C. A. MacDonald & Associates would like to acknowledge the assistance of a wide variety of individuals in putting together this report as part of the Fraser Institute 1995 Waste in Government competition. In particular, senior officials in most Canadian provinces and territories were extremely helpful in identifying issues and solutions at work within the various jurisdictions.

In the United States, the assistance of the executive and members of the United Council on Welfare Fraud (UCOWF), as well as many state and local officials were also of great assistance. This project could not have been completed without their help.

C. A. (Tina) MacDonald                                                                                     Sheila Blair

TABLE OF CONTENTS

EXECUTIVE SUMMARY

INCOME SECURITY FRAUD IN CANADA
        The Canadian Context
        The Issue of Client Overpayment
        The Magnitude of the Problem
        Current Solutions
        The Missing Pieces

LEADERSHIP - THE CRITICAL COMPONENT
        The Federal Role
        Leadership Components
        Implementation Strategies
        Implementation Costs
        Issues

CLIENT IDENTIFICATION
        Identifying Clients
        The Current Situation
        The Wages of SIN
        Social Insurance Number Administration
        Direct Benefits
        Implementation Strategies
        Implementation Costs
        Issues

DATA-MATCHING
        Description of Problem
        The American Experience
        Current Canadian Strategies
        Description of Initiative
        Direct Benefits
        Implementation Strategies
        Implementation Costs
        Issues

TAX RECOVERY PROGRAMS
        Description of Problem
        Current Strategies
        Tax Refund Intercept Programs
        Direct Benefits
        Implementation Strategies
        Implementation Costs
        Issues

ADMINISTRATIVE DISQUALIFICATION
        Description of Problem
        Current Strategies
         American Administrative Disqualification Programs
        Direct Benefits
        Implementation Strategies
        Implementation Costs
        Issues

EARLY PREVENTION AND DETECTION
        Description of Problem
        Current Strategies
        Description of Initiative
        Direct Benefits
        Implementation Strategies
        Implementation Costs
        Issues

OVERALL STRATEGIES AND TIMING
        Putting It Together
        Implementation Considerations and Strategies
        Possible Timing of Activities
        Conclusion

There are several critical types of fraud in the various income support systems which are difficult to identify and prevent because of current federal legislation. The following types of fraud account for the majority of inappropriate expenditures, overpayments and waste:

1. Undeclared Income - Clients in all of the income support programs are required to declare their income. Various program regulations define how this income is handled, ranging from income exemptions, through benefit reductions or complete disentitlement. Undeclared income results in significant overexpenditures and unacceptable drains on publicly funded programs.

 
2. Multiple program claims - Clients are often eligible for more than one income support program. When income from other programs is reported by the client, various regulations clarify how that income affects benefit levels and eligibility. However, when this income is not reported, overpayments result.

 
3. Cross-jurisdictional claims - Clients who are willing to travel can establish eligibility in similar programs offered by several jurisdictions (eg. two or more provincial welfare programs plus municipal welfare benefits). This results in overpayments, particularly at the provincial level, but these overpayments may have been cost-shared at 50% by the federal government.

In the United States, the federal government has taken leadership in the area of fraud control through legislation requiring states delivering the federal programs of Food Stamps, Aid to Families with Dependent Children (AFDC) and Social Security to implement fraud control programs. These fraud control programs are also cost shared by the federal government. Other federal legislation around security and privacy supports these initiatives.

In Canada, not only is there no leadership in this area, current legislation actively works against detection and prevention of fraud and abuse in the various income support programs. The following main thrusts would establish a strong federal leadership role and a climate for appropriate fraud and abuse control in all Canadian income support programs.

1. Establish a priority for fraud prevention and detection - Through the development of new legislation and/or adapting current legislation (eg. CAP legislation), identify a requirement for federal and provincial income support programs to actively prevent and detect fraud and abuse. Requirements for reduction of administrative and client error could also be included. (Only a few provinces are currently active in fraud prevention).

 
2. Require the use of a common client identifier - The Social Insurance Number (SIN) is the obvious choice for a client identifier. However, current legislation on the use of the SIN prevents some programs (most noticeably provincial income support programs) from requiring applicants and spouses to disclose or obtain a SIN number as a condition of eligibility. Cross-program and jurisdictional comparisons cannot occur without a common identifier.

 
3. Improve access to federal tax information - The confidentiality provisions in the current income tax system prevent income support programs from accessing data to identify current income sources and amounts except in individual cases. Routine data matching is required to ensure early detection and prevention rather than the current system of using the income tax information in specific cases to confirm the existence of fraud found through other methods.

 
4. Facilitate routine cross-program and cross-jurisdictional data matching - As the federal government embarks on a major program to redesign the Income Security computer systems, a key objective should be to ensure the feasibility of determining from which other programs an applicant or current client is receiving assistance. In effect, this means the creation of a client index for all federal programs (including federally funded natives receiving assistance on reserves) which could also be accessed by provincial and municipal administrations.

 
5. Allow administrative disqualification for assistance - In Canada, enforced by CAP legislation, benefits are based on need only, and past or current fraudulent behaviour has no impact on current and future eligibility. In the United States, one identified instance of fraud or program abuse can result in a six-month disqualification, two instances can result in one year of disqualification, and more than two instances of fraud can result in a lifetime disqualification. If this were implemented in Canada, information on fraud and abuse disqualifications should be shared between programs and jurisdictions.

 
6. Expand the use of the tax system to recover income support overpayments - Currently, the record on recovering overpayments in provincial income security programs, particularly on closed files, is dismal. The tax system is regularly used to recover UIC overpayments. Expanding the use of the tax system to include overpayments in other federal and provincial and municipal income support programs would result in millions of dollars being recouped annually. This is referred to as a "tax-intercept program" in the United States.

 
7. Encourage the establishment of early fraud detection and prevention systems - Over 30 American states, at the direction of the federal government, have established early detection systems which intercept probable cases of fraud before assistance is granted by integrating fraud prevention and detection into the intake system. Cases are referred for investigation if certain criteria are met (such as lack of or forged identification, needs consistently exceeding income, etc.). In Florida, California, and Texas, for example, approximately 70% of the cases result in assistance being denied or benefits reduced.

 

As the details in each chapter explain, it is often difficult to estimate a specific cost-benefit for many of the initiatives. However, the American experience clearly demonstrates that all these proposals offer a positive, and sometimes very high, return on dollars invested. Canadian Income Security programs cost over $85 billion, $39.2 billion of this expenditure supports programs considered to be at high risk of fraud and client error. A conservative estimate of the amount of fraud in these programs is 4%, and client error is estimated at 5%. Therefore, the magnitude of the fraud and error "problem" in income security programs in Canada is approximately $3.5 billion dollars, spread across federal, provincial, territorial, and municipal programs. If it is assumed that the initiatives proposed in this document offer the potential to reduce the level of fraud and client error by one quarter to one-third, this would result in a total savings to these Canadian income security programs of something between $875 to $1,155 million annually when fully implemented.

This is an ambitious proposal with many facets. However, it offers significant savings at a time when the country needs deficit reduction strategies. The mood of the electorate also makes this a timely proposal. It is clear that taxpayers want to see expenditures reduced rather than taxes raised. However, they also do not want to see their social programs decimated. This proposal allows managed reductions at all levels of government without affecting the benefits available to those who truly need assistance.

The Canadian Context

A national effort, involving all levels of federal, provincial, territorial, and municipal governments, to control this costly leakage from the overall income security system in Canada would engender wide-spread support from all Canadian provinces and the vast majority of Canadian taxpayers since it demonstrates an active concern for reducing government waste. This is consistent with current proposals to reduce "cheating" in the Canadian tax system.

The Canadian income security system is a complex mix of federal, provincial, and municipal programs, including the following:

• Canada Pension Plan
• Old Age Security
• Guaranteed Income Supplement
• Provincial and municipal "welfare"
• Veteran's Affairs
• Student loans
• Unemployment Insurance
• Worker's Compensation
• Child Benefit
• DIAND Income Support Programs
• Special support programs (eg. AIDS)
• Training Allowances

The Issue of Client Overpayment

All of the categories of error and fraud overlap, and it is often a matter of convenience or legal requirements which determine how a particular case is labelled. Generally, the overall caseload profile may be characterized as shown in the accompanying graphic.

A key question for many jurisdictions is the total amount of fraud and error which actually exists in the income security system. Obtaining a definitive answer to this question is difficult, if not impossible, since empirical studies and published data are rare, and unfortunately the data that are available is often a reflection of the degree of emphasis that the jurisdiction has placed on uncovering fraud rather than an absolute measure of the amount of fraud. A high level of reported fraud is usually indicative of an active fraud detection and prevention system and a low level of reported fraud often demonstrates nothing more than a state of blissful ignorance. This is a difficult concept for politicians, the public and the media to grasp. However, in our experience, after surveying many Canadian and American jurisdictions, this clearly appears to be the case. One of the few Canadian jurisdictions which has attempted to answer the question of the level of fraud and error in income security programs in an empirical way is the province of Alberta. Through a process known as Eligibility and Benefit Verification, a random sample of active welfare cases was examined in depth every month for several years. This process included a complete administrative file review, re-interviewing of the client in their home and re-establishment of eligibility, often involving contact with landlords, employers, banks, etc.

The most recent public data available⁽¹⁾

is for the 1988-89 fiscal year when this process resulted in 22% of cases being recommended for closure due to an inability to verify continued eligibility. These closures were recommended because:

• the address did not exist
• the client was not at the reported address
• the client requested case closure due to recent employment
• the client refused to take part in the review

• 4% were referred for fraud investigation
• 15% of the files were complete and up to date
• 20% of the files contained minor errors not related to eligibility or benefit levels
• 17% of the files contained underpayments
• 22% of the files contained overpayments (only 32% of these files, or 7% of the total, were a result of administrative error - the rest were a result of client errors)
• 22% of the files were recommended for closure because they could not be verified at all.

For purposes of any overall typology of fraud and error control programs, both legal fraud and program abuse are handled under the category "fraud", since the mechanisms and procedures used to identify the incident are identical. Non-financial errors, which are more of a case management issue, are not addressed directly in this document. However, once a non-financial error has been reviewed and clarified as having a financial impact, it may then be classified as administrative or client error.

There are a number of mechanisms which income security clients typically use if their intent is to defraud the government out of additional funds. These techniques include:

• Undeclared income
• Multiple program claims
• Cross-jurisdictional claims
• Forged endorsement of stolen cheque
• Additional adult with income in the home
• False declaration of lost-cheques/cash
• False description of family size/structure
• Fraudulent needs declaration or verification
• Undeclared assets
• Multiple IDs
• Lending of client identification

1. Undeclared Income - Clients in all of the income support programs are required to declare their income. Various program regulations define how this income is handled, ranging from income exemptions, through benefit reductions or complete disentitlement. Undeclared income results in significant overexpenditures and unacceptable drains on publicly funded programs.

 
2. Multiple program claims - Clients are often eligible for more than one income support program. When income from other programs is reported by the client, various regulations clarify how that income affects benefit levels and eligibility. However, when this income is not reported, overpayments result.

 
3. Cross-jurisdictional claims - Clients who are willing to travel can establish eligibility in similar programs offered by several jurisdictions (eg. two or more provincial welfare programs plus municipal welfare benefits). This results in overpayments, particularly at the provincial level, but these overpayments may have been cost-shared at 50% by the federal government.

The Magnitude of the Problem

One American state which has provided information to assist people to understand the difficulty of estimating the amount of fraud is California. Although the California Strategic Plan⁽²⁾
does not specifically state the amount of fraud in California's Aids to Families with Dependent Children (AFDC) program, it does present a variety of project results which estimated the amount of fraud in the AFDC program. These results are presented below. It should be noted that the methodologies of these reported studies are quite different, and the results from each project are not directly comparable. However, the information does indicate that the amount of fraud detected, depending on how fraud is defined and the methodology used, is extremely variable, ranging from 4% to 100%. This supports the conclusion that the more you look, the more you find. The more "pointed" the investigation, the more fraud can be found in income security files.
 
 

Bringing Integrity to Welfare in California A Strategic Plan for Eliminating Fraud, Waste and Abuse - January, 1994 What is the Extent of Fraud in AFDC?   4 percent of cases randomly selected for Quality Control review in 1992-93 were estimated to involve fraud. (Because state QC reviewers lack many of the tools that investigators use to detect fraud, this QC- based figure should be regarded as a low estimate.) 10-13 percent of all cases and 5-6 percent of all payments: 1977 Conputer-Aided Techniques Against Public Assistance Fraud prepared by the MITRE Corporation for the US Department of Justice. 15.8 percent of all cases and 10 percent of all payments: 1970 Recipient Fraud Incidence Study conducted by the Fraud Review Panel for the California State Department of Social Welfare (now Social Services) 62 percent of a sample of 215 cases subjected to a fraud investigation: 1993 Child-Only Fraud Pilot Project prepared by the Orange County Social Services Agency for the California Department of Social Services. "AFDC Fraud is a billion dollar problem: 1987 State Investigation of Fraud in the Aid to Families with Dependent Children prepared by the Department of Health and Human Services, Office of the Inspector General. 100 percent of a sample of 25 cases subjected to in-depth study had unreported income: 1990 The Real Welfare Problem prepared by the Centre for Urban Affairs and Policy Research, Northwestern University. Taken from California's Fraud Strategic Plan, June, 1994

Even at the lowest fraud estimate, the loss to California is staggering.

Another way of looking at the amount of fraud is to examine public information provided by the American federal office responsible for the Food Stamp program (which provides benefits to over 25 million Americans). The following table demonstrates the variability of reported fraud and error rates, depending on the fraud and error control programs in place in the particular jurisdiction. The average known fraud rate in the United States is 4.4% of total program disbursements. However, this ranges from a high of 13.62% in Massachussets, to a low of 1.07% in Washington.
 

FRAUD AND ERROR  RATES - 1992 UNITED STATES FOOD STAMP PROGRAM
	Fraud Claims  as % of total $	Client Error Claims  as % of total $	Agency Error Claims  as % of total $
Alabama	6.24	6.04	2.28
Alaska	3.20	2.51	2.87
Arizona	1.43	4.51	4.61
Arkansas	4.86	4.97	1.16
California	2.63	9.44	5.60
Colorado	7.93	6.10	3.05
Connecticut	4.17	1.54	0.22
Delaware	5.44	4.40	1.64
District of Columbia	4.48	8.20	0.83
Florida	1.70	3.37	2.92
Georgia	7.95	3.82	6.77
Hawaii	6.86	9.02	2.57
Idaho	4.30	2.60	3.43
Illinois	6.13	31.43	4.44
Indiana	2.95	7.17	1.91
Iowa	2.82	8.78	2.71
Kansas	4.22	6.32	2.40
Kentucky	2.93	1.53	1.29
Lousiana	2.59	3.98	0.91
Maine	2.03	4.03	1.08
Maryland	1.20	7.91	1.20
Massachusetts	13.62	7.96	1.70
Michigan	9.20	7.67	2.40
Minnesota	6.64	20.75	6.85
Mississippi	6.14	2.86	2.46
Missouri	5.79	6.81	2.10
Montana	2.30	2.72	0.24
Nebraska	2.84	10.35	3.69
Nevada	2.68	4.11	0.68
New Jersey	10.47	27.30	1.86
New York	2.08	10.77	1.44
New Hampshire	5.66	7.75	3.05
New Mexico	1.73	4.95	10.63
North Carolina	3.40	2.11	1.21
North Dakota	2.37	5.55	1.91
Ohio	7.75	7.07	2.43
Oklahoma	3.21	7.04	6.08
Oregon	4.00	12.10	2.53
Pennsylvania	5.59	24.17	2.25
Rhode Island	2.33	2.40	0.30
South Dakota	1.98	3.45	0.59
South Carolina	5.21	6.72	2.76
Tennessee	5.62	3.08	2.91
Texas	3.11	6.26	1.93
Utah	3.77	8.52	1.31
Vermont	3.37	2.63	2.50
Virginia	2.63	3.90	1.73
Washington	1.07	10.10	2.33
West Virginia	2.80	3.90	5.50
Wisconsin	2.10	15.20	3.69
Wyoming	5.59	11.78	1.97
UNITED STATES	4.40	9.50	2.85

There are some anomalies in the data on client error. Four states (Illinois, Minnesota, New Jersey and Pennsylvania) show very high rates of client error, above 20%. It is possible that the methodology used in these states for measurement purposes is more comparable to an eligibility review process method than a file review mechanism. Even including these very high results, the average error across all states is 9.5%, ranging from a low of 1.53% in Kentucky to 31.43% in Illinois. The average agency error rate in the United States Food Stamp Program is 2.85%.

These data raise a wealth of questions about both interpretation of the data and the mix of programs in each state which contribute to the absolute magnitude of the rates. It is beyond the scope of this proposal to provide that level of detail for all 51 states. Although the American programs are not directly comparable to the variety of Canadian programs, it is likely that the Canadian experience falls somewhere within this range, although it is possible that Canadian rates are actually higher, since the American programs generally have more detection and prevention programs in place.

For the purposes of estimating the cost-benefits attributable to this proposal, the value of overpayments attributable to fraud in Canadian income security programs is estimated to be 4%, and the value of overpayments attributable to client errors is estimated to be 5%. This implies that the magnitude of the fraud/error "problem" can be calculated, as a percentage of total expenditures, of $39.2 billion, on income security programs in Canada, as approximately $3.53 billion.

However, until Canadian jurisdictions get serious about measuring the amount of fraud and error, the question of the total amount of fraud in the system will likely remain unanswered in any definitive way. Experience, in other jurisdictions, clearly demonstrates that the more you look for fraud, the more fraud you find. The public and the media, unfortunately, often treat high levels of uncovered fraud as a public scandal rather than as a demonstration of active fiscal management of the public purse. Continued public education may be the only solution to this dilemma.
 

Current Solutions

1. DETERRENCE - Deterrence activities and controls are intended to cause clients, staff or vendors to choose not to commit fraud or to desire to minimize errors. Given that only the individual "deterred" has any knowledge of whether a certain activity caused them to choose the desired course, this area is practically impossible to quantify. Field practitioners vouch for the efficacy of deterrence activities, but they do not lend themselves to any form of cost-benefit analysis. An example is the fingerprint imaging process, as implemented in some counties in California and New York, which deters clients from applying twice under different names.

 
2. PREVENTION - Prevention activities and controls are designed to ensure that, even if an individual intends to commit fraud, the individual would be unable to follow through on their intentions because of some intervening circumstance. Fraud and error prevention occurs when computer systems, procedures or individual knowledge intervenes in the payment process to correct or prevent the fraud or error before an overpayment actually occurs.

 
3. ACTIVE DETECTION - Active detection programs are predicated on a jurisdiction's belief that fraud and error are an expensive reality. These programs, through an assessment of ways that an individual could make an error or commit a fraud, involve a conscious effort on the part of the jurisdiction to "go out looking" for incidences of overpayment. These are often referred to as "early detection" or "front-end" programs. Public awareness of these programs also has both deterrent and preventive effects.

 
4. PASSIVE DETECTION - Passive detection programs depend on regular internal controls or on someone or something outside the organization to identify a potential case of overpayment. Once that identification of potential fraud or error has been made, however, post-fraud investigation is required to confirm and correct the overpayment.

 
5. COLLECTION - These processes and activities relate to "righting the wrong", and may involve "outsiders" to the program, such as police, courts, personnel or finance departments.

There are a surprisingly large number of jurisdictions, including the federal government programs who make no attempt to measure the degree of error or fraud that they have within their programs. In these cases, neither the program department itself, nor the related Auditor-General's department, have identified the potential of cost-control through reduction in fraud and error levels as a priority. These jurisdictions also invariably have few, if any, programs for controlling or reducing the level of fraud and error. It is highly unlikely that these jurisdictions are operating an error-free or fraud-free program. However, if fraud and error levels are not routinely measured and identified, there is no public or internal pressure to "solve the problem".

The public sensitivity towards levels of fraud and error in Income Security programs, coupled with an ever tightening fiscal environment, make it critical that jurisdictions be able to demonstrate internally and publicly the integrity of their income security programs. Therefore, a typology of fraud and error control programs was designed (C. A. MacDonald & Associates. Typology of Fraud and Error Programs. August, 1993) to assist jurisdictions which recognize the need and are attempting to improve their levels of fraud and error in their income security programs.

The concepts presented in this proposal have been selected from this typology and are those which hold promise of a high cost-benefit ratio, and have not been implemented to any degree in Canada.

The Missing Pieces

In Canada, not only is there little apparent leadership in this area, current federal legislation actively works against detection and prevention of fraud and abuse in the various income support programs. The following main themes would establish a strong federal leadership role and a climate for appropriate fraud and abuse control in all Canadian income support programs.

1. Establish a priority for fraud prevention and detection - Through the development of new legislation and/or adapting current legislation as well as other techniques, identify a requirement for federal and provincial income support programs to actively prevent and detect fraud and abuse. Requirements for measurement and reduction of administrative and client error could also be included. (Only a few provinces, Revenue Canada and the UI benefits administration are currently active in fraud prevention).

 
2. Require the use of a common client identifier - The Social Insurance Number (SIN) is the obvious choice for a client identifier. However, current legislation on the use of the SIN prevents some programs (most noticeably provincial income support programs) from requiring applicants and spouses to disclose or obtain a SIN number as a condition of eligibility. Cross-program and jurisdictional comparisons of the client base cannot occur efficiently and with accuracy without a common identifier.

1. Improve access to federal tax information - The confidentiality provisions in the current income tax system prevent income support programs from accessing data to identify current income sources and amounts except in individual cases and normally only under court order or with the express permission of the client. Routine data matching is required to ensure early detection and prevention rather than the current system of using the income tax information in specific cases to confirm the existence of fraud found through other methods.

 
2. Facilitate routine cross-program and cross-jurisdictional data matching - As the federal government continues work on a major program to redesign the Income Security information systems, a key objective should be to ensure the feasibility of determining from which other programs an applicant or current client is receiving assistance. In effect, this means the creation of a client index for all federal programs (including federally funded assistance programs for natives on reserve) which could also be accessed by provincial and municipal administrations.

 
3. Allow administrative disqualification for assistance - In Canada, under current CAP legislation, benefits are based on need only, and past or current fraudulent behaviour has no impact on current and future eligibility. In the United States, one identified instance of fraud or program abuse can result in a six-month disqualification, two instances can result in one year of disqualification, and more than two instances of fraud can result in a lifetime disqualification from receiving program benefits. If this were implemented in Canada, information on fraud and abuse disqualifications should be shared between programs and jurisdictions.

 
4. Expand the use of the tax system to recover income support overpayments - Currently, the record on recovering overpayments in provincial income security programs, particularly on closed files, is dismal. By redirecting funds to the program to which funds are owing, the tax system is used to recover UIC overpayments from refunds that would otherwise go to the client. Expanding the use of the tax system to include overpayments in other federal, provincial and municipal income support programs would result in millions of dollars being recouped annually. The process is referred to a "tax-intercept program" in the United States and is federally mandated and widely used.

 
5. Encourage the establishment of early fraud detection and prevention systems - Over 30 American states, at the direction of the federal government, have established early detection systems which identify probable cases of fraud before assistance is granted by integrating fraud detection and prevention into the intake process. Cases are referred for investigation if certain criteria are met (such as lack of or forged identification, needs consistently exceeding income, etc.). In Florida, California, and Texas, approximately 70% of the cases referred for investigation result in assistance being denied or assistance being granted at a reduced rate.

The Federal Role

• legislation requiring all States to identify alleged fraud and develop procedures for referring those cases to law enforcement officials.

 
• cost-sharing the administrative cost of fraud and error control programs.

 
• for programs funded totally by the federal government, but administered by the states, such as Food Stamps, a "savings-sharing" formula has been established which permits a proportion of savings to be retained by the states as an incentive to prevent and detect program abuse.

 
• for programs jointly funded at both the federal and state/county levels, such as Aids to Families with Dependent Children (AFDC), providing direction to the states regarding which fraud control techniques should be implemented at the state/county level.

 
• on-going evaluation of federal, state and county fraud initiatives coupled with dissemination of information on effective techniques for fraud control.

 
• mandating and facilitating state and county access to federal data banks such as the Internal Revenue Service and Social Security benefits information systems.

 
• partially supporting the costs of national fraud conferences.

Without a strong federal presence, and a national priority placed on fraud detection and prevention, Canada will continue to lose billions of dollars to client fraud and error.

Leadership Components

There are a number of components to federal leadership in this area, each of which will be discussed separately below.

Legislative Framework

With the imminent demise of the Canada Assistance Plan, the federal government will very rapidly be developing legislation to replace this act and to support the block funding approach to social services with the provinces. It is probable that this legislation will contain some form of national standards, similar to the Canada Health Act. Discussion with federal officials would indicate that the federal government will continue to require that assistance be granted without regard to residency requirements in the province of issue. This will support the free flow of income security clients across the country, enabling clients to move to where opportunities for employment are, without artificial barriers to movement.

A standard which is currently in place in the Canada Assistance Plan is that need is the only determinant of eligibility. This means that provinces cannot place conditions on individuals receiving benefits, or implement certain forms of penalties on clients who have defrauded the system of money. It is critical that this standard be removed, and replaced with a more general standard which makes explicit that the receipt of income security benefits in Canada is a privilege, not a right, and that appropriate behaviour and certain responsibilities are expected of recipients in return, both in terms of honesty and movement towards independence as rapidly as possible. Then, if clients do not live up to these responsibilities, certain penalties can be applied by the provinces or municipalities, which may involve reduction in, termination of, and disqualification for future benefits. This would be equally effective in encouraging clients to move towards employment and as retribution for illegal activity.

There should be a clear requirement in the new legislation for provinces and municipalities delivering income security programs to measure the degree of fraud and error in those programs and to implement programs which would reduce the level of overpayment attributable to client fraud and error. There are some provinces who freely admit off the record that measurement is not occurring because if the public knew the amount of overpayments, it would become a public political scandal in the province. In fact, one provincial income security manager typified this approach as "living in a state of blissful ignorance".

In at least two provinces, British Columbia and Alberta, the result of the public becoming aware of the magnitude of the fraud and error problem has been the on-going development and implementation of two of strongest fraud and error control programs in the country. In Alberta, the Auditor-General blew the whistle in 1988-89 identifying $35 million in overpayments in the previous year. In British Columbia, in 1993, a confidential report identifying problems with internal controls was leaked to the public. In both cases, the public outcry resulted in a serious attack on fraud and error. Few other provinces even approach the kind of comprehensive programs which are now in place in these two jurisdictions. It is unfortunate that sometimes the only way to change government priorities is to have such a public outcry! Certainly, the $3.53 billion estimated earlier on page 8 would seem to be sufficient cause for taxpayers to insist on a more effective approach to controlling this waste of both federal and provincial government resources especially in light of recently announced cutbacks in most other program areas..

New legislation should allow for the Minister of Human Resources Development Canada to mandate fraud and error control programs, establish effective initiatives for the provinces to implement in order to retain federal funding for programs. This is similar to the approach taken by the United States federal government in suggesting all other states implement the "Orange County" approach to fraud control. This approach is described in Chapter 8, Early Detection and Prevention.

Lastly, any new legislation should contain a statement about confidentiality and privacy with regard to client information. In the legislation, trained and regulated investigators should be given the same rights and powers as law enforcement officials in terms of access to information, regardless of privacy legislation. This is the approach taken in Ontario in their privacy legislation, where fraud investigators are considered to be law enforcement officials.

Establishment of Principles

Such new legislation may also be an appropriate mechanism for establishing a set of principles for controlling fraud and error. These principles should contain at least the following elements:

UNOBTRUSIVE       Fraud and error control programs should be as unobtrusive as possible to
                                    individual clients unless strong suspicion or specific evidence of wrong-doing
                                    has been established.

TARGETED              Fraud and error control programs should be targeted towards that segment
                                    of the client population which are determined to present the highest risk of
                                    committing fraud or error.

CONFIDENTIAL     Information which comes to light during an investigation of an individual must
                                    remain confidential until or unless it is used in a court of law or a
                                    quasi-judicial setting such as an appeal hearing.

BENEFIT OF DOUBT Any individual under investigation for potential wrong-doing shall be given
                                    the benefit of the doubt in any circumstance for which the evidence is
                                    questionable.

PRESUMPTION OF Any individual client shall be presumed to be honest unless
HONESTY                 specific evidence exists which demonstrates that this is untrue. In other
                                    words, income security programs should be designed on the belief that the
                                    vast majority of client are honest.

Other principles may also be required to set the appropriate "tone" for a national focus on fraud and error, but these at least are critical to establish the intent of the fraud and error control initiatives.

Facilitation of Fraud Control

For any individual provincial, municipal or territorial jurisdiction in Canada, the number of specialized fraud staff is very small as a proportion of the overall staff delivering income security. As a result, there is little specialized training for these individuals, and in fact, because of these small numbers, these individuals tend to get ignored as part of the overall organization. Often they are seen as a nasty, unwelcome reminder that not all clients are honest, and tensions often exist between these staff and front-line workers.

The federal government should support the setting up of a Canadian branch of the American United Council on Welfare Fraud (UCOWF) and should encourage provincial and territorial jurisdictions to purchase memberships ($25 per year per membership) for all fraud and eligibility investigators and other staff interested in the area of fraud and error control. This would provide an extremely cost-effective mechanism to obtain access to national and regional fraud conferences, information sharing and exchange with southern colleagues, a certificate program for training fraud investigators, and access to a wealth of training materials. Some Canadian investigators are already members of the UCOWF organization, but Canada is not identified as a separate region, with a Canadian Director on the Executive Board. Although many fraud related issues and investigative techniques are the same in both countries, Canadian programs are different, and this would be recognized and reflected if a formal Canadian branch of the organization were established. Then, individual jurisdictions could depend on the availability of fraud and error control staff training without incurring major additional costs. Selected Canadian members of UCOWF have been approached to attend a meeting in Atlanta, Georgia, at the next annual UCOWF national convention to discuss the establishment of a Canadian branch or region of UCOWF.

Funding for Fraud Control

Although not specifically required by the Canada Assistance Plan, fraud and error control programs are cost-shareable. In addition, any savings achieved by the provinces is also shared between the province and the federal government. However, it was up to an individual province or municipality to decide what and how much was expended in fraud control programs. Under the new block funding scheme, the incentive to save money through fraud and error control is even higher than under the previous cost-sharing arrangement. For example, if a province implements a $1 million program, which saves $4 million, under CAP the province would have only paid $500,000 of the cost and would have retained $2 million of the savings, leaving the province with a net gain of $1,500,000. Under the block funding scheme, for the same program, the province would pay the full $1,000,000 cost, but would retain all the savings, giving a $3,000,000 net gain, so for an extra $500,000 expenditure, an additional $1,500,000 would be gained. The relative cost-benefit remains the same but the absolute cost savings increases significantly.

For provinces without any significant resources in fraud and error control, a new federal mandate that provinces must enter into fraud control programs would result in additional costs to the province. However, this is not a true cost, since all evidence points to the fact that whatever the cost, the savings will be significantly higher.

Establishment of a National Office of Fraud and Error Control Initiatives

The fraud and error control programs proposed in this document all require significant federal-provincial and inter-provincial cooperation, involving all provinces and territories and at least two federal departments (several divisions of Human Resources Development Canada plus Revenue Canada). This inter-jurisdictional cooperation would be facilitated if a national office of fraud and error control were to be established.

The Terms of Reference for this office would include: drafting of those sections of the new legislation supporting the fraud strategies: overseeing development of compatible information systems; working with national poverty lobby groups to foster commitment and understanding of fraud and error control programs; coordinating the development of federal-provincial data sharing agreements; evaluating new approaches to fraud and error control; and establishing and maintaining a national income security client index (see Early Detection and Prevention).

Implementation Strategies

1. Establishing fraud control as a key item on the next First Ministers Conference, followed by a joint communique identifying the priority of, and action plan for, a concerted national thrust for attacking fraud and error in Canadian income security programs.

 
2. Ensuring that the area of fraud and error control is an agenda item for the next federal-provincial-territorial meetings of the Ministers of Social Services, the Deputy Ministers of Social Services, and the Ministers and Deputy Ministers of Finance.

 
3. Establishing a special meeting of federal-provincial-territorial Directors of Income Security to discuss cooperative approaches to dealing with fraud and error.

1. Establish the National Office of Fraud and Error control

 
2. In consultation with provinces and municipalities, develop a national workplan for implementation of the various initiatives.

 

Implementation Costs

The majority of the other costs would be less direct - legislation will be required whether or nor fraud and error control sections are incorporated, senior intergovernmental meetings will occur whether or not fraud is on the agenda. A special meeting of Executive Directors would cost approximately $2,000 per provincial delegate.

These costs are negligible compared to the immense benefits which a national attack on fraud and error would engender, including increased public support for income security programs and increased public confidence in program administration.

Issues

Because of the lack of measurement of fraud and error rates and losses in Canadian income security programs, the area of fraud and error control has not been seen as a high priority. And, in fact, politicians do not like reporting high levels of fraud and error in their programs. However, a well-designed communications strategy which focuses on the positives of high reported fraud rates (demonstrating active detection, prevention, and prosecution of fraud) can as equally well demonstrate fiscal stewardship as fiscal malfeasance.

If such an initiative were announced, without a doubt there would be concerns raised by client advocacy groups that this represented "picking on" the poor, who could not defend themselves. The fact that fraud and error is probably higher in the income tax system might even be used as a rationale for not addressing fraud in income security programs. Advocacy groups can be invited to participate in the planning and design of programs as a way of engendering understanding of the magnitude of the problem and the need to protect programs for the truly needy. A communications strategy which emphasized that fraud is committed by a very small proportion of the overall caseload, and that this small group gives a "bad name" to all clients of the system may ameliorate this back-lash. However, the "silent majority" will be strongly in support of the initiative. Not only does everybody know someone who is supported by a Canadian income security program, most people know of someone who is cheating the system! The public is far more willing to concede the magnitude of the problem than are politicians.

These issues can primarily be addressed by publicizing the amount of money which could be saved through fraud and error control initiatives. For every dollar saved by not overpaying clients of income security programs, one less dollar will have to be cut from the federal and provincial budgets in another area to address the deficit. Clearly, the communications strategy should be to tout these fiscally responsive initiatives as directly addressing the reduction of the deficit and the protection of income security programs for the needy.

Identifying Clients

A different, but related, problem occurs when the identification material (cards, documents, etc.) is absolutely valid, but is being carried by a person other than the one to whom the identification belongs. This could occur for several reasons:

• the identification has been lost, and found by another;
• the identification has been stolen;
• the identification has been inadvertently issued to a person not the actual owner of the identification;
• the identification has been sold, with the original owner perhaps paying a small fee for replacement identification; or
• the identification has been lent to another by the owner, who will expect its return after the current carrier of the identification has obtained the goods or services for which the identification assisted in establishing eligibility.

• Clients with multiple sets of ID obtaining benefits fraudulently under several identities;
• Client with no identification obtaining benefits on a temporary basis prior to confirmation of identity;
• The lack of, or requirement to use, any common identifier (i.e., SIN) across programs and jurisdictions means that data linkages may relay on a name/birthdate as an identifier, which often leads to missed or inaccurate matches.

Each jurisdiction facing these issues could address the problem independently, developing improved client identification mechanisms. If several provinces and territories or individual federal income security programs began to independently develop and implement various solutions to the question of fraudulent identification, the various levels of government run the risk of significant additional costs resulting from duplication of administration and development costs. In addition, although the resulting potentially incompatible systems may solve each jurisdiction's problem, inter-jurisdictional issues may not be addressed. The problem of fraud resulting from inadequate client identification is an inter-jurisdictional issue, requiring consistent solutions and cooperation between the federal provincial and municipal governments to gain maximum results from efforts to minimize duplication and unnecessary costs.

The Current Situation

In the United States, all clients must present their Social Security Number in order to apply for or receive any income security benefits. In Canada, however, even though a comparable unique identifying number exists, the Social Insurance Number (SIN), federal legislation restricts the use of this number to specific labour related situations. This is very much a "Catch-22" for provincial governments. Revenue Canada requires provinces to produce a T-4 slip for all clients who receive benefits from income security programs, linked to the client's social insurance number. Yet, provincial income security programs can request, but cannot insist on a client producing a SIN number, since the use of the number is legally restricted.

Most federal income security program require the use of a SIN number when applying for assistance. Some provinces, such as Newfoundland, insist on the production of a SIN card in spite of the restrictive legislation on use of the SIN. Ontario, faced with a recent appeal by a client who did not wish to produce his SIN number, is much less insistent (the client won the appeal and did not produce his SIN card). In addition, the current "need is the only basis for eligibility" standard in the current CAP legislation means that benefits cannot be denied on the grounds of not producing or obtaining a SIN number.

Several provinces and municipalities, including British Columbia and Metro Toronto are attempting to strengthen their client identification requirements. However, without federal leadership through changing the SIN legislation to allow income security programs to require the SIN number as client identification for all clients (head-of household, spouse and teenage children), these initiatives may solve each jurisdiction's problems, but will not assist in resolving the inter-jurisdictional and inter-program issues related to data matching mentioned earlier.

Again, federal leadership is required to sanction and facilitate the use of a common client identifier which all income security programs can use. This common identifier is a necessary underpinning for the remaining initiatives in this proposal.

The Wages of SIN

Two changes at the federal level are required:

1. Revising the section of the legislation on the use of the Social Insurance Number to extend use of the SIN to include provincial/territorial/municipal income security programs; and

 
2. In any new set of national standards governing provincial/territorial/municipal income security programs, ensure that lack of a SIN number (or production of a SIN within a reasonable time) for all working-age dependents in the household is valid grounds for denying benefits.

Once all clients and other working-age recipients were identified by a Social Insurance Number, access to all federal income-related files would be more effective. This would of course not help in the area of discovering non-reported sources of income such as income earned in the underground economy, through illegal pursuits such as prostitution, gambling or drug-dealing, or from many forms of self-employment income. However, over 70% of all income in Canada is in the form of wages and salaries, where the employer is deducting both income tax and, often, UIC premiums at source. In addition, significant portions of the remaining 30% are reported to Revenue Canada directly by tax-filers. This information is all accessible via linkage through the Social Insurance Number.

Social Insurance Number Administration

1. Linkage between temporary SIN numbers for new immigrants and a later "permanent" number. Metro Toronto Social Services is currently pursuing this issue with the federal government on a pilot basis.

 
2. Improved federal record keeping and control-oriented processes for replacement of lost and stolen cards. Currently, the process seems to be a simple re-issue of a new card with the original SIN number, possibly resulting in two individuals with access to the same number--the original owner, and the person who fraudulently obtained a replacement card or stole or found the original card.

 
3. Linkage of SIN numbers to the proposed National Death registry, currently being explored by Statistics Canada in conjunction with provincial Vital Statistics departments. SIN numbers belonging to deceased individuals must be permanently retired and not re-issued. Individuals using SINs belonging to deceased persons would only be doing so for illegal purposes.

In the United States, a commercial computer program is available, to income security administrations for approximately $30, which "unscrambles" any individual Social Security Number, and identifies the date and place of issuance. The Canadian SIN for any individual also identifies where and when the card was issued. With this information available through a simple computer program to front-line intake workers in federal, provincial or municipal income security offices, the intake worker would have a major new tool to assist in identifying stolen or forged cards. For example, if a 40-year old male, reporting 20 years of employment history applies for assistance using a SIN which was issued two years ago, a whole new line of exploration regarding the individual's eligibility for assistance is uncovered.

Direct Benefits

One side-effect of improved administration of the Social Insurance Number with regard to lost and stolen SINs would be the general benefits to all Canadians by affording some additional protection from theft of identity with all of the ensuing tax problems, and criminal record problems presently caused by having the activities and behaviour of the individual who has stolen an identity attributed to the wrong person.

Implementation Strategies

The passage of the legislation would require federal, provincial and municipal income security agencies to contact all clients (and other working-age adults such as spouses, teenage children) who did not have SIN numbers on file. Depending on the province or municipality, this could range from 0% to 15% of the total caseload. Clients would be informed of the legislative change, the reason for requiring a SIN (to link data files with other programs and jurisdictions), and given a specific time period in which the SIN would have to be reported. Clients who still refused to provide, or obtain, a SIN would be terminated from assistance. This refusal to provide a SIN would be considered prima facie evidence of fraud or abuse and may warrant a fraud investigation to determine if overpayments had been occurring and could be recovered, or whether prosecution was warranted.

Implementation Costs

The cost of improving administration of the Social Insurance Number system is somewhat peripheral to the proposal outlined in this document, and has not been addressed directly, but may include the costs of establishing new processes and installation of new computer systems.

Issues

It is only those who had neglected to report income from another program or jurisdiction or from earned or unearned income sources who would suffer from an "invasion of privacy". And, since these individuals are clearly breaking the law, it is strongly suggested that these cases of "invasion of privacy" would be supported by politicians and the general public.

Description of Problem

When the question of establishing data matches is raised, the immediate public response is that "Big Brother" is watching, and that with developments in computer technology, privacy and confidentiality require stringent legal protection. The fact of the matter is, "Big Brother" has been watching for some time. through a variety of mechanisms. There are several other types of data linkage which already occur in government income security programs. The following listing identifies these types of current data exchanges.

1. Downloading of basic client data from a central data base - most federal income security programs, and provinces/municipalities have established a central data base of client data which can be accessed when a client reopens his/her application for assistance. The main data base will download to the remote terminal the basic client data on the data base, including the name, birth date, names of dependents, and other relevant information on client eligibility. This reduces administrative workload attributable to the re-entry of client data; it prevents clerical errors (such as the misspelling of the individual's name); and allows for updating of variable information on the main data base, such as address, change of name (through marriage for example), or adding dependents to a case file.

 
2. Identity cross-checking for initial registrations - Some provincial and federal programs have automated internal cross-checking of client data against other registrants on the file. For example, some provinces check current client addresses against the addresses of other registrants to identify whether there are other recipients listed at the same address. Similarly, checks for duplicate name/birthdays, SIN, or duplicate registration of dependents are also performed.

 
3. Legislatively required data linkage - There are information links which are identified by government policy or legislation (for example, the free provision of health care services to income security clients is common across Canada). In these cases, the data linkage is often one-directional, with the "base" program, income assistance, automatically registering the individual for services on the health care data base. These linkages minimize administrative duplication, expedite service to individuals, and minimize intrusion into the lives of individuals.

1. Access to information required by program legislation - Some programs require personal or financial information in order to ensure eligibility for benefits. For example, eligibility for the Alberta seniors benefit is tied to eligibility for the Guaranteed Income Supplement. Federal data tapes are transmitted to Alberta to allow calculation of eligibility for the provincial program. In addition, most provincial income security programs require information on assets owned by individuals applying for assistance. In Alberta, for example, on-line access to Vehicle Registration files to confirm assets owned by individuals has been established as part of the Eligibility and Benefit Verification process referenced earlier (page 4). Again, this linkage is one-directional - Income Security can access Vehicle Registration, Vehicle Registration cannot access Income Security files. The rationale for this type of linkage is based on the eligibility requirements of each program.

 
2. Improvement in Individual Service - The Pharmacare Network project in British Columbia is an excellent example of data linkage which improves service to individuals. With the linking of prescription records across the province, individuals can be assured that their prescriptions, possibly obtained from several doctors and pharmacies, are safe if taken in combination. This system is also linked to the Pharmacare billing system to clarify in a non-judgemental or non-stigmatizing way what proportion of the cost of the prescription is payable by the individual. From the perspective of the individual pharmacist, this system increases the privacy of individuals. Since the pharmacist cannot access billing information directly, the pharmacist has no way of knowing which government program is covering the cost of the prescription, whether the individual is receiving a seniors discount, has reached the deductible cost of pharmaceutical, or is receiving Income Security.

 
3. Improvement in Vendor Service - Data linkage can also improve payment of government accounts to individual vendors. Again, the BC Pharmacare Network Project enables payments to be made to pharmacists within a week of the prescription being filled, a major improvement in speed of payment.

 
4. Identification of Fraud - Data linkages already occur which assist programs in identifying fraudulent access to programs. An example is the recent interprovincial agreement between British Columbia, Alberta, and Saskatchewan, with other provinces exploring joining the network, to identify individuals receiving assistance in more than one province, using the SIN number as the primary identifier where available, and name/birthdate where it is not. Similarly, several provinces, including Alberta, Ontario, and New Brunswick are experimenting with on-line linkage to UIC files, to ensure that new applicants are not already receiving undeclared UIC payments. These data-matching processes do not identify any individual unless there are strong grounds to suspect that illegal activity is occurring.

 

Current methods of linking data between programs include transmission of paper documentation (forms), one-way transmission of electronic data, case-by-case requests for information, tape-to-tape data matches and on-line access. The effectiveness of these mechanisms suffer from several drawbacks:

1. Lack of a common identifier - current data linkages often occur on the basis of a combination of SIN and name/birth date/address matching. The lack of any consistent common identifier means that these matches are very difficult and result in having to sort through large numbers of potential duplicate registrations due to an inability to easily confirm an actual match of client files.

 
2. Incompatible file formats - Using name/address/birth date matching is only possible if the electronic file formats are similar. Unless this is the case, each individual file match would require a lot of systems work to rewrite the files to put them in compatible formats. This degree of data manipulation makes these matches an expensive, time consuming and inexact process.

 
3. Too much information is shared - A department requesting case-by-case access to another program's files often receives additional extraneous information which was not required or allowed by their legislation. This compromises the privacy of individuals.

 
4. Dependent on clerical accuracy - The accuracy of data linkage can often be compromised, either by matching an individual with the wrong file, or not finding a match at all due to clerical errors in entering the name or birth date.

 
5. Administratively time consuming - Obtaining information on a client-by-client basis can be administratively time consuming and is such an inefficient use of time that it is often not done by harried caseworkers.

Technology exists which would use the new common client identifier to build on existing program automated file structures to link information where supported by legislation and policy. The rapidly developing field of local and wide area PC networks, applications of CD-ROM technology for storage of and rapid access to huge amounts of information, and Windows-based SQL inquiry processes, to name but a few of the new technical developments, ensures that the technology exists to do anything that is required from a practical perspective. Computer security features, using password and access code control provides a far more secure environment for individual information than the current systems which are still largely paper-based. However, the detailed technical aspects of mechanisms to link files are beyond the scope of this proposal.

The real issues are not of a technical nature.

The American Experience

Congress implemented Project Match, a federal project to match Aid to Families with Dependent Children (AFDC) and Medicaid files.

1979- Congress passed a law requiring State agencies to use income information (wage matching) in determining eligibility for AFDC. In 1983, Congress added that requirement for the Food Stamp program as well.

Congress introduced the Family Assistance Management Information System (FAMIS) to promote computerized information systems for the States.

Congress passed the Agriculture and Food Act and the Omnibus Reconciliation Act, requiring the States to match social security numbers of Food Stamp recipients with computerized wage data.

These initiatives led towards a far more encompassing data matching undertaking, the Income and Eligibility Verification System.

The Income and Eligibility Verification System (IEVS) was established by Congress under the 1984 Deficit Reduction Act to reduce errors in determining eligibility and benefit levels in the Food Stamp, Aid to Families with Dependent Children (AFDC) and Medicaid programs.

Implementing regulations require State welfare agencies to compare income reported by program applicants and recipients with income from several data sources:

1. Internal Revenue Service (IRS) data on interest, dividends, and other types of unearned income;

 
2. Social Security Administration data for Retirement, Survivors, and Disability Insurance benefits, Supplemental Security Insurance benefits, and annual earnings; and

 
3. State quarterly wage reports and unemployment insurance benefits.

Since the federal legislation mandating computer matches was passed in 1984, individual States have made various degrees of progress towards implementation of the full range of data matching. As well, some States are conducting matches in addition to those mandated and some States are requesting access to additional Federal data banks such as Veterans Administration benefits, Federal employee salaries and pensions, U.S. Savings Bond holdings, and State death records maintained by the Social Security Administration.

There are two aspects to IEVS:

1. Applicant System - client identifying information, i.e., name, SSN, sex, DOB, is submitted via computer by eligibility or intake workers once an application for benefits is completed. Information from the data sources listed above is usually available within 3 - 5 working days for those States with this system, and can be used to confirm eligibility or to deny benefits as appropriate. However, Counties cannot unduly delay initial assistance while awaiting match results. As well, prior to submitting client identifying information for data matching, Counties must inform clients, in writing that the match will be conducted.

 
2. Recipient System - this system provides information from the matching of active clients with a variety of other systems at various times of the year, depending on the match type. Various threshold levels are used to ensure that only information that could impact eligibility or benefit levels is sent to the worker for follow up.

• Payment Verification System (monthly) - provides information on recipients that receive, or will receive Retirement Survivors Disability Insurance (RSDI), State Unemployment Insurance (UI), or State Disability Insurance (DI).

 
• Integrated Fraud Detection System (quarterly) - identifies unreported wages and duplicate aid for AFDC, Food Stamps and Supplemental Security Income/State Supplementary Program (SSI/SSP).

 
• State Franchise Tax Board Asset Match (annual) - a state-wide automated system that matches the welfare recipient file against FTB's interest and dividend file. FTB's file contains information from financial/investment institutions based in California or having offices in California. Only those cases with interest and dividend income which exceed specified levels are reported.

 
• IRS Asset Match System (annual) - an automated system that matches the welfare file against the IRS Unearned Income File.

 
• Beneficiary Earnings Exchange Record (monthly) - BEER reports contain wage information from the Social Security Administration (SSA) which includes self-employment, out-of-state wages, military wages, and Federal wages.

The Office of the Inspector General study found that:

1. 88% of the IEVS computer matching systems are used by the Food Stamp and AFDC programs,

 
2. 69% are used by the Food Stamp and Medicaid programs, and

 
3. 26% are used by the Food Stamp, AFDC and Medicaid programs.

The state agencies responsible for administering assistance programs have been developing increasingly sophisticated computer systems to support program operations, particularly for using computer matching activities to corroborate client information or to detect discrepancies in information. Federal funding incentives encourage the development of new systems and concerns about initial client eligibility have prompted agencies to examine the increased use of up-front matching processes as a preventative cost-avoidance measure.

Applicant matching (up-front matching) has been shown to improve the morale of eligibility workers by helping them establish the integrity of case determination. As well, the availability of information about individual applicants was seen to improve the delivery of services to clients. The information allows eligibility workers to more easily meet case disposition deadlines and documentation requirements. The computerized databases can be a way to save applicant's time by relieving the client of the need to locate the required documentation/verification themselves.

Overall, increasing concern for the detection and prevention of fraud, waste, abuse and error in Government programs has stimulated the development of techniques using information technology to detect and prevent these problems.

Current Canadian Strategies

Provincial data links with other provincial programs focus on links to Workers' Compensation Boards, Maintenance Enforcement data bases, Land Titles, Motor Vehicles and Student Finance Boards. Newfoundland has also established links with the public service payroll files and public service pension plans. An interesting linkage which Metro Toronto is pursuing is with the Insurance Crime Prevention Bureau to purchase client information on income received from insurance companies to recompense individual who lose regular employment income due to being off work as a result of a motor vehicle accident.

(X=data sharing agreement in place; P=On-line pilot underway; E=Exploring/negotiating an agreement):

	U.I.C.	OAS/GIS	C.P.P.	Revenue  Canada	Immigration Canada	Inter-provincial Matches	Provincial  Program Matches
British Columbia	X					X	X
Alberta	X (P)	X	E	E		X	X
Saskatchewan	X		E	E		X	X
Manitoba	X					X
Ontario (provincial)	X		E		X	E	X
Metro Toronto	X (P)				X	X
New Brunswick	X (P)	E		E		E	X
Prince Edward Island	X	X	X				X
Newfoundland	X						X
North West Territories	X
Yukon

In all cases of federal-provincial agreement except the UIC on-line pilot, these matches are tape-to-tape matches. In addition, even without a formal data sharing agreement, provinces have access to most federal files on a case-by-case basis. That is, after a province has determined that an individual has committed fraud, information to confirm that fraud can be obtained from the relevant federal file, usually via court order. However, this method does not assist in identifying, or preventing at an early stage, new cases of fraud, which a tape-to-tape match of caseloads would do.

In the case of every province that was contacted, the initiative to establish the data link came from the provincial/municipal level, and each agreement is unique. Even in the case of UIC/Income Security agreements, which every province has in place, the amount and type of data obtained differs for each province. This perhaps unintentional "divide and conquer" strategy of the federal government has resulted in different provinces obtaining different data. For example, Prince Edward Island has a data agreement in place, but cannot pursue the on-line linkage feature being pilot tested by Alberta, Ontario and New Brunswick, since the on-line linkage allows access to more data than is allowed by the specific PEI/Canada agreement. The agreement will have to be renegotiated before on-line access will be a possibility. The primary reason for different agreements across the country is that agreements are negotiated by each individual UIC Regional Office, with no obvious central coordination.

Description of Initiative

It is apparent that the federal government has already created linkages among some of its own programs, at least through Revenue Canada for purposes of clawback and overpayment recovery. There are several recommended themes for the federal government to follow in the area of data matches:

1. Ensure that appropriate linkages occur between all federal programs to ensure that fraud and error within the federal system is minimized.

 
2. Establish a consistent data sharing mechanism for on-line access to CPP and OAS/GIS information to which all provinces can link.

 
3. Provide access, either as an on-line system or as a periodic tape-to-tape match, to information on current UIC contributions and income-tax withholding data in Revenue Canada files to provide the most up-to-date information on income.

 
4. Provide access on an annual basis (June of each year) to previous year's income tax data for those current clients who were also on assistance for the previous year to aid in determining whether all income and assets were reported. This will require careful programming to restrict the amount of information available to the provinces to only that data which is required for program administration. Particular care must be given to only providing information for the period of time during which the individual was actually receiving assistance and the month prior. This turns the current one-way flow of information (of provinces providing T4s for clients) into a two-way link.

 
5. The federal government should work with native bands to establish a registry of natives receiving income security benefits through the Department of Indian and Northern Development which can be made available to provinces. In return, provinces should allow access to native bands to provincial client index data to assist native communities in determining whether an applicant is already receiving assistance elsewhere.

 
6. The federal government should monitor and evaluate the cost-effectiveness of each of these linkages to determine which are the most beneficial, and to estimate the true magnitude of "double-dipping".

 

Direct Benefits

Lastly, the interprovincial matches that were started by British Columbia and Alberta, and have slowly moved east to include Saskatchewan and Manitoba, with other eastern and northern jurisdictions soon to join, have identified hundreds of cases of "double-dipping" since their inception. These hundreds of cases translate into hundreds of thousands of dollars saved each month. These matches are of individuals registered under the same name in more than one province, and do not include individuals using different identities in different provinces. The first match is invariably the largest as all the cases with undeclared income come to light, with subsequent months settling down to a smaller amount, as the match becomes more of a deterrent for those clients who are aware of the matching process.

Implementation Strategies

Given the potential impact of increasing tape-to-tape, electronic or on-line matching between provinces and the federal government, and the amount of follow-up that individual caseworkers would be required to do to follow-up on each "match" , a staggered implementation is suggested. It is proposed that a series of pilot sites be established between the federal government and individual provinces, each site focusing on linkage with one federal data source, such as CPP or Revenue Canada. Careful monitoring and evaluation of the results of each pilot would help quantify the actual savings attributable to each source of information.

Implementation Costs

Policy/program design specialist, including drafting agreements 1 month            $5,000
Systems analyst to establish initial match 1.5 months                                         $9,000
System analyst, testing system, setting up procedures 1 month                          $6,000
On-going computer costs (storage time, CPU time) Per month                         $1,300
Cost to add additional province to system 1 week                                            $1,500
Hardware/network installation Per Office                                                         $5,000

In addition, the costs of caseworker staff training (for 1 day) are equivalent to one day's salary for the workers in the office. Local computer costs average about $500 per office, per month.

The federal government is currently redesigning the supporting information systems for federal income security programs. It is probable that these costs can be absorbed within the overall $127 million budget (not including any potential over-runs which may occur) which has been identified for the project, providing decisions were made very quickly.

Issues

The federal government in the United States has clearly specified the protections with which the data involved in the IEVS matches will be ensured. Federal and State law and regulations provide that agencies receiving IEVS information or information provided by other agencies through IEVS, must protect the confidentiality of the information from unauthorized access or disclosure.

• The information shall be used only to the extent necessary to assist in the valid administrative needs of the program receiving such information and shall be disclosed only for these purpose.

 
• The requesting agency shall not use the information for any purpose not specifically authorized.

 
• The information shall be stored in a place physically secure from access by unauthorized persons.

 
• Information in electronic format, such as magnetic tapes or discs, shall be stored and processed in such a way that unauthorized persons cannot retrieve the information by means of computer, remote terminal, or other means.

 
• Precautions shall be taken to ensure that only authorized persons are given access to on-line files.

 
• The county shall instruct all personnel with access to IEVS information regarding the confidential nature of the information and the sanctions against unauthorized disclosure are to be specified in state statutes.

Section 1137 of the Social Security Act (PL 98-339) mandates that, during the application process, all applicants be informed that:

• Everyone who wants assistance must apply for and/or provide an SSN;

 
• SSNs will be compared with records from the SSA (Social Security Administration);

 
• SSNs will be used in computer matches to check income and assets with records of tax, welfare, employment, and other agencies;

 
• Differences between the information provided by the applicant and that provided in the computer matches will be checked out and may have an impact on eligibility or benefit level.

Description of Problem

1. Some provinces "advance" money to clients for one-time-only purchases, and then take a repayment agreement to allow the client to repay the debt. This is often the case for items such as security deposits on accommodations. There is absolutely no wrong-doing or error on the part of the client as this agreement is freely entered into jointly by the provincial government and the client.

 
2. The client makes an honest error in reporting, with no intent to defraud, but it results in an overpayment. For example, a single parent may renegotiate custody with her husband, resulting in her children moving from living full-time with her to living with her ex-spouse, and she retains visiting rights on alternate weekends. She does not understand that this arrangement may affect her benefit level.

 
3. The client chooses to falsely report circumstances, but it is impossible to prove intent in court. For example, a man with a history of mental illness reports that his cheque is missing, and the province replaces the cheque. After forensic examination, it is proven that he cashed the cheque himself. He denies any knowledge of cashing the cheque. Did he intend to defraud? Or was it a lapse of memory? No fraud investigator or public prosecutor would be willing to take this case to court--so an overpayment is filed against the client.

Generally, overpayments filed as a result of repayment agreements or client error are relatively small. Overpayments as a result of fraud tend to be much larger. In 1992, the Ontario Auditor General reported that the outstanding debt due to overpayments for clients who are or were on Ontario provincial income security programs was $80 million. Based on a confidential survey of Canadian provinces⁽³⁾ , overpayments average about 3 to 4% of the total budget for income security at the provincial level. Thus, at any given time, approximately $230 million is outstanding in recoverable debts from clients to provincial governments.

Current Strategies

This means that even a small repayment (such as a $600 damage deposit) would take over two years to repay, assuming the client did not win a hardship appeal. And, it is not uncommon for a client to incur another overpayment long before the first is paid off. Added to this, the majority of clients leave the caseload long before their overpayments are recovered through benefit reduction.

After a case is closed, even in cases where a repayment agreement has been signed by the client, provinces generally make very half-hearted efforts to recover outstanding overpayments. Every year, provincial Treasury, Finance or Crown Debt Collection departments "write off" millions of dollars in unrecoverable overpayments. These funds are written off with no lasting effect on a client's credit rating or any other long-term impact.

When a client has been prosecuted for, and found guilty of, fraud, the court can order seizure of assets, court-ordered restitution, legally defined repayment schedules, or at the very least, legally establish the existence of the overpayment. If the client does not repay as ordered, jail sentences and other penalties as well as civil recovery action can ensue. However, for general repayments, there are no penalties to speak of, since provinces do not take action to establish these debts for civil recovery or refer these debts to commercial credit bureaus.

Tax Refund Intercept Programs

This is similar in nature to the use of the Canadian tax system to intercept income tax rebates before they are issued if the tax-filer has been overpaid by UIC, or is in arrears with student loan payments. These debts are "netted" against the income tax rebate owing to the tax-filer and returned to the federal government before monies are paid to the individual.

Individual provinces, which are "writing off" millions of dollars in outstanding overpayments every year on closed cases, where the individual or family is no longer receiving government income security benefits, have no ability to access the national tax system to recoup these outstanding debts.

In the Canadian context, a decision would have to be made at the federal level whether this could be considered as taxes owing, whether only tax rebates would be affected by this strategy or whether child tax benefits and GST rebates would also be affected. Some states have also had success intercepting larger lottery winnings in order to repay overpayments against state income security programs.

Direct Benefits

It is difficult to estimate, without a careful evaluation of a pilot test, the amount of monies which may be recouped through tax and lottery intercept programs. However, it is likely that at least 10% to 20% of the outstanding overpayments could be recovered each year through this mechanism. This would translate into $23-$46 million a year of additional revenues returned to the provinces which they were unable to obtain before.

Implementation Strategies

1. Establishing, through federal and provincial legislation, that overpayments would be recouped through the tax system.

 
2. Establishment of a national data base of outstanding overpayments for closed cases. It would be quite possible for an individual to have overpayments registered against several jurisdictions (federal, provincial, municipal, and first nations).

 
3. Federal/provincial bilateral agreements that the federal government would intercept tax and other rebates in order to collect monies on behalf of the provinces and would remit same to the provinces.

 
4. Establishment of a pilot test to develop procedures and quickly estimate the amount of money which could be recovered.

 
5. Develop a public/client information campaign to inform individuals of the new process.

Implementation Costs

Although this cost of establishing a national registry could be estimated at approximately $1.5 million, the recoveries of funds to the provinces should far exceed this amount. It may be unreasonable, in today's fiscal climate, that the federal government would agree to this kind of expenditure with no return. A federal provincial cost-sharing arrangement, perhaps dependent on the proportion of tax points for each province, might be an incentive for the federal government to cooperate. Alternatively, the provinces might fund the development of the program.

Issues

As long as the intercept is focused purely on a rebate of overpayment of income tax or lottery winnings, public reaction will probably be positive, however, if child tax benefit and GST rebates are also recovered, there will probably be a public outcry that "the deficit is being carried on the backs of the poor".

Very generally, the public is unaware of the sheer amount of the monies owing by past or present income security clients which is written off as bad debts. Much of this negative reaction could be dealt with through a proactive communication and media campaign.

Description of Problem

Current Strategies

There is nothing more demoralizing to a front-line worker than being forced by policy and legislation to continue to support an individual who has been charged and convicted of fraud, or who has admitted culpability in obtaining an overpayment. They know very clearly that the probability of recouping the funds is very slight.

Generally. there is no recourse in Canada for a federal, municipal or provincial program but to reinstate an individual who has been convicted of fraud or who has admitted to fraud as long as qualifying conditions are met. Only in Prince Edward Island, where provincial legislation allows recovery of an established overpayment through deduction of the entire amount of the individual's benefit, can the overpayment be quickly recovered.
 

American Administrative Disqualification Programs

Individuals found to have committed intentional program violations are notified in writing of the specifics of the overpayment, and for cases where criminal prosecution is not to be undertaken, the client can choose to request an Administrative Disqualification Hearing (ADH) or to sign a waiver of right to a hearing.

Hearings are held at the state level, in some cases by telephone conference, and afford clients the opportunity to explain why they believe they did not commit a violation. Some states find that providing an "evidence packet" which outline the evidence supporting the violation in advance of the ADH is quite successful in convincing clients to sign the hearing waiver at that time.

On the basis of a signed hearing waiver, an Administrative Disqualification Hearing which confirms the violation, or a successfully prosecuted criminal case, the overpayment is established and the client becomes ineligible for further benefits as follows:

• for a period of six months for a first violation
• for a period of one year for a second violation
• permanently for a third violation

States are required to report, to the federal Food and Nutrition Administration, information on each individual disqualified. A national registry is maintained for access by state welfare agencies. In 1991, federal legislation was passed allowing states to extend the Administrative Disqualification process to the Aids to Families with Dependent Children program.

Direct Benefits

Implementation Strategies

1. Assessment of what will be the new or revised Canada legislation in place of CAP and the removal of restrictions provincial and municipal authorities now face on refusing benefits to clients who can show they are "in need". Once this has been done, provincial and municipal authorities may require amendments in some cases to their own legislation.

 
2. A process would have to be developed to allow clients relatively fast access to Administrative Disqualification Hearings - this may be an additional role for already existing Appeal Committees in each jurisdiction, or other systems would have to be established. Some American states have found ways of streamlining this process that could be considered for implementation here.

 
3. A national registry would have to be established with an effective and efficient system for jurisdictions to register information on each disqualified person, assuming disqualification in one jurisdiction would apply across the country. This may be a task for the National Office of Fraud and Error Control Initiatives.

 
4. An intensive public campaign would be required to ensure all clients and potential clients were made aware of the consequences of abusing the program.

 
5. A few pilot test sites in several provinces or municipalities may be desirable to determine the impact of such a process.

 

Implementation Costs

Issues

There will be a concern raised that, by denying benefits to one household member, the system is, in effect, penalizing all household members, especially children, for the transgression of one. However, the argument of personal responsibility would say the offending client very deliberately made the decision to abuse the system, knowing the potential consequences.

It should be noted that when Colorado implemented the Administrative Disqualification process, a class action suit was launched challenging the constitutionality of the initiative. This suit was ultimately unsuccessful in the courts.

Description of Problem

Current Strategies

Description of Initiative

As part of the state efforts to utilize federal funding for fraud detection and prevention programs, many U. S. states have embarked on early fraud prevention programs (specific information was obtained from many states, including Washington, California, Texas, Florida, and Ohio), in addition to their active post-benefit investigations. A comparison of the benefits of these activities follows:
 
 

Table 1 - Cases Reviewed with Savings (1991-92)                                                                FLORIDA                              TEXAS                                                                            Cases               Savings                    Cases              Savings  Post Fraud Referrals (after                             18,549              $5,843,803                   12,852             $15,800,000  benefits were paid)  Fraud Prevention Referrals                             4,081             $7,236,443                    4,880            $12,900,000  (before benefits were paid)  TOTAL                                                             22,630            $13,080,246                   17,732             $28,700,000

The fraud prevention program savings are calculated on the basis of the amount of benefits that the eligibility specialist would have awarded without benefit of the additional information gathered by the investigator, minus the amount of benefits awarded, if any, after the investigation. Two observations can be made about this data. First, the savings identified from Post Fraud investigative activities are speculative at best - the state is often unable to recover the monies owed. Secondly, fraud prevention savings are cost-avoidance, but are fully realized.

For those states undertaking any type of pre-benefit issuance investigation, fraud prevention has been integrated into the intake process, through the detailed training of intake workers in recognizing probable fraud, and referral for investigation prior to benefit issuance rather than depending only on post-fraud investigation, as was the case prior to the introduction of the front-end program. Specific referral criteria have been developed in each state. If these criteria are met, then the case is referred for investigation prior to eligibility being approved. Depending on the state and the program in question (Food Stamps, Aids to Families with Dependent Children, Medicaid), between 4% (Florida) and 21% (Orange County, California) of intakes are referred for investigation prior to benefits being granted. These programs, since they are preventative in nature, result in cost-avoidance rather than identified overpayments. They also reduce the necessity for post-benefit investigations, costly criminal prosecutions, and attempts to recover the benefits that were fraudulently received. Combined data from the fraud prevention investigations in the Food Stamp and AFDC programs in Florida during the period July 1, 1991 to June 30, 1992 show the following results:
 
 

Table 2 - Action Taken                                                                                                                                   Cases                % OF TOTAL Cases Approved without Benefit Adjustment                                                    1,234                          30.8% Cases Approved with Benefit Reduction                                                                541                          13.5%  Cases Denied Benefits                                                                                            1,973                           49.2% Cases where the Application was Voluntarily Withdrawn                                   261                            6.5% TOTAL CASES DISPOSED OF IN FY 91-92                                                        4,009                        100.0%                                                PROGRAM SAVINGS  Food Stamp Program                                                                                       $3,915,232 AFDC Program                                                                                                $3,321,211 TOTAL SAVINGS IN FY 91-92                                                                      $7,236,443

Similar results were found in Orange County, California. Out of 10,192 cases referred from the Food Stamp and AFDC programs, 6,854 (67%) were denied benefits. Based on the assumption that clients would have received benefits for 24 months, a savings of $70,928,214 has been estimated with a savings of $33 for every dollar spent. In the county-operated General Assistance program (which would include all single clients), 5,398 clients were denied benefits out of a total of 5,987 referrals (90% denial of benefits). In 1992 in Ohio, 6,005 referrals resulted in 2,814 cases where cost-savings resulted, for a projected saving of $4,849,559.

In 1988, the federal government passed legislation recommending that every county nation-wide establish Early Detection Units using the Orange county model. A recent survey by the United Council on Welfare Fraud reported that 22 of the 30 states included in the survey had implemented a front-end prevention system. In the first six months of FY 1991, these 22 states had completed 48,663 investigations, with a total cost avoidance of $68,700,000.

On the basis of interviews with Canadian provinces and territories, there is no Canadian jurisdiction which has undertaken a front-end fraud prevention program. From a review of the detailed program documentation provided by the U.S. states contacted, there does not seem to be any specific legislative or other barrier to their implementation in Canada. In addition, despite significant differences between the American Food Stamp and AFDC programs and the Canadian provincial social assistance programs, there is no reason not to anticipate a similar level of cost-effectiveness when this process is applied to Canadian programs.

American program directors indicate that these programs work particularly well in conjunction with extensive data matching and administrative disqualification processes.

Direct Benefits

• Provides higher eligibility worker (intake worker) morale.
• Stops fraud from taking place and thus:
• eliminates post investigations
• eliminates prosecution of recipients
• eliminates the need to recover lost funds
• Provides eligibility workers (intake workers) with a sense of importance because they are controlling the process.
• Provides a cost-effective measure to reduce error rates.
• Provides a method of accountability and cost reduction.
• Increases teamwork between eligibility workers (intake workers) and investigators.
• Provides avoidance of the client becoming involved in the criminal justice system (saves the client from himself).

Implementation Strategies

In Canada, federal support is not necessary to implement these programs, although federal assistance with facilitation of a comprehensive data matching program would certainly improve the cost-effectiveness of the programs.

Implementation Costs

In Canada, the turnover rate in provincial income security programs is approximately 10% per month, that is, there is roughly 1 new or re-opening applicant for every 10 on-going cases on a monthly basis. And, there are over 1,000,000 active welfare files at the provincial and municipal levels. Thus, on a monthly basis, there are approximately 100,000 new applicant/cases. This means that Canada-wide, approximately 330 new investigators would be required across all ten provinces, at an annual cost of approximately $60,000 per investigator (includes salary, benefits, and some administrative support), for a total of nearly $20 million. If the cost-benefit ratio, based on the US experience, is conservatively calculated at approximately 3 to 1, this $20 million investment will return $60 million in savings.

Issues

Putting It Together

As each chapter has identified, it is often difficult to estimate a specific cost-benefit for many of the initiatives. However, American experience clearly demonstrates that all these proposals offer a positive, and sometimes very high, return on dollars invested. To reiterate, Canadian income security programs cost over $85 billion annually, $39.2 billion of which support programs which are considered to be at high risk of fraud and client error. A conservative estimate of the amount of fraud in these programs is estimated to be 4%, and client error is estimated at 5%. Therefore, the magnitude of the fraud and error "problem" in income security programs in Canada is approximately $3.5 billion dollars, spread across federal, provincial, territorial, and municipal programs. If it is assumed that the programs proposed in this document offer the potential to reduce the level of fraud and client error by one quarter to one-third, this would result in a total savings to these Canadian income security programs of something between $875 to $1,155 million annually when fully implemented.

Implementation Considerations and Strategies

• sheer program inertia
• political defensiveness
• differing political perspectives

On the positive side, the imminent demise of the Canada Assistance Plan leaves a vacuum which will enable many of these proposals to be implemented far more easily than if CAP were still firmly in place.
 

Possible Timing of Activities

1995-96       Establish federal leadership and establish the National Office of Fraud and Error
                    Control Initiatives.

                    Extend current agreements to all provinces (eg. province-to-province data matches,
                    on-line UIC and CPP links).

                    Develop Revenue Canada on-line linkage software.

1996-97       Implement pilot programs in provinces/territories/municipalities (tax intercept,
                    administrative disqualification, early prevention, on-line access to Revenue Canada
                    data).

                    Federal establishment of a national client index and national overpayment collection
                    information system.

                    New data sharing agreements with Revenue Canada established.

1997-98      Pilot project evaluations

1998-99      Full implementation begins.

2000           Implementation complete.

This is an ambitious schedule, especially since many provinces, and the federal government will be facing elections midway through implementation. However, it is an achievable target for implementation with senior political commitment.

Conclusion

1. ⁰ Alberta Family and Social Services. Income Support Branch. Report on the Eligibility and Benefit Verification Project and Related Initiatives, October, 1989.

2. ⁰ California. Office of the Governor. California's Fraud Strategic Plan. June, 1994.

3. ⁰ C. A. MacDonald & Associates, 1993.