred to the Fraud Investigation Service (FIS). Penalties are applied to fraudulent cases. The FIS is notified of the result. Each case is classified as correct, customer error, official error or fraud. Next: Data Analysis

53 UK: Measuring the level of error and fraud in the benefit system 53 Data Analysis The level of error and fraud found in the sample is extrapolated to the total population of claimants. Disaggregated by type of benefit, region, type of error and fraud These estimates are qualified by confidence intervals that take into account the fact that a relatively small sample is used in relation to the size of the actual caseload. Methodology approved by National Audit Office and National Statistics. Uses of the resulting data. These figures are used by: The Government, to measure progress against objectives (Public Service Agreement targets) The Department: Identify trends and levels of loss. Improve understanding of the causes and patterns of fraud and error. Inform the counter fraud strategy. Allow to refine the benefit processes to reduce the potential for administrative or customer mistakes. Used to estimate the likely savings for counter fraud or counter error activities.

54 54 Reduction in the Food Stamp Error Rates