Summary

Statewide Fingerprint Imaging System: The State Must Weigh Factors Other Than Need and Cost-Effectiveness When Determining Future Funding for the System January 2003 2001-015 rotiduA etatS ainrofilaC S T I D U A E T A T S F O U A E R U B The first five copies of each California State Auditor report are free. Additional copies are $3 each, payable by check or money order. You can obtain reports by contacting the Bureau of State Audits at the following address: California State Auditor Bureau of State Audits 555 Capitol Mall, Suite 300 Sacramento, California 95814 (916) 445-0255 or TDD (916) 445-0255 x 216 OR This report may also be available on the World Wide Web http://www.bsa.ca.gov/bsa/ The California State Auditor is pleased to announce the availability of an online subscription service. For information on how to subscribe, please contact David Madrigal at (916) 445-0255, ext. 201, or visit our Web site at www.bsa.ca.gov/bsa Alternate format reports available upon request. Permission is granted to reproduce reports. � � � ��������� ���� ������ ������������� ������������������� ������������ ����������������������� January 2, 2003 2001-015 The Governor of California President pro Tempore of the Senate Speaker of the Assembly State Capitol Sacramento, California 95814 Dear Governor and Legislative Leaders: As required by Chapter 111, Statutes of 2001, the Bureau of State Audits presents its audit report concerning the Department of Social Services’ (Social Services) Statewide Fingerprint Imaging System (SFIS), which was designed to detect duplicate- aid fraud. This report concludes that Social Services implemented SFIS without determining the extent of duplicate-aid fraud throughout the State. In its eagerness to implement SFIS, Social Services based its estimates of the savings that SFIS would produce on an evaluation of Los Angeles County’s fingerprint imaging system, rather than conducting its own statewide study. We have concerns that the methods Los Angeles County used to develop its savings estimate do not allow for the results to be extrapolated statewide. Further, Social Services’ use of this data assumes that conditions in Los Angeles County hold true in other counties. Similar concerns were expressed by the United States Department of Agriculture as early as 1998. Social Services did not implement SFIS in a manner that would allow it to collect key statewide data during its implementation of SFIS. Therefore, we are unable to determine whether SFIS generates enough savings from deterring individuals from obtaining duplicate aid to cover the estimated $31 million the State has paid for SFIS or the estimated $11.4 million the State will likely pay each year to operate it. Further, the exact cost of SFIS is unknown because Social Services does not track the counties’ administrative costs. Social Services estimated the total administrative costs that all counties except Los Angeles incurred for fiscal year 2000–01 would be roughly $1.8 million, yet Riverside County told us that its estimated costs for the same fiscal year were roughly $1.4 million. Therefore, Social Services may be understating the cost of implementing and operating SFIS substantially. The primary benefits that the State derives from continuing to use SFIS are the proven effectiveness of fingerprint imaging technology to identify duplicate fingerprints and its ability to identify applicants who may travel from county to county seeking duplicate aid. However, opponents of SFIS raise what appear to be valid concerns. For example, they question whether the benefits of SFIS outweigh its potential negative effects on the Food Stamp program, such as the fear it may provoke in immigrant populations. Therefore, the State must weigh these factors in deciding whether to continue SFIS. Respectfully submitted, ELAINE M. HOWLE State Auditor ������������������� �������������������������������������������������� ���������������������������������������� ������������������ CONTENTS Summary 1 Introduction 5 Audit Results Social Services Implemented SFIS Without Determining the Extent of Duplicate-Aid Fraud 13 In Implementing SFIS, Social Services Did Not Require Data Collection, Thus Insufficient Information Exists to Substantiate SFIS’s Cost-Effectiveness 19 Decision Makers Should Consider the Benefits and Drawbacks of SFIS When Deciding Future Funding for the System 32 Recommendations 41 Appendix Selected Information From Our County Survey 43 Responses to the Audit Health and Human Services Agency 55 Department of Social Services 56 California State Auditor’s Comments on the Response From the Department of Social Services 65 SUMMARY RESULTS IN BRIEF The California Department of Social Services (Social Services) began rolling out its Statewide Fingerprint Imaging System (SFIS) to the 58 counties in March 2000 to detect duplicate- Audit Highlights . . . aid fraud in the State’s California Work Opportunity and Responsibility to Kids (CalWORKs) and Food Stamp programs. Our review of the California Duplicate-aid fraud occurs when an individual is receiving aid Department of Social Services’ under two or more active accounts. SFIS prevents duplicate (Social Services) Statewide Fingerprint Imaging System participation by matching fingerprint images of program (SFIS) revealed: applicants against a database containing the fingerprint images of existing program participants. These welfare programs provide þ Social Services implemented SFIS without determining benefits to roughly 1.2 million CalWORKs recipients and the extent of duplicate-aid 1.8 million food stamp recipients. fraud throughout the State. þ It based its estimate of the Federal regulations require states to implement a system to savings that SFIS would detect duplicate aid using, at a minimum, Social Security produce on an evaluation numbers, birth dates, or addresses. Moreover, these regulations of Los Angeles County’s require states that detect a large number of duplicates to fingerprint imaging system, rather than conducting its implement other measures, such as more frequent checks or an own statewide study. increased emphasis on prevention. These regulations do not require the use of fingerprint imaging technology. Most states þ Because Social Services use computer matching against existing databases to verify did not collect key statewide data during its applicants’ information. implementation of SFIS, we are not able to determine Before it implemented SFIS, Social Services did not determine the whether SFIS generates enough savings to cover extent of duplicate-aid fraud throughout the State, and therefore the estimated $31 million it did not properly establish the State’s need for a fingerprint the State has paid for imaging system. Prior to SFIS, most of the State’s counties relied SFIS or the estimated on computer matching and tips received on fraud hotlines to $11.4 million the State will likely pay each year to identify duplicate-aid recipients. When we surveyed counties operate it. regarding the number of duplicate-aid fraud cases they identified prior to SFIS, the data did not suggest that duplicate-aid fraud was a þ In deciding whether to continue SFIS, the serious problem in the majority of counties. Only Los Angeles and Legislature should a few other counties told us that they used their own countywide consider the benefits fingerprint imaging systems to detect duplicate-aid fraud. SFIS provides as well as what appears to be valid concerns regarding the In its eagerness to implement SFIS, Social Services based system, such as the fear it its estimates of the savings that SFIS would produce on an may provoke in immigrant evaluation of Los Angeles County’s fingerprint imaging system populations eligible for rather than conducting its own statewide study. In doing so, it the Food Stamp program. California State Auditor Report 2001-015 11 was remiss in not heeding advice it received from the federal government. Specifically, the federal Department of Health and Human Services’ Office of Inspector General questioned whether Los Angeles County’s fingerprint imaging system was a cost- effective tool to prevent, detect, and deter duplicate-aid fraud. Additionally, in 1998 the United States Department of Agriculture expressed concern about Social Services’ inability to identify the extent of duplicate-aid fraud throughout the State and about its decision to use Los Angeles County’s evaluation to substantiate the savings that SFIS would produce. In fact, Social Services decided not to use federal funds to implement SFIS and instead to proceed using only state funds, in part because the federal government was requiring it to perform a cost-benefit analysis as a condition of using federal funds. We also have concerns that the methods Los Angeles County used to develop its savings estimate do not allow for the results to be extrapolated statewide. Further, Social Services’ use of this data assumes that conditions in Los Angeles County hold true in other counties. Despite these concerns, Social Services did not implement SFIS in a manner that would allow it to collect key statewide data during its implementation of SFIS, such as the number of applicants the counties denied or the number of recipients who are no longer receiving aid because they chose not to comply with the State’s fingerprint imaging requirements. These data would have allowed Social Services to quantify the amount of savings SFIS was generating. According to Social Services, it did not require the counties to collect these data because the Legislature did not provide it with the authority or the resources to require counties to collect data during implementation. However, state laws and policies establish the State’s expectations for implementing information technology (IT) systems, including the need to justify proposed projects’ costs and benefits. Therefore, Social Services was remiss in not bringing its concerns with the lack of authority and resources to the Legislature’s attention so that Social Services could effectively implement SFIS. Due to Social Services’ decision not to collect pertinent data during the implementation of SFIS, we were not able to determine whether SFIS generates enough savings from deterring individuals from obtaining duplicate aid to cover the estimated $31 million the State has paid for SFIS or the estimated $11.4 million the State will likely pay each year to operate it. Further, the exact cost of SFIS is unknown because Social Services does not track the counties’ administrative costs. 22 California State Auditor Report 2001-015 California State Auditor Report 2001-015 33 The primary benefits that the State derives from continuing to use SFIS are the proven effectiveness of fingerprint imaging technology to identify duplicate fingerprints and its ability to identify applicants who may travel from county to county seeking duplicate aid. On the other hand, most of the matches that SFIS identified have turned out to be administrative errors made by county staff, and the level of detected duplicate-aid fraud has been small. Furthermore, opponents of SFIS raise what may be valid concerns. For example, they question whether the benefits of SFIS outweigh its potential negative effects on the Food Stamp program, such as the fear it may provoke in immigrant populations. Both the federal government and the Legislature have expressed a desire to increase participation in the Food Stamp program among persons eligible to receive benefits. The Legislature recently required Social Services to develop a community outreach and education campaign to help eligible families learn about and apply for the Food Stamp program. The use of SFIS may run counter to these efforts. Therefore, the State must weigh these factors in deciding whether to continue funding SFIS. RECOMMENDATIONS To ensure that its implementation of future IT projects meets State expectations, Social Services should do the following: • Collect sufficient data to measure the benefits and costs against the project objectives. • Identify promptly any obstacles that may prevent it from implementing the project effectively. Additionally, to improve its management of SFIS, Social Services should identify the full costs of operating SFIS by requiring counties to track their administrative costs separately. To ensure that its estimates are representative of the entire state and its key assumptions are defensible, Social Services should study the conditions of a sample of counties instead of assuming that conditions in one county hold true in other counties. The Legislature should consider the pros and cons of repealing state law requiring fingerprint imaging, including whether SFIS is consistent with the State’s community outreach and education campaign efforts for the Food Stamp program. 22 California State Auditor Report 2001-015 California State Auditor Report 2001-015 33 AGENCY COMMENTS Although Social Services agrees with most of our recommendations, it disagrees with many of the report’s findings and conclusions. Social Services’ strongest concern is that the report gives the reader the impression that Social Services made the decision on its own to implement fingerprint imaging using unreliable data. Rather, Social Services points out that the decision to implement fingerprint imaging was the result of legislation, after extensive analysis, discussion, and debate among stakeholders. n 44 California State Auditor Report 2001-015 California State Auditor Report 2001-015 55 INTRODUCTION BACKGROUND The California Department of Social Services (Social Services) is responsible for the direct operation or supervision of a wide variety of programs that benefi t Californians. For example, in fi scal year 2002–03 Social Services plans to deliver roughly $20 billion in services and benefi ts to more than 3 million Californians. Its stated mission is to serve, aid, and protect needy and vulnerable children and adults in ways that strengthen Social Services’ Major Program Areas and preserve families, encourage personal Welfare: Provides temporary fi nancial responsibility, and foster independence. Social assistance to eligible needy and dependent Services accomplishes its mission through the use persons to enable them to achieve self- of staff located in more than 40 offi ces throughout suffi ciency. Also monitors, administers, and improves the quality of all welfare programs, the State, the 58 county welfare departments, such as CalWORKs, foster care, child care, and a host of community-based organizations. food stamps, refugee cash assistance, and adoption assistance. Social Services administers four major program areas: welfare, social services, community care Social Services: Provides services to the licensing, and disability evaluation. As part of its elderly, blind, disabled, and other adults and children; protects them from abuse, neglect, and welfare program area, Social Services oversees the exploitation; and helps families stay together. California Work Opportunity and Responsibility Community Care Licensing: Regulates group to Kids (CalWORKs) program and the federal Food homes, nurseries, preschools, foster homes, Stamp Program. It spends roughly $7.8 billion each and day care facilities to assure that they meet the established standards for health and safety. year on these two programs. Disability Evaluation: Evaluates the eligibility Federal law established the Temporary Assistance of applicants for federal and state programs that aid the aged and disabled in an effi cient for Needy Families (TANF) program, which replaces and equitable manner. the Aid to Families with Dependent Children (AFDC) program. TANF provides assistance to needy families so that children can be cared for in their own homes or in the homes of relatives and so that needy parents can end their dependence on government benefi ts; TANF does so by promoting job preparation, work, and marriage. Families must include either a pregnant individual or a minor child who resides with the family. TANF imposes mandatory work requirements. Generally, families must participate in work activities such as on-the-job or vocational educational training for an average of at least 35 hours per week. TANF also limits assistance to fi ve years, although a few exceptions exist for minor children or families experiencing hardship. 44 California State Auditor Report 2001-015 California State Auditor Report 2001-015 55 California legislation enacted in 1997 renamed the State’s AFDC program to CalWORKs and imposed work participation requirements and time limits for the receipt of aid. CalWORKs is the State’s largest cash-assistance program for children and families, with estimated assistance payments in fiscal year 2001–02 of $3.2 billion and other costs totaling $2.2 billion for activities such as assisting with child care and administering the program. CalWORKs is funded by the federal TANF block grant, as well as state and local funds. In federal fiscal year 2001 families were able to receive CalWORKs grants ranging from $58 to $645 per month, depending on income levels. As Figure 1 shows, for fiscal years 1995–96 through 2000–01, the average annual caseload for AFDC and CalWORKs steadily declined. Social Services attributes the recent decline to CalWORKs’ stricter work requirements and the five-year time limit for receiving cash assistance, as well as a booming economy. FIGURE 1 CalWORKs Average Annual Caseload Fiscal Years 1990–91 Through 2000–01 ��� �� �� �� �� � ������ � � ����� � � ����� � � ����� � � ����� � � ����� � � ����� � � ����� � � ��� � � � � � � ����� � � ������ ���������� Source: Unaudited Social Services Research and Development Division data released August 2002. Note: California implemented CalWORKs, which replaces AFDC, in January 1998. 66 California State Auditor Report 2001-015 California State Auditor Report 2001-015 77 ����������������� California also participates in the federal Food Stamp Program, the nation’s largest food stamp assistance program. The federal government fully funds the cost of providing food stamps to needy individuals so that they can obtain a more nutritious diet; however, state and local governments share the program’s administrative costs. To be eligible for food stamps, a household’s assets, gross income, and net income must not exceed certain levels that vary by household size, composition, and location. Social Services’ preliminary estimates indicate that the average size of California households receiving food stamps in federal fiscal year 2001 was 2.7 persons, and the average monthly benefit was $197 per household. Most permanent-resident aliens are ineligible to participate in the federal Food Stamp Program. However, in September 1997, the State implemented the California Food Assistance Program to provide food stamp benefits to noncitizens of the United States who meet the eligibility criteria in effect on August 21, 1996, for the federal Food Stamp Program, but whose immigration status under current federal laws makes them ineligible for benefits. In fiscal year 2001–02 California issued roughly $1.6 billion in food stamp benefits, including $1.5 billion under the federal Food Stamp Program and $80 million under the California Food Assistance Program. For purposes of our report, we refer to both programs as the Food Stamp program. In fiscal year 2001–02 California estimates that it will spend almost $802 million to administer the Food Stamp program at both the State and county levels. As shown in Figure 2 on the following page, since fiscal year 1996–97, the number of California households receiving food stamp benefits has been steadily declining. According to a July 2002 United States Department of Agriculture report on the characteristics of households receiving food stamps, changes in the economy and legislation are among the factors that affect participation in the federal Food Stamp Program. For example, federal law included as part of federal reform of the nation’s welfare program generally disqualifies able-bodied adults without dependents who, during the preceding 36-month period, received food stamp benefits for at least 3 months but worked fewer than 20 hours per week. 66 California State Auditor Report 2001-015 California State Auditor Report 2001-015 77 FIGURE 2 Average Households Receiving Food Stamps Fiscal Years 1990–91 Through 2000–01 ������������������������������� ����������������������� ��� ��� �� �� �� �� � ������ � � ����� � � ����� � � ����� � � ����� � � ����� � � ����� � � ����� � � ��� � � � � � � ����� � � ������ ���������� Source: Unaudited Social Services Research and Development Division data released June 19, 2002. Note: The California Food Assistance Program began in September 1997. Within Social Services’ Welfare-to-Work Division is its Fraud Bureau, which is charged with the responsibility of safeguarding the funds of these and other public-assistance programs by preventing, detecting, and investigating welfare fraud and misspent funds. The Fraud Bureau employs almost 30 people and accomplishes its mission by providing oversight and guidance to the 58 county welfare departments. State law requires all applicants and recipients of benefits under the CalWORKs and Food Stamp programs to provide their fingerprint images, unless they are dependent children or persons who are unable to be fingerprinted for reasons such as the loss of all their fingers. Additionally, state regulations require each adult parent or caretaker living in the home of a child who is receiving or applying for aid or on whose behalf an adult is attempting to seek aid for an unaided excluded child to provide two fingerprint images and a digital photo image as part of the application process. An example of an unaided excluded child is one who receives payments under the Supplemental 88 California State Auditor Report 2001-015 California State Auditor Report 2001-015 99 ���������������������� Security Income/State Supplementary Payment Program. If any such adult refuses to provide the images, the entire household becomes ineligible for the aid. The Fraud Bureau is responsible for overseeing the Statewide Fingerprint Imaging System (SFIS), which is designed to detect duplicate-aid fraud. This type of fraud occurs when an individual is receiving aid under two or more active accounts. SFIS contains an Automated Finger Identifi cation System, which federal law enforcement agencies have used since the 1980s. This computerized subsystem scans and digitizes a fi ngerprint image by creating a map of the individual’s unique ridge patterns and translating that map into a code that the computer can then search for and possibly match. The subsystem conducts two types of searches: an open search, which compares the fi ngerprint image to others in the database, and a closed search, which compares the fi ngerprint image to those on a specifi c account. In addition to fi ngerprint images, SFIS also collects and stores in the database each applicant’s or recipient’s digital photo image. Figure 3 shows an SFIS workstation. FIGURE 3 An SFIS Workstation 88 California State Auditor Report 2001-015 California State Auditor Report 2001-015 99 Some opponents of fingerprint and photo imaging have raised policy and legal concerns about exactly who must be fingerprinted, as well as about the requirement that all individuals be photographed in order to receive aid. Proponents have argued that fingerprint and photo images must be mandatory for all applicants to ensure the effectiveness of the eligibility process. For example, Social Services had interpreted the law as requiring fingerprint images of caretakers who are not themselves recipients of aid but are applying for benefits on behalf of another. The fingerprint imaging system that Social Services uses also includes a photo-imaging component that helps staff determine whether fingerprints are duplicates. Social Services reads the law to permit it to require fingerprint and photo images of parents and caretaker relatives who are not applying for or receiving aid but are in the home with recipients of aid. Others believe this reading of the law is too broad. In a recent lawsuit, Sheyko v. Saenz, the plaintiff challenged Social Services’ interpretation of the law, and the Superior Court ordered Social Services to refrain from requiring images when one parent or caregiver has already applied for aid. The court also ordered Social Services to provide a copy of the order and judgment and written instructions to all county welfare departments. However, the court also ruled that requiring photo imaging of persons who are subject to fingerprint imaging requirements does not violate the law. The court further ruled that the law permits the State to impose a sanction on the entire household when an applicant or recipient of aid refuses to provide fingerprint images and that this does not violate the law. The parties have appealed the ruling, which as of December 2002 was pending before the Third Appellate District in the California Court of Appeal. Social Services believes that the order is stayed, pending appeal. SCOPE AND METHODOLOGY Chapter 111, Statutes of 2001, directed the Bureau of State Audits (bureau) to conduct an audit of Social Services’ SFIS. This system was designed to detect duplicate-aid fraud. The bureau was asked to report on the level of fraud detected through SFIS; the level of fraud deterrence resulting from SFIS; SFIS’s deterrence of eligible applicants, especially the immigrant population, from applying for public benefits; and SFIS’s cost-effectiveness. To gain an understanding of SFIS and its role in detecting duplicate-aid fraud, we reviewed relevant federal and state laws and regulations, as well as Social Services’ policies and 1100 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1111 publications relating to SFIS, CalWORKs, and the federal Food Stamp Program. We also interviewed staff at Social Services, the Health and Human Services Agency Data Center (data center), county welfare departments, and district attorneys’ offices in order to develop an understanding of their roles in relation to the implementation, operation, and maintenance of SFIS. Further, we reviewed various planning and approval documents relating to SFIS’s implementation. Finally, we evaluated Social Services’ efforts to ensure that it keeps the fingerprint images it collects confidential, in accordance with state laws. To evaluate SFIS’s effectiveness in detecting duplicate-aid fraud, we attempted to assess the extent of duplicate-aid fraud in California prior to implementing SFIS and the systems that Social Services and the counties had in place then. In doing so, we reviewed relevant federal laws and regulations relating to the State’s Income and Eligibility Verification System. In addition, we interviewed Social Services and data-center staff to obtain an understanding of how SFIS tracks fraud activity. We also reviewed SFIS data to identify the number of possible fraud matches detected since its implementation. Lastly, we surveyed all 58 counties to determine, among other things, the level of duplicate-aid fraud detected before and after the SFIS implementation. However, we did not perform independent tests of the accuracy of the information the counties provided to us in the surveys. (Please refer to the Appendix for selected information from our county survey.) We were unable to assess SFIS’s effectiveness in deterring duplicate-aid fraud, because Social Services did not perform a control study comparing the actions of fingerprinted applicants to those who were not fingerprinted, nor did it make data collection mandatory during the implementation of SFIS. To determine SFIS’s cost-effectiveness, we identified its costs by reviewing monthly invoices that the data center sent to Social Services, reports from the State’s accounting system, and Social Services’ estimate of county administrative costs. We also, with the aid of a consultant, evaluated the methodologies that Social Services used to calculate savings attributable to SFIS. Finally, we reviewed information from other states to identify the various methods they employed to determine the cost- effectiveness of fingerprint imaging systems. We also spoke with representatives of some states to determine why they chose not to implement fingerprint imaging systems. 1100 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1111 To assess whether SFIS deters eligible applicants, especially immigrant populations, from applying for aid, we analyzed the citizenship status of aid recipients before and after the implementation of SFIS. We also interviewed food advocacy groups to hear their concerns regarding the effect SFIS has on the declining rate of participation in the Food Stamp program. During this audit it came to our attention that Social Services is not ensuring that counties comply with a state law requiring them to report to Social Services the names, birth dates, and Social Security numbers of people whose period of incarceration in jail has exceeded 30 days, rendering them ineligible to receive benefits. Because this issue is beyond the scope of our review of SFIS, we have reported it in a separate letter to the secretary of the Health and Human Services Agency. n 1122 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1133 AUDIT RESULTS SOCIAL SERVICES IMPLEMENTED SFIS WITHOUT DETERMINING THE EXTENT OF DUPLICATE-AID FRAUD The Department of Social Services (Social Services) did not adequately establish the extent of duplicate-aid fraud before implementing the Statewide Fingerprint Imaging System (SFIS). The federal government requires states to have a system to ensure that no individual participates in the federal Food Stamp Program more than once in a month or in more than one jurisdiction or household. Most states use computer matching to comply with this requirement. In 1996 the Legislature passed a law mandating the use of fingerprint imaging for the Food Stamp and California Work Opportunity and Responsibility to Kids (CalWORKs) programs. Prior to its use of SFIS, Social Services was aware of potential duplicate- aid fraud only when the counties brought such cases to its attention. Data from counties using systems in place prior to SFIS’s implementation do not suggest that duplicate-aid fraud was ever serious enough to warrant a costly fingerprint imaging system. Finally, a report that Social Services could have used to gauge duplicate-aid fraud prior to SFIS, referred to as the DPA 266, contains unreliable and limited information. Therefore, Social Services did not know the extent of California’s duplicate- aid fraud prior to implementing SFIS. Other States Comply With Federal Regulations for Preventing Duplicate-Aid Fraud Without the Use of Fingerprint Imaging Technology Duplicate-aid fraud occurs when an individual is receiving aid under two or more active accounts. All states must have a system Although permissible, to ensure that no individual participating in the federal Food federal regulations Stamp Program commits duplicate-aid fraud. Further, federal do not require the regulations require states to identify aid recipients by name and use of fingerprint Social Security number at a minimum and by other identifiers imaging technology. such as birth dates or addresses as appropriate. If states detect a large number of duplicates, federal regulations require them to implement other measures such as more frequent checks or increased emphasis on prevention. Although permissible, federal regulations do not require the use of fingerprint imaging technology. 1122 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1133 Each state is responsible for developing a system to counteract duplicate-aid fraud that is feasible and appropriate for the particular state. The method by which states fulfill this federal requirement differs. California legislation enacted in 1996 mandates fingerprint imaging for applicants and recipients Only four other states receiving aid from the CalWORKs (formerly the Aid to Families use fingerprint imaging with Dependent Children [AFDC]) and Food Stamp programs, technology to detect with a few exceptions. Four other states use fingerprint imaging duplicate-aid fraud. technology to detect duplicate-aid fraud: Arizona, Connecticut, New York, and Texas. These states use their systems for either their Temporary Assistance for Needy Families (TANF), Food Stamps, or General Assistance programs. They believe that their systems are generating savings. For example, Texas estimated that in fiscal year 1999 annual savings for its Food Stamp program ranged between $6 million and $12 million. Other states explored the use of fingerprint imaging technology but chose not to implement it after determining either that it was not cost-effective, that other methods of reducing duplicate- aid fraud were more feasible, or that budget cuts did not allow for it. For example, the state of Maryland conducted a study to evaluate the cost-effectiveness of fingerprint imaging as a method of preventing fraud in its Family Investment and Food Stamp programs. Maryland’s Department of Human Resources found that duplicate-aid fraud cases made up less than 4 percent of all fraud it identified. Therefore, fingerprint imaging would eliminate only roughly 60 of its 50,000 temporary cash- assistance cases, which did not justify the additional costs. Moreover, Maryland found that other antifraud methods, such as preventing fraud at the front end by investigating applicants’ eligibility (this entails visits to schools, neighbors, and the applicants’ homes) provided a better return on its investment. The United States Department of Agriculture (USDA) also commented that because it had no indication that duplicate- aid fraud was a significant problem in Maryland and because the state’s front-end investigation of applicants to determine their eligibility has had proven results, Maryland might more prudently invest in its investigations than in a fingerprint imaging system. The state of Illinois also conducted a demonstration project at three offices to determine the cost-effectiveness of preventing duplicate-aid fraud through the use of fingerprint imaging technology. According to the chief of Illinois’s Bureau of Operations Support, the project lasted fewer than two years, 1144 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1155 and Illinois chose not to continue because the new rules for the federal TANF program reduced the state’s number of cases from 250,000 to 60,000. Thus, implementing a fingerprint imaging system was not cost-effective. Finally, the state of Michigan also considered a fingerprint imaging system, but it cut funding for the system as part of numerous reductions to balance the state budget. Many states, including California, use computer matching to address the problem of duplicate-aid fraud. Computer matching verifies the accuracy of information that an applicant provides by comparing it to information contained in existing databases. For example, the state of Illinois matches many identifiers, such as names, addresses, and Social Security numbers, to verify whether an applicant is receiving benefits in more than one county. Illinois also has online access to similar data from its neighboring states and compares its records with theirs. Thus, states have found many ways, other than using fingerprint imaging technology, to fulfill the federal requirement of detecting duplicate-aid fraud. Social Services Did Not Know the Extent of Duplicate-Aid Fraud Before Implementing SFIS Social Services implemented SFIS to ensure that individuals cannot commit duplicate-aid fraud in the State’s CalWORKs and Food Stamp programs. However, California did not know the extent of its duplicate-aid fraud before it implemented SFIS. Moreover, data reported by the counties prior to SFIS’s implementation did not suggest to us that duplicate-aid fraud was ever serious enough to warrant a costly fingerprint imaging system. Before SFIS was in place, estimating how much duplicate-aid fraud actually existed in the State was difficult. Social Services Data reported by counties was aware only of potential cases of duplicate-aid fraud that responding to our survey the counties brought to its attention. The methods the counties regarding the number of used to detect duplicate-aid fraud prior to SFIS met the federal duplicate-aid fraud cases requirement and were similar to those used in other states. identified prior to SFIS According to our survey, the counties used computer matches did not suggest to us that as the primary method to detect possible duplicate-aid fraud, duplicate-aid fraud was a followed closely by tips from concerned citizens or other serious problem. organizations. Data from the counties responding to our survey regarding the number of duplicate-aid fraud cases identified prior to the implementation of SFIS did not suggest to us that duplicate-aid fraud was a serious problem. 1144 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1155 Only Los Angeles and a few other counties told us they used their own countywide fingerprint imaging systems to identify duplicate-aid fraud. Specifically, these counties’ available data indicate that from July 1995 through the date of their conversion to SFIS in 2000, the counties investigated 74,770 possible fraud cases, of which 972 were substantiated. The counties also reported that the dollar value of the fraud associated with these substantiated cases was $654,905. In addition, Alameda County reported that although it was able to provide the number of possible fraud cases investigated, it was unable to provide any further information because it does not track investigative outcomes by type. Los Angeles County reported that it experienced an increase in the number of possible fraud cases investigated in fiscal years 1999–2000 and 2000–01 due to the inability of its fingerprint imaging system to recognize certain information in its newly implemented eligibility system. Los Angeles County further states that although the majority of these matches were not due to fraudulent activity, the county had to investigate to rule out fraud. Similarly, available data reported by 46 counties that used computer matching or other sources indicates they investigated Forty-six counties reporting 1,408 potential duplicate-aid cases from July 1995 through that they used computer the date of their conversion to SFIS. However, they indicated matching or other that only 583 of these cases were substantiated. Only two sources to detect fraud counties reported a possible explanation for the difference said they substantiated between the number of cases investigated and substantiated. only 583 cases while For example, Solano County told us that most cases it closed participation levels in as unsubstantiated were due to either an unfounded allegation the CalWORKs and or an amount of fraud that had no adverse financial impact. As Food Stamp programs of the end of March 2000, shortly after Social Services began were roughly 281,000 to roll out SFIS in the counties, the number of CalWORKs and 351,000, respectively. cases and the Food Stamp program households for these 46 counties were roughly 281,000 and 351,000, respectively. Given the participation levels in both programs and the number of substantiated cases, it appears that the extent of known duplicate-aid fraud was not significant. Federal law requires state agencies to develop and implement an Income and Eligibility Verification System (IEVS) to verify the eligibility and benefit levels for applicants and recipients of the CalWORKs and Medicaid programs. California chose also to use IEVS for its Food Stamp program. California implemented IEVS in 1987 and continues to use it. The California Department of Health Services (Health Services) operates the applicant portion of the State’s IEVS. When an individual applies for aid 1166 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1177 through CalWORKs, Medi-Cal (California’s Medicaid program), or the Food Stamp program, counties submit that person’s identifying information, such as name and Social Security number, to Health Services. Health Services then cross-matches the applicant information against its statewide Medi-Cal Eligibility Data System and other data maintained by the State’s Employment Development Department and Franchise Tax Board, as well as by the federal Social Security Administration and the Immigration and Naturalization Service. For example, if an individual currently receiving welfare benefits in a county applies for aid in another county using the same Social Security number, the Medi-Cal Eligibility Data System will alert the welfare department in the second county. Social Services operates the recipient portion of the State’s IEVS, which generates information by matching the Medi-Cal Master Extract File and county files with various income and eligibility sources. The Integrated Earnings Clearance/Fraud Detection System (fraud detection system), which is now a subsystem of IEVS, has been in effect since 1972 and can detect cases in which recipients either fail to report or underreport wage information to the counties. Additionally, since 1983 the fraud detection system has been able to generate a duplicate-aid match when a recipient of the CalWORKs, Food Stamp, or Social Security Income/State Supplementary Payment programs has two or more records containing the same Social Security number, gender, birth date, and three to five characters in the first and/ or last name. Finally, the fraud detection system can report For several years, information regarding possible recipients of duplicate aid within Social Services has operated a county; between counties; and from the states of Oregon, a fraud detection system Nevada, and Arizona. Thus, the fraud detection system contains that contains sufficient sufficient information to assist Social Services in estimating information to assist the extent of duplicate-aid fraud. For example, according to Social Services in estimating Social Services, the fraud detection system identified roughly the extent of duplicate-aid 9,600 potential cases of duplicate-aid fraud in the CalWORKs fraud, but it did not use this and Food Stamp programs in 2001. However, Social Services did information to evaluate not use this information to evaluate the extent of duplicate-aid the extent of duplicate-aid fraud in the State. Although Social Services believes that most fraud in the State. of the potential cases the fraud detection system identifies are usually due to intercounty transfers resulting from individuals moving across county lines or administrative errors, we are unable to verify its belief. Specifically, Social Services told us that it does not require the counties to report on the potential cases they substantiate as fraud because it does not have the resources to track this type of information and would rather focus on the number of cases processed and overpayments 1166 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1177 established during a particular period. Moreover, Social Services points out that the fraud detection system is limited in that it checks for duplicate matches only among individuals who use a single Social Security number, as opposed to individuals who inappropriately use multiple numbers. Although Social Services tracks statewide statistics on duplicate- Social Services has not aid investigation requests, it does not maintain statistics on resolved fully its problems the results of these investigations. It tracks these statistics via with the accuracy of the a monthly report, referred to as the DPA 266, which it requires DPA 266 that we identified each of the counties to complete. In this report, counties must in a March 1995 audit indicate the number of investigation requests for 18 categories, report. If Social Services including duplicate aid. However, Social Services asks counties had been able to capture to group the results of the investigations for these categories more detailed and reliable and to separately identify only the results of their early fraud data, it may have been able prevention and detection investigations. Social Services to present a clearer picture recognizes that the DPA 266 is limited because it does not break of the extent of duplicate- down the results of investigations by type or dollar impact, but aid fraud in the State. the department states that the counties are adamantly opposed to any further reporting requirements. We raised concerns regarding the DPA 266 in our March 1995 report, titled Department of Social Services: Review and Assessment of the Cost Effectiveness of AFDC Fraud Detection Programs. Specifically, we found that Social Services’ instructions to the counties on how to complete the DPA 266 were not thorough enough to assure consistent completion by all counties, especially considering that many counties have unique programs, techniques, and organizational structures. Further, we found that several of the counties did not properly interpret the instructions for gathering and reporting fraud activity, which led them to submit inaccurate data to the State. We recommended in 1995 that Social Services develop an ongoing desk-review program of the DPA 266 to review the reports for consistency and reasonableness and provide timely feedback to the counties when it notes errors. In April 1996 Social Services told us that staff from its Information Services Bureau had completed a desk review of all 58 counties’ DPA 266 reports for fiscal year 1993–94 and ensured that counties with deficiencies had corrected their errors. However, despite these efforts, it appears that Social Services has not resolved fully its problems with the DPA 266. Our survey results indicate that the counties do not report information consistently on the DPA 266, and therefore it is an unreliable report. The 57 counties responding to our survey reported almost 44,000 possible duplicate-aid fraud cases requiring investigation between July 1999 and June 2002. 1188 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1199 However, during this same period, unaudited monthly statewide DPA 266 reports show roughly 4,500 duplicate-aid investigation requests. For example, the DPA 266 reports show that only 2,300 duplicate-aid investigation requests occurred in Los Angeles County, yet this county reported in our survey that it received almost 42,100 such requests during the same period. If the county’s survey data is correct, the DPA 266 underreports investigation requests in Los Angeles by 95 percent. According to the chief of the Fraud Bureau, Social Services no longer verifies the accuracy of the information the counties report, because it does not consider the DPA 266 to be a statistical or claiming document but merely an activity report. However, this is inconsistent with Social Services’ instructions for completing the DPA 266, which state that information collected on the DPA 266 is used to prepare a federal program activity report and special reports for the Legislature. Specifically, federal regulations require state agencies to submit to the USDA an annual program activity statement that includes data on investigations of precertification and postcertification fraud. If Social Services had captured more detailed and reliable data using the DPA 266, it might have been able to present a clearer picture of the extent of duplicate-aid fraud identified by the counties. Social Services had a few options available for determining In its eagerness to the known extent of duplicate-aid fraud in the State prior to implement SFIS, Social implementing SFIS. For example, it could have surveyed the Services based its estimates counties as we did or requested counties to analyze their fraud of the savings that SFIS detection system and DPA 266 data to determine the extent of would produce on an duplicate-aid fraud. Instead, as we discuss later, Social Services relied evaluation of Los Angeles on the experience of one county to implement a statewide system. County’s fingerprint imaging system, rather than conducting its own IN IMPLEMENTING SFIS, SOCIAL SERVICES DID statewide study. NOT REQUIRE DATA COLLECTION, THUS INSUFFICIENT INFORMATION EXISTS TO SUBSTANTIATE SFIS’S COST-EFFECTIVENESS In its eagerness to implement SFIS, Social Services based its estimates of the savings that SFIS would produce on an evaluation of Los Angeles County’s fingerprint imaging system, rather than conducting its own statewide study. In doing so, Social Services was remiss in not heeding advice it received from the federal government. For example, as early as 1998, the federal government expressed concern about Social Services’ inability to identify the extent of duplicate-aid fraud in the State 1188 California State Auditor Report 2001-015 California State Auditor Report 2001-015 1199 and about its use of Los Angeles County’s evaluation of the cost- effectiveness of its own fingerprint imaging system to substantiate the savings that SFIS would produce. However, Social Services did not implement SFIS in a manner that addressed the federal government’s concern, which would have allowed it to collect key statewide data when it began using SFIS, such as the number of applicants the counties denied or the number of recipients who are no longer receiving aid because they chose not to comply with the State’s fingerprint imaging requirements. According to Social Services, it did not require the counties to collect these data because Los Angeles County had already proven that a fingerprint imaging system would be cost-effective, and Social Services did not want to burden the counties with additional reporting requirements. Social Services asserts that in mandating the use of SFIS, the Legislature did not provide it with the authority or the resources to require counties to collect data during implementation. Nevertheless, due to Social Services’ decision not to collect pertinent data during the implementation of SFIS, we are unable to determine how much savings SFIS generates by deterring individuals from obtaining duplicate aid. Therefore, we cannot determine whether or not the savings it may generate are enough to cover the estimated $31 million the State has paid for SFIS or the estimated $11.4 million the State will likely pay each year to operate SFIS. Social Services Was Remiss in Not Heeding Advice to Adequately Evaluate SFIS During the development of SFIS, Social Services did not follow the advice of the federal government, which identified the need to evaluate the cost-effectiveness of a fingerprint imaging system. Instead, Social Services proceeded to develop and implement SFIS without evaluating the extent of duplicate-aid A 1995 business plan fraud in the State or the cost-effectiveness of SFIS. Consequently, prepared by the data we cannot determine the true value of SFIS. center for SFIS included a methodology that would A Tentative Proposal for Evaluating SFIS Was Reasonable likely have been able to quantify any statewide The Budget Act for fiscal year 1995–96 authorized the Health savings attributable to and Welfare Agency Data Center, later renamed the Health SFIS, but Social Services and Human Services Agency Data Center (data center), to and the data center did develop the necessary plans and documents to implement not adhere to the plan a biometric identification system, that is, a system that when implementing SFIS. identifies individuals based on physical characteristics, such as fingerprints. Accordingly, in 1995 the data center developed a 2200 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2211 business plan that outlined a description of the SFIS project, its preliminary costs and benefits, the procurement process, and a tentative methodology for attributing savings to SFIS. The data center submitted the business plan to the Department of Finance’s Technology Investment Review Unit and received conditional approval in November 1995. The business plan proposed a methodology for quantifying savings attributable to SFIS that included implementing SFIS over a six-month period by adding a subset of counties each month. Each month two randomly selected groups of cases would be drawn to establish a control group and an experimental group of AFDC recipients. Individuals in the control group would not be fingerprinted, but individuals in the experimental group would be fingerprinted. Then the amount of benefits paid to each group in the first calendar month in which SFIS had its full effect on the experimental group would be used to calculate an initial savings amount. The recidivism rate, the rate at which individuals previously terminated from receiving aid return to aid, would be tracked for each county for one year and used to adjust the initial savings. This is a reasonable methodology that would likely have been able to quantify any statewide savings attributable to SFIS. The State Relied Inappropriately on the Results of Los Angeles County’s Demonstration Project to Establish the Need for SFIS As the SFIS project moved through the State’s planning and approval process, Social Services and the data center did not adhere to the proposed methodology outlined in the business plan but relied instead on the results of Los Angeles County’s fingerprint imaging demonstration project. Data- center staff were unable to explain why the State chose not to adhere to the business plan’s methodology, but the deputy director of Social Services’ Welfare-to-Work Division presented the following reasons: (1) the business plan was developed prior to welfare reform and block-grant provisions that eliminated the need for the federal Department of Health An April 1996 planning and Human Services approval for funding; (2) the release of the document prepared by independent evaluation of Los Angeles County’s Automated the data center suggests Fingerprint Image Reporting and Match (AFIRM) system the business plan was made further data collection and evaluation unnecessary; and not followed because (3) the State had no authority, nor were resources provided the State was eager to in the legislation, to require counties to collect data during implement a fingerprint implementation. However, an April 1996 planning document imaging system statewide. from the data center states that, in considering the success of the demonstration project for Los Angeles County’s AFIRM 2200 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2211 system, the State was eager to implement a fingerprint imaging system statewide. Additionally, the document indicates that another reason for expediting the implementation of SFIS is that the State expected to replace the county’s AFIRM system with SFIS. Specifically, the data center was concerned that the waiver granted by the federal Department of Health and Human Services to conduct the county’s demonstration project would expire on March 31, 1999, or when the State implemented SFIS, whichever was earlier. A key condition of the federal waiver required Los Angeles County to conduct an outcome and process evaluation of its use of AFIRM. The basic evaluation was to include the identification and longitudinal tracking, or tracking over a period of time, of an experimental group and a control group of AFDC cases. Further, an independent evaluator was to conduct the cost- benefit analysis. The final evaluation report of Los Angeles County’s AFIRM system states that the experimental group and a control group were randomly selected from the county’s AFDC caseloads as of April 1, 1994. The county sent the experimental group an appointment letter for fingerprinting, but it did not Los Angeles County fingerprint the control group until after the demonstration demonstrated that costs project’s conclusion. The county tracked the amount of benefits for its AFIRM system were paid to these two groups for 26 months, from August 1994 not larger than the savings to September 1996. The county used the net difference in the attributable to it; however, payments to the two groups, resulting from those individuals the inspector general in the experimental group who chose either not to continue noted that, if AFIRM was the application process or be fingerprinted and thus to to be principally a tool to discontinue receiving aid, to estimate the savings AFIRM would prevent, detect, and deter generate. Using different methods, the consultant calculated only duplicate-aid fraud, three savings estimates of $59 million, $73 million, and available data did not $85.2 million and compared these estimates to Los Angeles clearly demonstrate its County’s actual program costs of $20.6 million. Thus, the cost-effectiveness. consultant demonstrated that AFIRM’s costs were not larger than the savings attributable to it, which was a key condition of the federal approval. As a further condition of the waiver, Los Angeles County was required to determine the reasons for noncooperation and to establish the incidence of possible fraud in the population. To meet this requirement, the county selected a sample of 137 cases where the people did not cooperate with the fingerprint imaging requirement, and it performed a detailed investigation. The consultant reviewed the approach, methods, and procedures the county used to select and evaluate the cases. The Table shows the results of the county’s investigation. 2222 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2233 TABLE A Review of 137 Discontinued Cases in Los Angeles County Total duplicate-aid fraud 31 Confirmed duplicate-aid fraud 17 Verified duplicate-aid fraud* 8 Phantom clients† 9 Highly probable duplicate-aid fraud‡ 14 Fraud other than duplicate aid 63 No overpayment 10 No fraud found 33 Total cases investigated 137 Source: Ernst & Young LLP, February 1996 report on Los Angeles County Department of Social Services, AFDC AFIRM Demonstration Project. *Of the eight conclusive cases of duplicate-aid fraud, all were in jurisdictions outside Los Angeles County, and most were outside California. † A phantom case is one in which investigators could find no evidence of the individual’s existence. ‡ Highly probable duplicate-aid fraud includes those clients who had previously demonstrated behavior that the county believes would be consistent with persons who establish duplicate cases, such as using false identification documents. One of the conclusions the consultant drew was that AFIRM prevents, deters, and detects duplicate-aid fraud. However, during and subsequent to the three-year evaluation period of the AFIRM project, others raised concerns about the data and its use to justify SFIS. Specifically, in May 1996 the federal Department of Health and Human Services’ Office of Inspector General (inspector general) conducted a review of Los Angeles County’s demonstration project to evaluate the results obtained and to determine whether project costs would be offset by program savings or whether the project would be cost-neutral. The inspector general concluded that the county had achieved significant savings in addition to meeting the cost-neutrality provisions of the AFIRM project. Further, the data shows that AFIRM had achieved positive results in combating fraud. However, the inspector general also noted that if AFIRM was to be principally a tool to prevent, detect, and deter only duplicate- aid fraud as opposed to other types of fraud, available data did not clearly demonstrate its cost-effectiveness. As shown in the Table, only eight of the discontinued cases were verified as actually receiving duplicate aid. The inspector general did not make any recommendation with regard to the ongoing AFIRM project. 2222 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2233 In November 1997, in response to the data center’s request for approval and funding to support the implementation of In June 1998 the regional SFIS, the regional director of the federal Food Stamp Program director of the federal within the USDA gave contingent approval and asked the data Food Stamp Program center to provide a cost-benefit analysis. In March 1998 the data expressed concerns about center submitted a cost-benefit analysis that it prepared using the data center’s inability data from Los Angeles County’s AFIRM demonstration project, to identify the duplicate- stating that this project was large enough to have statistical aid population to be relevance for expansion to statewide projections. In June 1998 deterred by SFIS, its use the regional director of the federal Food Stamp Program of data from one large responded by expressing concerns about the data center’s urban county as a model inability to identify the duplicate-aid population to be deterred for projecting savings by SFIS; its use of data from one large urban county in the State statewide, and its use as a model for projecting savings statewide; and its use of data of data that was not an from Los Angeles County’s demonstration project, which was accurate representation not current and therefore was not an accurate representation of of current circumstances. current circumstances. The regional director asked the State to provide supporting documentation to show that the projected savings were due to the prevention of duplicate participation. Our review of the final evaluation report of Los Angeles County’s AFIRM demonstration project raised similar concerns. For example, our consultant informs us that the evaluation was based on a systematic sampling design, which selects cases using a sampling interval. This sampling design does not provide for a margin of error that would take into account any limitations in the sampling design or available data. Because the county used a sampling design in which the variance of the estimate could not be estimated from the data, none of the savings estimates presented in the report reflects an upper or lower margin of error. Thus, the true range of the savings attributable to AFIRM is unknown. Further, our consultant informs us that it is inappropriate for Social Services to extrapolate Los Angeles County conditions when developing statewide estimates because doing so assumes that these conditions hold true in other counties. Therefore, we disagree with the data center’s statement that the demonstration project had statistical relevance for expansion to statewide projections. In August 1998 the Health and Welfare Agency (agency), later renamed the Health and Human Services Agency, requested the governor’s approval to proceed with the procurement and implementation of SFIS without full federal approval. The agency stated that the SFIS implementation schedule for rolling out the system to the 58 counties in phases did not allow for a cost-benefit evaluation to occur between phases. 2244 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2255 This evaluation approach would have been similar to the one outlined in the business plan. In addition, the agency stated that addressing the concerns of the regional director of the federal Food Stamp Program would result in delays to the procurement and implementation schedule that would mean a loss of state The agency told decision savings amounting to roughly $73,000 per day. Finally, the makers that addressing agency stated that, based on the results of Los Angeles County’s the USDA’s concerns AFIRM demonstration project, it did not feel that analyzing would result in delays costs and benefits after each phase of the implementation was to the procurement and necessary. The agency believed that once actual cost-benefit implementation schedule, data were available, the USDA might well retroactively approve resulting in a loss of state funding for SFIS. The governor’s office granted approval to savings amounting to proceed with SFIS using only state funds. However, since rolling roughly $73,000 per day. out SFIS to the 58 counties in March 2000, Social Services has never produced actual cost-benefit data, and it continues to fund SFIS using only state funds. During Implementation, Social Services Missed Its Opportunity to Determine SFIS’s Cost-Effectiveness Capturing critical data during the implementation phase would have allowed Social Services to quantify the savings attributable to SFIS. The failure of the data center and Social Services to adhere to the methodology outlined in the business plan and to address the concerns of the regional director of the federal Food Stamp Program has caused the State to miss its opportunity to obtain information it needs to accurately estimate any savings that SFIS generates. On February 29, 2000, Social Services asked counties planning to collect information regarding the denial and discontinuation of aid related to SFIS to share the information voluntarily with Social Services so that it could measure the impact of SFIS on the CalWORKs and Food Stamp programs. Social Services suggested a report format but added that it would accept data in any format. The suggested format tracks the number of applications and recipient cases the county denied or discontinued due to either an applicant’s or recipient’s failure to fulfill the fingerprint imaging requirement or a duplicate-aid match identified by SFIS. However, the report does not track the number of applicants or recipients who later complied with the fingerprint imaging requirement and received aid. Therefore, Social Services does not have an accurate picture of SFIS’s impact on the denial or discontinuance of aid. 2244 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2255 The deputy director of Social Services’ Welfare-to-Work Division told us that in mandating SFIS, the Legislature did not provide any statutory authority or resources to require counties to collect Social Services and the data. Although we agree that state law mandating SFIS neither data center had roughly explicitly mandates the collection of data nor provides funding four years to inform the for these efforts, it does require Social Services and the data Legislature that they center to design, implement, and maintain the system. Other lacked the statutory state laws and policies establish the State’s expectations for authority and resources to implementing information technology (IT) projects. For example, require counties to collect state law holds the head of each agency responsible for the data, but did not do so. management of IT in the agency that he or she heads, including the justification of proposed projects in terms of costs and benefits. Further, state policy requires agencies to establish reporting and evaluation procedures for each approved IT project and to prepare a post implementation evaluation report that measures the benefits and costs of a newly implemented IT system against the project objectives. The State does not consider a project complete until the Department of Finance approves the post implementation evaluation report. Data collection is a key component in preparing this report. Therefore, the data center and Social Services were remiss in not bringing the lack of authority and resources to the Legislature’s attention so they could effectively implement SFIS. Moreover, because counties did not begin to use SFIS until March 2000, roughly four years after the passage of the law, it is reasonable to conclude that the data center and Social Services had ample opportunity to do so. At a minimum, the data center and Social Services could have studied the conditions of a sample of counties throughout the State during the implementation of SFIS. If they had done so, they would have found some counties feel the SFIS system is not beneficial because the costs and resources necessary to maintain it exceed the minimal risk of duplicate-aid fraud in those counties. For example, Nevada and Sierra counties believe that SFIS is not a cost-effective method of preventing duplicate-aid fraud for small counties. Moreover, Orange County believes that SFIS adds additional workload and costs for its clerical, eligibility, and welfare fraud staff, but provides minimal deterrent or cost-savings result. Finally, Riverside County stated that its estimated staffing costs to operate SFIS for fiscal years 2000–01 and 2001–02 exceeded its estimated restitution and 2266 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2277 savings by $2.7 million. The data center told us that its post implementation evaluation report on SFIS is due by July 2003. However, it will rely on Social Services to calculate the benefits and costs associated with SFIS. Incomplete Cost Data and a Flawed Method for Estimating Savings Renders Social Services’ Cost-Benefit Analysis for SFIS Unreliable Social Services tracks some of the costs associated with SFIS, but it does not track county administrative costs. As a result, it does not know the full costs of operating SFIS. Further, because Social Services did not capture the data necessary to determine the savings attributable to SFIS during its implementation, Social Services developed an estimate based on the results of Los Angeles County’s AFIRM demonstration project. However, the methodology it used to estimate the State’s savings of roughly $150 million over five years for SFIS is flawed and therefore unreliable. According to Social Services’ records, the State spent an estimated The State spent an $31 million on SFIS between fiscal years 1995–96 and 2001–02. estimated $31 million on Included in these costs are three main components: data- SFIS between fiscal years center costs that include payments to Electronic Data Systems 1995–96 and 2001–02, Corporation (EDS) for the use of SFIS equipment, including but we were unable to the operating software; Social Services’ administrative costs, determine the counties’ including salaries, wages, rent, utilities, and office supplies; actual costs because and the counties’ administrative costs. County administrative Social Services did not costs should include staff time to fingerprint and photograph require counties to track applicants and to resolve any matches SFIS identifies. Figure 4 them separately. on the following page summarizes the Social Services and data-center costs. Although we were able to substantiate the data center’s and Social Services’ costs, we were not able to determine the counties’ actual costs because Social Services did not require counties to track SFIS administrative costs separately. Counties recoup their CalWORKs and Food Stamp programs’ administrative expenses by submitting a claim to Social Services. However, we found in our review of Social Services’ instructions to counties on how to complete the claim and in our conversations with a few counties that the counties bundle their SFIS administrative expenses with other eligibility costs, making it difficult to identify SFIS administrative costs. 2266 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2277 FIGURE 4 SFIS Costs Fiscal Years 1995–96 Through 2001–02 �������������� ���������� ��� �� � � � � � ������� ������� ������� ������ � � �������� ������� ������� ���������� Source: Social Services accounting records and Health and Human Services Agency Data Center unpaid monthly invoices for an additional billing in fiscal year 2000–01 and for the period of March through June 2002. Note: Costs do not include county administrative costs. To estimate the counties’ administrative costs, Social Services uses a variety of factors. However, we found that certain assumptions in its methodology for obtaining this estimate are questionable. Specifically, Social Services’ estimate assumes that it takes county staff 5 minutes to print and photograph an applicant and 15 minutes to deny or discontinue cases when applicants fail or refuse to provide fingerprint images. However, Social Services did not conduct a time study to support these figures because it considers SFIS to be a part of the eligibility determination process. Social Services told us that it used the best information available at the time it built the estimate in November 2000 and believes the 5 minutes and 15 minutes are logical assumptions. We disagree. Social Services estimated that the total administrative costs that all counties except Los Angeles 2288 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2299 ����������������� incurred for CalWORKs and the Food Stamp program for fiscal year 2000–01 would be roughly $1.8 million, yet Riverside County told us that its estimated costs for the same fiscal year were roughly $1.4 million; Riverside County alone estimated its costs as amounting to 78 percent of the costs Social Services estimated for the 57 counties. Additionally, Social Services’ estimate does Social Services estimated not include the cost that counties incur for investigating possible that the administrative fraudulent activity because it believed these costs would be costs for all counties except minimal. However, the data from one county suggests that these Los Angeles for fiscal costs may not be minimal at all. Specifically, Shasta County year 2000–01 would be estimates that completing an investigation and writing the roughly $1.8 million, yet report can take between 1.5 hours and 15.5 hours, at a cost of Riverside County estimated about $20 per hour. Furthermore, Social Services chose not to that its costs for this period include any administrative costs for Los Angeles County in its were roughly $1.4 million, estimate because the county had not implemented SFIS at the or 78 percent of time Social Services released this estimate. Los Angeles County Social Services’ estimated represents roughly 38 percent of CalWORKs cases and 40 percent costs for 57 counties. of Food Stamp program households. Therefore, Social Services may be understating the cost of implementing and operating SFIS substantially. Social Services’ November 2000 estimate also attempts to quantify benefits or savings that would accrue to the CalWORKs and Food Stamp programs. This estimate comprises the following components: savings attributable to the avoidance of awarding CalWORKs grants to individuals who attempt to commit duplicate-aid fraud and savings in the cost of administering both CalWORKs and the portion of the Food Stamp program that aids households that do not receive public assistance. Figure 5 on the following page shows the estimated savings for each component. The estimate does not include savings attributable to the avoidance of duplicate-aid fraud in the Food Stamp program because the data was not available. Further, Social Services also did not include savings resulting from Los Angeles County’s use of SFIS because the county was not yet using SFIS when Social Services built the estimate. Finally, Social Services used data from Los Angeles County’s demonstration project to support key assumptions in its development of the SFIS savings estimate, which is inappropriate because it assumes that these conditions hold true in other counties. 2288 California State Auditor Report 2001-015 California State Auditor Report 2001-015 2299 FIGURE 5 Social Services’ Estimate of SFIS Savings Fiscal Years 1999–2000 Through 2003–04 ��������� ��������������������� �������� ��������������������� �������� ������������ ��� �� �� �� �� �� � ��������� ������� ������� ������� ������� ���������� Source: Social Services’ Estimates Branch. Note: Fiscal year 2003–04 estimate represents the first quarter of savings for CalWORKs because the State’s contract with the SFIS vendor terminates on September 6, 2003. In developing its estimate, Social Services used the actual caseload and approved applications to estimate the number of applicants or recipients that SFIS would deter—those who were committing or contemplating duplicate-aid fraud but would decide not to because of the fingerprint imaging requirement. To this number, it applied an outdated percentage that it claims was computed during the evaluation of Los Angeles County’s AFIRM demonstration project when the project was used solely for the county’s General Relief program. Specifically, according to Social Services, Los Angeles County’s General Relief program’s experience with AFIRM during the period of June through December 1991 showed that 6.67 percent of the recipients were discontinued from receiving aid because they did not show up 3300 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3311 ������������������� for the scheduled fi ngerprint imaging appointment. However, Los Angeles County staff could not provide documentation to support the 6.67 percent fi gure because the data is more than 11 years old, and the county’s policy is to maintain documents for seven years. Social Services also multiplied the 6.67 percent by 33 percent. It told us that it initially built the Social Services’ Savings Methodology estimate about 1995 using the best information then available, including Los Angeles County’s 1. Estimate the number of people SFIS deters. experience with its AFIRM demonstration project. CalWORKs and Food Stamp programs’ caseloads Social Services states that it recognized the need or applications are multiplied by 6.67 percent, to adjust for the difference between Los Angeles which is then multiplied by 33 percent. The result is divided by the SFIS implementation County’s General Relief program and the State’s period for the respective programs to arrive CalWORKs and Food Stamp programs. Based on at the average monthly cases or applications. This fi gure is then adjusted using Los Angeles coordinated discussions between its Estimates County’s recidivism data. Branch staff and CalWORKs and Food Stamp programs’ staff, Social Services felt that 33 percent 2. Identify the average savings per case or application. was a reasonable fi gure to account for the fact that fewer recipients would be inclined to commit fraud Social Services applies the following to the number of cases or applications computed in the CalWORKs and Food Stamp programs—and above to yield the monthly savings: thus fewer would be deterred—because of the • $500 for the CalWORKs grant severity of the consequences, which would entail • $50 for CalWORKs administration being disqualifi ed from receiving aid permanently. • $30 for Food Stamp program administration However, Social Services was unable to provide any data to support its theory or the use of 33 percent Social Services then multiplies the monthly as opposed to some other percentage. This use of a savings by the average length of time a recipient typically stays on aid: rate that is not based on sound statistical reasoning • 42 months for CalWORKs applicants also causes us to question the validity of Social Services’ savings estimate. • 21 months for existing CalWORKs cases • 14 months for food stamp applicants Finally, Social Services applied Los Angeles • 7 months for existing food stamp cases County’s AFIRM demonstration project’s monthly 3. Total the amount of savings for each recidivism rates, the rate at which individuals program component. previously terminated from receiving aid from its AFDC program returned to aid, to the adjusted caseload and applications for the State’s CalWORKs and Food Stamp programs. Again, we are concerned about Social Services’ assumptions that Los Angeles County’s experience holds true in other counties. As we indicated previously, state policy requires the data center and Social Services to prepare a post implementation evaluation report to measure the costs and benefi ts of using 3300 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3311 SFIS. The Department of Finance requires agencies to maintain documentation supporting the project for at least two years Given the potential following approval of the post implementation evaluation understatement of SFIS report. Given the potential understatement of SFIS costs, the costs, the flaws in its flaws identified in Social Services’ methodology for estimating methodology for estimating the savings attributable to SFIS, and the lack of documentation savings, and the lack of to support its assumptions, Social Services will have difficulty documentation to support establishing the cost-effectiveness of SFIS. its assumptions, it will be difficult for Social Services to establish the cost- DECISION MAKERS SHOULD CONSIDER THE BENEFITS effectiveness of SFIS. AND DRAWBACKS OF SFIS WHEN DECIDING FUTURE FUNDING FOR THE SYSTEM The primary benefits that the State derives from continuing to use SFIS are the proven effectiveness of fingerprint imaging technology to identify duplicate fingerprints and its ability to identify applicants who may travel from county to county seeking duplicate aid. However, several factors could also support discontinuing the use of SFIS. For one, the State is spending $11.4 million or more annually to operate SFIS without knowing the actual savings that it may be producing. Additionally, although we were not able to verify some of the concerns that opponents of SFIS raised, other concerns appear valid. For example, the fingerprint imaging requirement may add an element of fear to the welfare application process and thus may keep some eligible people from applying for needed benefits. The State must weigh these factors in deciding whether to continue to fund SFIS. SFIS’s Primary Benefit Is the Effective Use of Fingerprint Imaging Technology to Detect Fraud Social Services began rolling out SFIS to the 58 counties in March 2000 to detect duplicate-aid fraud, which occurs when an individual is receiving aid under two or more active accounts. Fingerprint imaging technology is an effective tool for detecting duplicate-aid fraud and in a few instances has also allowed county staff to identify other types of fraud such as identity theft or the use of falsified documents. Law enforcement agencies and the public widely accept fingerprint imaging as a reliable means of human recognition. The Electronic Data Systems Corporation (EDS), under its contract terms, conducted accuracy tests during the implementation of SFIS. The results of these tests indicate that when applicants 3322 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3333 have two fingerprint images on file, SFIS’s matching subsystem, the Automated Finger Identification System, is more than 99 percent accurate for both an open fingerprint image search, When individuals which compares a new or existing image with all other images applying for benefits in SFIS, and a closed search, which compares a fingerprint under the Food Stamp or image in SFIS against the image an applicant is providing in CalWORKs programs have person. Therefore, when individuals apply for benefits under the two fingerprint images CalWORKs or Food Stamp programs, SFIS will detect whether on file, SFIS will detect they are already receiving benefits in any county throughout with 99 percent accuracy the State. SFIS is also helpful in detecting other types of fraud. whether they are already For example, Sacramento County reported that an SFIS match receiving benefits in any in calendar year 2000 led to the detection of recipients who county throughout had received roughly $103,000 by providing fraudulent birth the State. certificates for nonexistent children. Moreover, according to Social Services, 24 counties also use SFIS in their General Assistance/General Relief programs. Finally, 32 of the counties responding to our survey felt that SFIS was effective in deterring duplicate-aid fraud, although they were unable to quantify the savings associated with this deterrence. The State Could Also Realize Benefits by Discontinuing the Use of SFIS Although SFIS provides some benefits to the State, it also has some drawbacks, such as the high number of administrative errors that occur, the relatively low number of actual cases of fraud it detects, future operational costs estimated to be at least $11.4 million annually, and concerns raised by SFIS opponents. Decision makers should consider these drawbacks when they determine whether to fund SFIS in the future. The Majority of Matches SFIS Identifies Are Administrative Errors, and the Actual Level of Fraud It Detects Is Quite Small Although Social Services does not know how many applicants SFIS deters from attempting to receive duplicate aid, it can determine the number of applicants that SFIS detected who were attempting to receive duplicate aid. However, we found that the actual number of matches SFIS has identified as possible fraudulent activity is substantially fewer than the number of matches it identifies as administrative errors made by county staff. Moreover, because Social Services has not ensured that counties adequately report the results of their investigations, it also does not know the true extent of duplicate-aid fraud detected by SFIS. 3322 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3333 SFIS completes an open and closed fingerprint image search and enters any matches into a resolution queue. A match occurs when another recipient in the database has the same fingerprint According to SFIS data, image or when the recipient’s fingerprint image does not match between March 1, 2000, the one on file. Typically, a staff member in each county welfare and September 30, 2002, department, such as an eligibility supervisor, will research the fraud was found in just items in the queue and determine whether they are due to 45, or 0.2 percent, of the administrative error or possible fraud. An administrative error 18,157 items with a can occur if someone keys incorrect names and Social Security final disposition. numbers into SFIS or a recipient moves to another county. County welfare department staff correct any administrative errors they find and forward possible fraud items to fraud investigators. We found that the majority of the items in the resolution queue are administrative errors. Specifically, Figure 6 shows that between March 1, 2000, and September 30, 2002, SFIS detected a total of 25,202 matches, 7,045 which were still pending resolution as of September 30, 2002. Of the remaining 18,157 items with a final disposition, staff identified only 478 of the items, or roughly 3 percent, as possible fraud situations. Further, investigators found fraud in only 45 of the 478 possible fraud items, just 0.2 percent of the 18,157 items resolved, according to SFIS reports. FIGURE 6 Disposition of SFIS Resolution-Queue Items March 1, 2000, Through September 30, 2002 �������� ����� ������� ������� ���������� ���������� �������������� ��������� ����������� Source: Fraud Bureau monthly resolution reports, March 1, 2000, to September 30, 2002. 3344 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3355 These results are consistent with reports from some counties that the majority of their SFIS matches are administrative errors. The chief of the Fraud Bureau is not surprised that the actual level of fraud detected by SFIS is small, because most of its value is in deterring individuals from obtaining duplicate aid. Two of the four states using fingerprint imaging systems have estimated that the deterrent rate attributable to their systems is less than 1 percent. For example, in 1997 the University of Texas at Austin (university) conducted an evaluation of the state of Texas’ Lone Star Image System demonstration project over the first seven months of operation. The university found that the demonstration project had not reduced caseloads significantly by detecting or deterring duplicate aid and that it yielded no savings in benefit payments. Studies that the state of Texas conducted in 1998 and 1999 indicate that the deterrent rate of its fingerprint imaging system is between 0.44 percent and 0.94 percent of its new applications and recertifications. Additionally, the state of Arizona estimated that in fiscal year 2001, roughly 0.83 percent of the cases it closed or denied were attributable to its fingerprint imaging system. These states were able to estimate a deterrent rate because they had established a specific code when they first implemented the fingerprint imaging system to identify and track the number of applicants and recipients that did not complete the fingerprint imaging requirement. However, as we discussed previously, because the data center and Social Services did not collect that same key data during the implementation of SFIS, the State will never be able to substantiate the claim that the chief of the Fraud Bureau made regarding the deterrent value of SFIS. Finally, in order to determine how long items had been pending resolution, we asked the data center to produce a report identifying those items that were still awaiting resolution Roughly 3,000 of the as of October 21, 2002. We found that roughly 3,000 of 4,920 matches shown as the 4,920 matches shown as pending resolution in SFIS were pending resolution in SFIS more than 99 days old, and 1,100 had been pending for were more than 99 days a year or more. Data-center staff attribute the reduction old, and 1,100 had been of items pending resolution between September 30 and pending for a year or more. October 21 to Los Angeles County’s reclassification of almost 2,600 items. Nevertheless, 4,920 items in the queue are still pending resolution. Social Services told us that it generates monthly reports from SFIS that allow it to see whether counties are investigating and resolving discrepancies but that it reviews these reports in detail only twice a year. According to Social Services, one reason that so many items still await resolution is that the counties are investigating the matches but not clearing 3344 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3355 the resolution queue once they have resolved a case. However, although Social Services provides training and instructs counties to promptly resolve any matches that SFIS identifi es, it does not have a regulation, policy, or set of procedures requiring counties to do so. Additionally, Social Services has yet to develop written procedures for staff to follow when reviewing reports that SFIS generates. Social Services told us that it would like to contact county SFIS administrators to remind them to investigate pending items, but staffi ng shortages and its workload prevent it from doing so. Without policies and procedures, Social Services cannot ensure that SFIS information remains current, which can diminish its usefulness. Although We Could Not Confi rm Some of the Concerns That SFIS’s Opponents Raised, Other Concerns Appear Valid Opponents of SFIS express concerns that the State’s fi ngerprint imaging requirement is one of several key barriers to participation in the Food Stamp program, especially for the immigrant population. The opponents also assert that the most diffi cult fear for immigrants to overcome is the belief that law enforcement agencies and the Immigration and Naturalization Service (INS) will use their photo and fi ngerprint images. Finally, opponents question whether the benefi t of fi ngerprint imaging outweighs the negative Characteristics of Food Stamp Households impacts of hunger in the State. Fifty-one percent of all food stamp recipients The federal Food Stamp Program is an entitlement are children, and another 10 percent are age 60 program. Specifi cally, federal law provides that or older. eligible households within each state shall be Twenty-three percent of food stamp households given an opportunity to obtain a more nutritious receive TANF benefi ts. diet through the issuance of an allotment. Twenty-seven percent of all food stamp Opponents of the State’s fi ngerprint imaging households have earnings, which are their primary source of income. requirement say that their chief concern is the degree to which SFIS deters eligible, hungry Eleven percent of all food stamp households are above the poverty line, while 35 percent families from participating in the Food Stamp have incomes at or below half the poverty line. program, especially among California’s large immigrant population. However, although the opponents were able to provide us with some Source: USDA, “Offi ce of Analysis, Nutrition, and Evaluation Fiscal Year 2001 (Advance Report).” information concerning their experiences in working with applicants, they did not provide any quantifi able evidence that would help us determine whether SFIS deters eligible individuals from applying for benefi ts. Moreover, of the 49 counties that expressed an opinion on whether or not SFIS deters eligible people from applying, 34 felt that it did not. 3366 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3377 Given the existing data, we were unable to determine whether SFIS deters eligible people from applying for benefits. For example, although SFIS tracks information relating to an applicant’s name and birth date, it does not record an applicant’s immigration status; hence, we could not determine whether or not eligible immigrants applied. Furthermore, as we mentioned previously, Based on the available the data center and Social Services did not require collection of data, we cannot concur certain data during implementation of SFIS, such as tracking with the assertions denials of applicants or discontinuations of aid to recipients who that SFIS deters eligible failed to comply with the fingerprint imaging requirement. Only noncitizen applicants 16 of the 57 counties responding to our survey stated that their from participating in the eligibility systems could be used to track the data of applicants Food Stamp program. who refused to comply with SFIS fingerprinting requirements. Of these 16 counties, only Riverside County tracked citizenship data. Its data show that the majority of denials and discontinuances of aid attributable to SFIS were for persons who were citizens. Statewide demographic data also suggest that the proportion of citizens to noncitizens has not changed significantly among recipients of the CalWORKs and Food Stamp programs since SFIS was implemented. Figure 7 on the following page shows the breakdown of this data for each program. Thus, based on the available data, we cannot concur with the assertions that SFIS deters eligible noncitizen applicants from participating in the Food Stamp program. Opponents also told us that SFIS is a source of great fear for immigrants, many of whom are already reluctant to use food stamps because of their concerns that the State will share their photo and fingerprint images with the INS, which will use these to make public-charge determinations. The INS can find that an immigrant has become a public charge when an immigrant is likely to become primarily dependent on the government for subsistence, as demonstrated by the receipt of public cash assistance for income maintenance or by institutionalization of the individual for long-term care at government expense. An individual found to be a public charge may be ineligible to adjust his or her immigration status to legal permanent resident or may face deportation. Although benefits received under the federal Food Stamp Program are not subject to public-charge consideration, prior to January 2002, Social Services’ Food Stamp program applications did not contain sufficient information to address immigrants’ concerns. Social Services recognized this deficiency and in January 2002 revised its application to highlight important information for immigrants. For example, Social Services’ applications now clearly state that receiving food 3366 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3377 stamp benefits will not affect the applicant’s or the family’s immigration status. Further, the applications also stress that the immigration information that applicants provide is private and confidential. Thus, Social Services’ efforts may mitigate the fear immigrants have that obtaining food stamp benefits will affect their immigration status. FIGURE 7 Recipients by Citizenship Status ���� �� �� �� �� � ���� ���� ���� ����� ���� ���� ���� ����� ����������� �������� ��������������� Source: Unaudited Social Services Research and Development Division data. * In all years except 2000, citizenship status is unknown for less than 1 percent of recipients in the Food Stamp program. † Federal fiscal year 2001 is preliminary data that is pending Social Services’ final review. ‡ Federal fiscal year is October 1 to September 30. SFIS implementation occurred between March 14, 2000, and December 31, 2000. The immigrant’s fear of being determined a public charge because he or she receives food stamp benefits is unfounded. However, immigrants’ concerns about other government agencies’ use of information are potentially valid. Specifically, state law mandating fingerprint imaging specifies that Social Services, county welfare agencies, and all others shall not use or disclose the images for any purposes other than the prevention or prosecution of fraud. Thus, Social Services believes 3388 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3399 ������������������� ���������� ������� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� that it may only disclose fingerprint and photo images for the purpose of prevention or prosecution of fraud. Further, according to counsel, if Social Services were served with a subpoena or search warrant for those records in connection with some other purpose, it would not release the records without a court order. Nonetheless, we are aware of instances in which law enforcement, for example, has been provided access to records collected by government agencies even when the law authorizing collection of the records specifically limits the use of the records to a particular program. Thus the opponents may have some basis for their concern that the use of the images may not always be limited to the prevention or prosecution of welfare fraud. Finally, opponents question whether the benefits of fingerprint imaging technology outweigh the negative impacts of hunger in California. They recognize the need as well as the federal requirement for duplicate-aid fraud detection in the Food Stamp program, but they wonder if methods used in other states would be more appropriate and allow better access to the program. The majority of states rely on other methods to detect duplicate-aid fraud, such as preventing fraud at the front end by using computer matching and visits to applicants’ schools and neighbors. In addition, Social Services and the counties continue to use these methods as part of the eligibility determination process. The federal government has made efforts to reduce the negative impacts of hunger. The USDA monitors food security in U.S. households through an annual survey conducted by the SFIS opponents question United States Census Bureau. Its October 2002 report, titled whether the benefits Household Food Security in the United States, 2001, states that of fingerprint imaging about one-third of the nation’s 11.5 million food insecure technology outweigh people, or 3.5 million households, were food insecure to the the negative impacts of extent that one or more household members were hungry at hunger in California. least sometime during the year because they could not afford enough food. The USDA defines food insecurity as households that at some time during the year were uncertain of having, or unable to acquire, enough food for an active, healthy life for all household members. The report also points out that roughly 49 percent of these households with annual incomes less than 185 percent of the poverty line––which was $17,960 for a family of four in 2001––did not participate in any federal food assistance programs such as the federal Food Stamp Program. In calendar years 2001 and 2002 the USDA awarded almost $9 million to various organizations, including three in California, to improve access to the federal Food Stamp Program. These grants were used to target populations such as immigrants, the elderly, and minorities that do not speak English. 3388 California State Auditor Report 2001-015 California State Auditor Report 2001-015 3399 Similarly, California’s Legislature voiced its concern over low participation rates by requiring Social Services to develop a community outreach and education campaign to help families learn about and apply for the Food Stamp program. In an annual report to the Legislature dated April 1, 2002, Social Services stated that it believes its outreach efforts have had an effect on increasing the number of applications received and the caseload of the Food Stamp program. However, the Legislature specifically instructed Social Services to identify target populations and report on the results of its outreach efforts. Social Services identified two target populations: families terminating from CalWORKs and legal noncitizens. Although Social Services recognizes that the ultimate measurement of its outreach efforts’ success depends on its ability to reach the target population, it did not collect data to evaluate the participation rates of these two populations. Instead, it chose to rely on the USDA’s report of estimated state Food Stamp program participation rates, which presents information that is up to three years old. Furthermore, the USDA’s report does not have information specific to Social Services’ target populations. Therefore, Social Services does not know if its efforts to reach legal noncitizens have been successful. The manager of policy development for the Food Stamp Bureau told us that he plans to study the feasibility of developing a report to measure the participation rates of the target population. If SFIS does in fact deter eligible applicants, its use seems inconsistent If SFIS does in fact deter with the purpose of the Food Stamp program and the USDA’s and eligible applicants, its use California’s concerns with low participation. Thus, the Legislature seems inconsistent with the must decide whether the benefit of using SFIS outweighs its effect purpose of the Food Stamp on the Food Stamp program. Future costs for SFIS, at a minimum, program and the USDA’s will be at least $11.4 million annually, and this estimate does and California’s concerns not include the actual costs the counties would incur, for reasons with low participation. we discussed previously. The majority of this estimate, roughly $10.7 million, includes costs from EDS and the data center to operate and maintain SFIS. The data center estimates that if the Legislature were to eliminate the fingerprint imaging requirement, the State would bear costs of roughly $1.7 million. The majority of these costs would result from EDS’ removal of servers and workstations at sites located throughout the State. Social Services believes that discontinuing SFIS would require an adjustment to its budget for all or a portion of the savings previously included to account for the gradual return of those individuals SFIS deters. However, as we discussed previously, we were unable to assess SFIS’s effectiveness in deterring duplicate-aid fraud and the savings 4400 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4411 attributable to SFIS are unknown. The State’s contract with EDS expires on September 6, 2003, which allows the Legislature an opportunity to evaluate the necessity of SFIS. RECOMMENDATIONS To ensure that it reports accurate and complete information to the USDA, Social Services should require the Fraud Bureau to incorporate the review of DPA 266 data into its on-site visits to counties. To ensure that implementation of future IT projects meets state expectations, Social Services and the data center should do the following: • Collect sufficient data to measure the benefits and costs against the project objectives. • Identify promptly any obstacles that may prevent them from implementing the project effectively. Additionally, to improve its management of SFIS, Social Services should do the following: • Identify the full costs of operating SFIS by requiring counties to track their administrative costs separately. • Establish policies and procedures that require counties to resolve pending items in the resolution queue promptly. Additionally, the Fraud Bureau should develop written procedures for its staff to follow up on items pending in the resolution queue. The procedures should include Fraud Bureau staff requesting a monthly aging report to use as a tool to determine whether items pending in the resolution queue are current and, if necessary, contacting the appropriate counties. • Ensure that counties investigate and record the outcomes of their investigations in SFIS. To ensure that its estimates are representative of the entire state and its key assumptions are defensible, Social Services should do the following: • Study the conditions of a sample of counties instead of assuming that conditions in one county hold true in other counties. 4400 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4411 • Maintain adequate documentation, such as time studies or other empirical data, to support its estimates. To report accurately the results of its community outreach and education efforts to the Legislature, Social Services should establish a mechanism to track the participation rates of the target populations. The Legislature should consider the pros and cons of repealing state law requiring fingerprint imaging, including whether SFIS is consistent with the State’s community outreach and education campaign efforts for the Food Stamp program. To assist the Legislature in its consideration of the pros and cons of repealing state law requiring fingerprint imaging, Social Services and the data center should report on the full costs associated with discontinuing SFIS. We conducted this review under the authority vested in the California State Auditor by Section 8543 et seq. of the California Government Code and according to generally accepted government auditing standards. We limited our review to those areas specified in the audit scope section of this report. Respectfully submitted, ELAINE M. HOWLE State Auditor Date: January 2, 2003 Staff: Joanne Quarles, CPA, Audit Principal David E. Biggs, CPA Renju P. Jacob Paul P. Zahka 4422 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4433 APPENDIX Selected Information From Our County Survey As part of our efforts to evaluate the effectiveness of the Department of Social Services’ Statewide Fingerprint Imaging System (SFIS) in detecting duplicate-aid fraud, we surveyed all 58 counties to determine, among other things, the level of duplicate-aid fraud detected before and after the implementation of SFIS. Beginning on page 44, we present a copy of the survey that we sent to the counties. We asked the counties to coordinate the completion of the survey using information from their social services and fraud investigation departments or units. Although we did not independently test the accuracy of the information provided to us in the surveys, we believe the counties’ responses provide valuable information. In addition to including information reported by the counties, we summarize, beginning on page 52, the counties’ responses to questions that called for a yes or no answer or that asked counties to provide their opinions on a scale of 1 to 10. We received responses from 57 counties; Santa Cruz was the only county that did not respond. 4422 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4433 CALIFORNIA STATE AUDITOR BUREAU OF STATE AUDITS STATEWIDE FINGERPRINT IMAGING SYSTEM (SFIS) COUNTY SURVEY GENERAL INFORMATION County Name: _____________________________________________________________________________ Name of person responsible for completing the survey (please print): _____________________________________________________________________________ Position Title (e.g., Chief Counsel, Executive Officer, etc.): _____________________________________________________________________________ Telephone number where you can be reached (including area code): _____________________________________________________________________________ DUPLICATE AID FRAUD DETECTION PRIOR TO SFIS 1. Please provide a complete description of how your county identified duplicate aid fraud prior to SFIS. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 4444 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4455 2. a) Please submit the following data for duplicate aid fraud cases identified by your county prior to SFIS. Number of Number of Fraud Cases Fraud Cases Dollar Value Substantiated Substantiated Total Costs of Fraud That Were That Were Number of Number of Number of Associated Associated Referred to Adjudicated by a Possible Possible Possible With With the District County Fraud Fraud Cases Fraud Cases Substantiated Substantiated Attorney for Administrative Period* Cases Investigated Substantiated Cases Cases Prosecution Process 7/95 – 6/96 _________ _________ _________ _________ _________ _________ _________ 7/96 – 6/97 _________ _________ _________ _________ _________ _________ _________ 7/97 – 6/98 _________ _________ _________ _________ _________ _________ _________ 7/98 – 6/99 _________ _________ _________ _________ _________ _________ _________ 7/99 – 6/00 _________ _________ _________ _________ _________ _________ _________ 7/00 – 6/01 _________ _________ _________ _________ _________ _________ _________ * Provide data through the date of your conversion to SFIS b) If the number of possible fraud cases reported above is different than the number of possible fraud cases investigated, please briefly explain the major causes of the differences. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ c) If your county uses an administrative process to adjudicate substantiated fraud cases, please describe that process. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 3. How many investigators on average were assigned to work on duplicate aid fraud cases prior to SFIS? Full Time: _______________ Average Hourly Pay Rate: _______________ Average Hours Per Case: _______________ Part Time: _______________ Average Hourly Pay Rate: _______________ Average Hours Per Case: _______________ 4444 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4455 4. Please provide a complete description of the method used by fraud investigators to track the status of possible fraud cases under investigation. For example, if fraud investigators use an automated system, please describe or attach documentation that describes how investigators enter a case into the system, monitor its progress, and record its final disposition. Also, please indicate to whom the investigators must report their results, if applicable. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ DUPLICATE AID FRAUD DETECTION SINCE SFIS IMPLEMENTATION 5. Please describe how your county resolves fingerprint imaging matches and include a brief description of the various potential outcomes, such as administrative errors, fraud, or expected matches (matches expected to be reported by SFIS because the applicant told county staff that he or she had previously received benefits and had been fingerprinted). ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 6. a) Please submit the following data for duplicate aid fraud cases identified by your county since its conversion to SFIS. Number of Number of Fraud Cases Fraud Cases Dollar Value Substantiated Substantiated Total Costs of Fraud That Were That Were Number of Number of Number of Associated Associated Referred to Adjudicated by a Possible Possible Possible With With the District County Fraud Fraud Cases Fraud Cases Substantiated Substantiated Attorney for Administrative Period* Cases Investigated Substantiated Cases Cases Prosecution Process 7/99 – 6/00 _________ _________ _________ _________ _________ _________ _________ 7/00 – 6/01 _________ _________ _________ _________ _________ _________ _________ 7/01 – 6/02 _________ _________ _________ _________ _________ _________ _________ * Provide data since the date of your conversion to SFIS 4466 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4477 b) If the number of possible fraud cases reported above is different than the number of possible fraud cases investigated, please briefly explain the major causes of the differences. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ c) If your county uses an administrative process to adjudicate substantiated fraud cases, please describe that process. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 7. a) Please submit the following data for duplicate aid fraud cases identified by other sources (not SFIS) since your county’s conversion to SFIS. Number of Number of Fraud Cases Fraud Cases Dollar Value Substantiated Substantiated Total Costs of Fraud That Were That Were Number of Number of Number of Associated Associated Referred to Adjudicated by a Possible Possible Possible With With the District County Fraud Fraud Cases Fraud Cases Substantiated Substantiated Attorney for Administrative Period* Cases Investigated Substantiated Cases Cases Prosecution Process 7/99 – 6/00 _________ _________ _________ _________ _________ _________ _________ 7/00 – 6/01 _________ _________ _________ _________ _________ _________ _________ 7/01 – 6/02 _________ _________ _________ _________ _________ _________ _________ * Provide data since the date of conversion to SFIS b) If the number of possible fraud cases reported above is different than the number of possible fraud cases investigated, please briefly explain the major causes of the differences. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 4466 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4477 c) If the number of possible fraud cases investigated reported above is different than the number of possible fraud cases substantiated, please briefly explain the major causes of the differences. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 8. How many investigators on average have been assigned to work on duplicate aid fraud cases since SFIS’ implementation in your county? Full Time: _______________ Average Hourly Pay Rate: _______________ Average Hours Per Case: _______________ Part Time: _______________ Average Hourly Pay Rate: _______________ Average Hours Per Case: _______________ 9. Has your county performed an analysis of the savings resulting from SFIS? (If yes, please enclose a copy of your analysis or describe below.) Yes c No c ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 10. Has your county quantified the deterrent value of duplicate aid fraud using the SFIS? If yes, please attach a copy of your analysis. If no, please describe how you would attempt to determine the deterrent value of SFIS. Yes c No c ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 11. Did your county maintain statistics on denials or discontinuances resulting from an applicant’s or recipient’s failure to comply with the State’s fingerprint imaging requirement during your implementation of SFIS? Yes c No c If yes, please provide us with a copy of your analysis. Also, if possible, show a breakdown by ethnicity and immigration status. 4488 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4499 ELIGIBILITY SYSTEM 12. Does your county have an automated eligibility system that is linked to, and interfaces with SFIS? Yes c No c If yes, please attach a document describing the interface. 13. Does your automated eligibility system have a feature to detect whether recipients of CalWORKS or Food Stamps have not been fingerprinted? Yes c No c If yes, please describe. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 14. Does your automated eligibility system track denials or discontinuances resulting from possible duplicate aid fraud identified by SFIS? Yes c No c If yes, please submit data in the following format: Number of Number of Discontinuances Number of Denials Resulting Number of Resulting From Fiscal Year* Denials From SFIS Discontinuances SFIS ________ ________ ________ ________ ________ ________ ________ ________ ________ ________ * Provide data for each fiscal year since your conversion to SFIS. If possible, please show a breakdown by ethnicity and immigration status. 4488 California State Auditor Report 2001-015 California State Auditor Report 2001-015 4499 15. How does your county ensure that an applicant who has previously received aid in another county is no longer receiving aid from that county? ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 16. Does your county have procedures in place to ensure that fingerprint imaging information is kept confidential in accordance with Welfare and Institutions Code, Section 10850? Yes c No c If yes, please describe your procedures. ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 17. Does your county provide information to applicants for CalWORKS and Food Stamps that explains the confidentiality of the fingerprint images stored in SFIS? (Please describe and enclose examples of any available handouts). Yes c No c ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 18. Does your county have a system that tracks those applicants who refuse to comply with the fingerprint imaging requirements? Yes c No c If yes, please provide a list showing the number of applicants and their ethnicity. 5500 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5511 COUNTY OPINIONS Based on your county’s experience, please indicate on the following scales your level of agreement or disagreement with the following statements: 19. Attempts to commit duplicate aid fraud have been deterred by SFIS. Strongly Strongly No Disagree Disagree Neutral Agree Agree Opinion 1 2 3 4 5 6 7 8 9 10 c c c c c c c c c c c 20. Eligible recipients have been deterred from applying for benefits by SFIS. Strongly Strongly No Disagree Disagree Neutral Agree Agree Opinion 1 2 3 4 5 6 7 8 9 10 c c c c c c c c c c c 21. What are some of the most common complaints about SFIS your county encounters from applicants? ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ 22. What are some of the most common complaints about SFIS your county encounters from county staff working with the system? ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ ________________________________________________________________________________________ Return completed survey by September 6, 2002 5500 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5511 SUMMARY OF RESPONSES We summarize below the data that was easily quantifiable across counties, such as the responses to questions that requested yes or no answers or a county opinion from a scale from 1 to 10. Question 9: Has your county performed an analysis of the savings resulting from SFIS? Yes 1 county No 55 counties No opinion 1 county Question 10: Has your county quantified the deterrent value of duplicate-aid fraud using SFIS? Yes 1 county No 55 counties No opinion 1 county Question 11: Did your county maintain statistics on denials or discontinuances resulting from an applicant’s or recipient’s failure to comply with the State’s fingerprint imaging requirement during your implementation of SFIS? Yes 17 counties No 37 counties No opinion 3 counties Question 12: Does your county have an automated eligibility system that is linked to and interfaces with SFIS? Yes 5 counties No 50 counties No opinion 2 counties 5522 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5533 Question 13: Does your automated eligibility system have a feature to detect whether recipients of CalWORKs or food stamps have not been fingerprinted? Yes 29 counties No 25 counties No opinion 3 counties Question 14: Does your automated eligibility system track denials or discontinuances resulting from possible duplicate-aid fraud identified by SFIS? Yes 12 counties No 44 counties No opinion 1 county Question 16: Does your county have procedures in place to ensure that fingerprint imaging information is kept confidential in accordance with Welfare and Institutions Code, Section 10850? Yes 55 counties No 1 county No opinion 1 county Question 17: Does your county provide information to applicants for CalWORKs and food stamps that explains the confidentiality of the fingerprint images stored in SFIS? Yes 54 counties No 3 counties No opinion 0 counties 5522 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5533 Question 18: Does your county have a system that tracks those applicants who refuse to comply with the fingerprint imaging requirements? Yes 16 counties No 41 counties No opinion 0 counties Question 19: Attempts to commit duplicate-aid fraud have been deterred by SFIS. Strongly disagree (1-2) 5 counties Disagree (3-4) 3 counties Neutral (5-6) 11 counties Agree (7-8) 16 counties Strongly agree (9-10) 16 counties Mixed opinion 2 counties * No opinion 4 counties *Two counties provided two different responses to this question because its program staff and fraud investigation staff did not agree. Specifically, program staff indicated that they disagreed with the statement while fraud investigation staff agreed with it. Question 20: Eligible recipients have been deterred from applying for benefits by SFIS. Strongly disagree (1-2) 21 counties Disagree (3-4) 13 counties Neutral (5-6) 10 counties Agree (7-8) 2 counties Strongly agree (9-10) 0 counties Mixed opinion 3 counties* No opinion 8 counties *One county disagreed with the statement for CalWORKs but agreed for its Food Stamp program. Two other counties provided two different responses because its program staff and fraud investigation staff did not agree. In one instance, program staff agreed with the statement while fraud investigation staff were neutral. In the other instance, program staff agreed with the statement while fraud investigation staff disagreed with it. 5544 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5555 Agency’s comments provided as text only. Health and Human Services Agency 1600 Ninth Street, Room 460 Sacramento, CA 95814 December 16, 2002 Elaine M. Howle, State Auditor* 555 Capitol Mall Suite 300 Sacramento, CA 95814 Dear Ms. Howle: We have reviewed your draft report entitled “Statewide Fingerprint Imaging System: Because the Department of Social Services Never Established the Need for the System and Missed the Opportunity to Measure its Cost Effectiveness, Other Factors Should Be Considered When Determining Future Funding for the System.”† I am transmitting the enclosed cover letter and comments generated by the California Department of Social Services. Sincerely, (Signed by: Grantland Johnson) GRANTLAND JOHNSON Enclosures * California State Auditor’s comments begin on page 65. † Title refers to earlier draft version of report. 5544 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5555 1 2 3 1 13 4 555666 CCaalliiffoorrnniiaa SSttaattee AAuuddiittoorr RReeppoorrtt 22000011--001155 California State Auditor Report 2001-015 5577 4 5 5566 California State Auditor Report 2001-015 CCaalliiffoorrnniiaa SSttaattee AAuuddiittoorr RReeppoorrtt 22000011--001155 555777 CALIFORNIA DEPARTMENT OF SOCIAL SERVICES (CDSS) RESPONSE TO BUREAU OF STATE AUDITS REPORT ENTITLED: “STATEWIDE FINGERPRINT IMAGING SYSTEM (SFIS): BECAUSE THE DEPARTMENT OF SOCIAL SERVICES NEVER ESTABLISHED THE NEED FOR THE SYSTEM AND MISSED THE OPPORTUNITY TO MEASURE ITS COST EFFECTIVENESS, OTHER FACTORS SHOULD BE CONSIDERED WHEN DETERMINING FUTURE FUNDING FOR THE SYSTEM” BSA 2002-015 I. AUDIT RESULTS Before It Implemented SFIS, Social Services Did Not Determine the Extent of Duplicate Aid Fraud Throughout the State, and Therefore It Did Not Properly Establish the State’s Need for a Fingerprint Imaging System. The Bureau of State Audits (BSA) concludes that SFIS was implemented by CDSS without an understanding of the extent of duplicate aid fraud or the potential cost-benefit of a statewide system. The Department disagrees with this conclusion. The results of Los Angeles County’s Automated Fingerprint Imaging and Match (AFIRM) system evaluation by an independent contrac- 6 tor confirmed that duplicate aid was a significant and costly problem. The evaluation found 22.6 percent of the cases investigated were verified or determined with high probability to be multiple- 2 case fraud. The federal government also recognized that duplicate aid was a problem and required states to implement procedures to prevent receipt of it. Taking into account the AFIRM cost-benefit results, federal direction, and fingerprint imaging cost-benefit data from other states, state law was passed which required implementation of a statewide system. In addition, in 1995, California followed in the steps of Arizona, Texas, New York, and Connecticut and developed a fingerprint and photo imaging system as a way to meet federal requirements. The BSA reports that Maryland and Illinois determined fingerprint imaging was not cost-effective, partly due to their limited caseloads, and draws a comparison between California and these two states. A 7 more accurate comparison would be with larger states, such as New York and Texas, which have demonstrated fingerprint imaging to be cost-effective. The report suggests that the relatively low number of actual cases “detected” is a drawback of SFIS. The Department disagrees that the low number of cases detected by the system is a drawback. It is an expected result. “Preventing” duplicate aid is the primary benefit of such systems and is the 5 reason why other large states similar to California (i.e., New York and Texas) have implemented and are maintaining such systems. The report also asserts that the federal requirement to prevent duplicate aid can be met by using the existing Income and Eligibility Verification System, which uses social security numbers (SSNs) 8 as the primary client matching identifier. The Department disagrees with this conclusion. Relying on SSNs is not an acceptable way to prevent duplicate aid. False documents, including SSNs, can be readily obtained in many ways, including by purchase through the Internet. 5588 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5599 The Department also takes exception to the report narrative which consistently implies that CDSS made the decision on its own to implement fingerprint imaging using unreliable data. The direction 1 to implement fingerprint imaging was included in State legislation, Chapter 206, Statutes of 1996, after extensive analysis, discussion, and debate among stakeholders (e.g., county welfare depart- ments, advocate organizations, State control agencies, legislative staff and members, etc.). Section 10830. (a) states “the Department and the Health and Welfare Data Center shall design, implement and maintain a statewide fingerprint imaging system for use in connection with the determination of eligibility of benefits under the Aid to Families with Dependent Children (AFDC) program, and the Food Stamp Program”. Social Services Was Remiss in Not Heeding Advice to Adequately Evaluate SFIS and Dis- carded a Tentative Proposal for Evaluating SFIS 9 While the report accurately states that USDA did not approve federal funds for SFIS, it was not due to the federal concerns of a cost-benefit analysis as the report states. The Department’s decision to proceed without that funding was made because significant savings would be lost ($73,377 per calendar day) if the second procurement for the system was cancelled and the project had to go out to bid for a third time. SFIS had been competitively bid in August 1996. Pursuant to Proposition 209, the Wilson Administration made state contracting changes which conflicted with federal guide- lines. In July 1998, the Health and Welfare Data Center submitted a Request for Waiver to USDA to exclude these guidelines. The waiver was not granted. Ultimately, a decision was made to not cancel the second procurement and re-bid for a third time because of the lengthy period of time it would take to do so and the significant savings lost because of the delay. The report also faults the State for not using a cost-benefit analysis methodology identified in a November 1995 Business Plan. The Business Plan proposed in the preliminary conceptual design phase of the system had been developed prior to Los Angeles County completing the implementa- tion of its fingerprint imaging system and proceeding to have an independent evaluator review costs 3 and benefits. The February 1996 interim evaluation and July 1997 final evaluation of AFIRM pro- vided the necessary cost-benefit information, thus precluding a more protracted and costly evalua- tion. In Implementing SFIS Social Services Did Not Require Data Collection, Thus Insufficient Information Exists to Substantiate SFIS’ Cost-Effectiveness 0 Contrary to statements in the report, the evaluation of AFIRM utilized well-accepted, research and sampling techniques to arrive at its findings. Additional data collection would have required the development of costly system interfaces, reprogramming of existing county eligibility systems and 4 significant increases in local staffing costs. Data collection would have included extensive technical changes to systems at the State and local levels, and immediate investigations of applicants and recipients who did not show up for their fingerprint imaging appointment to determine the reason they did not comply. Additional county resources would also have been required to tabulate data and prepare reports for the State. Further, this all would have had to occur at a time when counties were heavily involved in redesign of their cash delivery systems resulting from the enactment of the California Work Opportunity and Responsibility to Kids (CalWORKs) Program. 5588 California State Auditor Report 2001-015 California State Auditor Report 2001-015 5599 The State is Spending $11.4 Million or More Annually to Operate SFIS Without Knowing the Actual Savings That It May Be Producing. The Department disagrees with the statement that “the methodology it used to estimate the State’s savings of roughly $150 million over five years for SFIS is flawed and therefore unreliable.” The Department’s estimating methodology for determining the costs and benefits of SFIS was and is based on reliable data as noted below: • As previously noted, prior to the legislative mandate to implement the system, Los Angeles 30 County, which represents approximately 40 percent of California’s public assistance caseload, had successfully implemented AFIRM as a system to prevent, detect and deter multiple case fraud for its General Assistance (GA) program. An independent evaluator conducted an in- depth analysis of the effect of AFIRM on the Aid to Families with Dependent Children (AFDC) caseload, which included the identification and longitudinal tracking of an experimental and q control group of AFDC cases. The conclusion of the independent analysis indicated that AFIRM provided a highly rewarding rate of return with a net savings of between $52.5 million and $64.6 million over a two-year period. • The “systematic” selection process used in the AFIRM evaluation (which includes randomized starting points and interval selection), is a well-accepted sampling methodology widely used in the research community. In addition to the important interval determination step, an integral 0 prior step in the process was the random selection of starting points. A randomly selected starting point is the truly randomized basis for a statistically valid sample. • The Department recognized the need to adjust the AFIRM 6.67 percent deterrent rate to account for the significant differences between the GA caseload and the CalWORKs caseload. A GA case represents a single adult as opposed to a CalWORKs case with an average family w size of 3.5. Adjusting the deterrent rate to equal 33 percent of the GA rate (or 2.2 percent) was a reasonable assumption given the number of persons impacted. • In developing the cost estimates for SFIS, the Department used reasonable assumptions to e account for county administrative time associated with SFIS. The estimate assumed it would take a county staff person five minutes to fingerprint and photo image an applicant. Based on the specifications for the fingerprint workstations, approximately twenty images per hour can be printed. Therefore, the assumption of five minutes to fingerprint and photo image an applicant is reasonable. The estimate did not include costs associated with investigating fraudulent activi- ties, as the value of the system is primarily deterrence of fraud. Based on the Department’s estimate, in order to offset the estimated ongoing costs of $12 million r annually, approximately 50 cases monthly would need to be detected or deterred. Based on Octo- ber 2001 through September 2002 CA 237 reports, this represents less than 1 percent (.14 per- cent) of the monthly CalWORKs applications. 6600 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6611 Advocate Concerns and SFIS Impact on Food Stamp Outreach Efforts and Program Participation The audit report also contends that concerns raised by SFIS opponents may be valid, and ques- tions whether the benefits of SFIS outweigh potential negative impacts on Food Stamp Program participation and outreach activities. The Department currently has no data demonstrating the impact that SFIS may have on food stamp participation. Please note, however, in Fiscal Year 2001- 02, non-assistance food stamp participation rates in California increased14 percent; it is projected that this caseload will increase by another 6 percent in the current year. II. REPORT RECOMMENDATIONS To ensure that it reports accurate and complete information to the USDA, Social Services should require the Fraud Bureau to incorporate the review of DPA 266 data into its annual onsite visits to counties. CDSS Response: CDSS concurs. Fraud Bureau staff will add this function to its visits. It should t be noted, however, that county site visits are not conducted annually. Generally, they are con- ducted at least once every three years with needed follow-up activities in the interim. To ensure that its implementation of future information technology (IT) projects meet State expectations, Social Services and the data center should collect sufficient data to measure the benefits and costs against the project objectives and promptly identify any obstacles that may prevent it from effectively implementing the project. CDSS Response: CDSS and the HHSDC concur. As has been the case on SFIS and all IT proj- ects, CDSS and HHSDC will continue to adhere to all appropriate IT policies and processes and identify obstacles that may prevent an appropriate analysis of pre and post impacts of the IT proj- ect. Social Services should identify the full costs of operating SFIS, by requiring counties to track their administrative costs separately. Social Services should establish policies and procedures that require counties to resolve promptly any pending items in the resolution queue. The Fraud Bureau should develop written procedures for its staff to follow up on items pending in the resolution queue. Social Services should ensure that counties investi- gate and record the outcomes of their investigations in SFIS. CDSS Response: The Department concurs with this recommendation in part. The Department agrees that it would be worthwhile to request a monthly aging report to use as a tool to determine if items pending in the resolution queue are current, and has already initiated this action. The Depart- ment will continue its efforts to ensure that counties promptly resolve pending items in the SFIS resolution queue. The Department will also assess the need for developing written procedures for Fraud Bureau staff. y However, the Department does not agree that counties should be required to separately track SFIS administrative costs. Under CDSS’ federally-approved Cost Allocation Plan, county costs associated 6600 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6611 with fingerprint imaging activities are captured under general eligibility determination activities. These costs are charged to CalWORKs, Food Stamps, Medi-Cal or to General Assistance, as appropriate. The Cost Allocation Plan also requires counties to record activities in 15 minute intervals. It is not reasonable to record activities in 5 minute increments, which would be the case if SFIS administrative costs had to be tracked separately. This change would require substantial system reprogramming and create significant additional workload for counties already strapped for resources. To ensure that its estimates are representative of the entire state and its key assumptions are defensible, Social Services should study the conditions of a sample of counties instead of assuming that conditions in one county hold true in other counties and maintain ade- quate documentation, such as time studies or other empirical data to support its estimates. CDSS Response: CDSS concurs that maintaining adequate documentation to support its esti- mates is important and has processes in place to assure that assumptions are appropriately docu- mented. The Department also concurs that, in most situations, a sample of counties would best represent conditions in the entire state. Depending on the situation, it is often necessary to prepare an estimate by other means, and use extrapolated data when actual data is unavailable. In the 13 SFIS estimate, AFIRM findings were used because they were actual results of a pilot in the State’s largest county, representing approximately 40 percent of the statewide caseload. The estimate was 0 considered valid and savings associated with it were the basis for enacting legislation to implement a statewide fingerprint imaging system. To report accurately the results of its community outreach and education efforts to the Legislature, Social Services should establish a mechanism to track participation rates of the target population. CDSS Response: CDSS concurs that a mechanism to track participation rates of target popula- tions would provide useful information to judge the results of community outreach and education efforts in the FSP. Non-citizens and persons leaving CalWORKs cash assistance are the Depart- ment’s target populations for outreach. The extent to which the Department will be able to continue or expand these efforts is contingent on resource availability (both staffing and funding). The Legislature should consider the pros and cons of repealing the state law requiring fin- gerprint imaging, including whether SFIS is consistent with the State’s Community outreach and education campaign efforts for the Food Stamp Program. This is an issue for the Legislature to consider. Please note that the Department currently has no data demonstrating the impact that SFIS may have on food stamps participation. To assist the Legislature in its consideration of the pros and cons of repealing the State law requiring fingerprint imaging, Social Services and the data center should report on the full costs associated with discontinuing SFIS. 6622 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6633 CDSS Response: CDSS concurs and has previously provided this information to the Legislature. u The CDSS budget assumes an annual savings of $68.7 million associated with the deterrence and detection of duplicate aid fraud through SFIS. Discontinuing SFIS would require that all or a portion of the savings be built back into the budget to account for the return to the program of the deterred caseload. The Department estimates that the elimination of SFIS July 1, 2003 would result in approximately $30 million in lost savings in the Budget Year, assuming the gradual return of the deterred caseload. The estimated costs to decommission the SFIS project are $1.7 million (mostly EDS services needed to support the system shutdown). The estimated annual ongoing costs as budgeted in the 2002 Budget Act to operate SFIS are approximately $12 million. Therefore, the net impact in the Budget Year to discontinue SFIS effective July 1, 2003, could be approximately $19.7 million. Additionally, DSS estimates that there could be future annual costs of over $60 million that would be attributable to ongoing, undeterred fraudulent duplicate aid. 6622 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6633 Blank page inserted for reproduction purposes only. 6644 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6655 COMMENTS California State Auditor’s Comments on the Response From the Department of Social Services To provide clarity and perspective, we are commenting on the response by the Department of Social Services (Social Services) to our audit report. The numbers below correspond to the numbers we placed in the margins of Social Services’ response. 1 Social Services is correct that our report makes no mention of the Legislature’s deliberations regarding the Statewide Fingerprint Imaging System (SFIS). On several occasions we asked Social Services to provide us with documentation to support its assertion of what took place during these deliberations, and it did not. Additionally, our review of the legislative analyses of the bill that introduced SFIS does not support Social Services’ assertions. Specifically, these analyses do not contain any discussions regarding a problem with duplicate aid throughout the State, information from other states, or the successful results of Los Angeles County’s Automated Fingerprint Image Reporting and Match (AFIRM) system. Thus, we lacked sufficient evidence to include these deliberations in our report. 2 Social Services’ characterization of the federal regulations may lead the reader to believe that fingerprint imaging is a requirement, which is not true. Although it is true federal regulations require all states to have a system in place to ensure that no individual participating in the federal Food Stamp Program commits duplicate aid fraud, as we state on page 13, federal regulations do not require the use of fingerprint imaging technology. In fact, as we state on page 14, only four other states use fingerprint imaging to detect duplicate-aid fraud. 3 Social Services is correct that an independent evaluation of the AFIRM system occurred; however, it fails to address the concerns raised about the independent evaluation data and the use of this data to justify SFIS. Specifically, as we state on page 23, in May 1996 the federal Department of Health and Human Services’ Office of Inspector General noted that if the AFIRM system was to be principally a tool to prevent, detect, and deter only duplicate-aid fraud, as opposed to other types of fraud, 6644 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6655 available data did not clearly demonstrate its cost-effectiveness. Moreover, as we state on page 24, in June 1998 the regional director of the federal Food Stamp Program expressed concern about the State’s inability to identify duplicate-aid population to be deterred by SFIS, its use of data from one large urban county in the State as a model for projecting statewide savings, and its use of data from Los Angeles County’s AFIRM system, which was not current and therefore was not an accurate representation of current circumstances. Finally, on page 24, we raise concerns that the sampling design used in the independent evaluation does not provide for a margin of error that would take into account any limitations in the sampling design or available data. Thus, the true range of savings attributable to AFIRM is unknown. 4 Social Services is attempting to downplay its responsibility for adhering to state policy outlined on page 26 that requires it to establish reporting and evaluation procedures for each approved information technology (IT) project and to prepare a post implementation evaluation report that measures the benefits and costs of a newly implemented IT system. Data collection is a key component in preparing the post implementation evaluation report. However, because of its failure to capture critical data during SFIS’s implementation, Social Services will have difficulty establishing the cost-effectiveness of SFIS. Furthermore, to plan an IT project without any consideration for how to measure its cost-effectiveness demonstrates poor planning by Social Services. 5 Social Services is incorrect. On pages 3 and 32, we acknowledge that the State derives benefits from the proven effectiveness of fingerprint imaging technology to identify duplicate fingerprints and its ability to identify applicants that travel from county to county seeking duplicate aid. Also, in response to Social Services’ concern that we did not include other large states in our discussion, on page 35 we point out that data from the states of Texas and Arizona indicate that the deterrent rate, or the rate of preventing duplicate aid, attributable to their systems is less than 1 percent. Moreover, the states of Connecticut and New York told us that they did not calculate a deterrent rate for their systems. Finally, as we state on page 35, because Social Services did not collect key data during SFIS’s implementation, the State will never be able to substantiate Social Services’ claim that the primary value of the SFIS is in deterring individuals from obtaining duplicate-aid fraud rather than detecting fraud that has occurred. 6666 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6677 6 Social Services is mischaracterizing the results of the final evaluation report of Los Angeles County’s AFIRM system. As we discuss on page 23, the report found that only 8, or roughly 6 percent, of the 137 discontinued cases in Los Angeles County were verified as actually receiving duplicate aid and that most of these cases were outside California. For the remaining 23 duplicate-aid fraud cases identified on page 23, or 16.6 percent, there was not conclusive evidence of duplicate-aid fraud. In fact, it was this datum that caused the federal Department of Health and Human Services’ Office of Inspector General to note that if AFIRM was to be principally a tool to prevent, detect, and deter only duplicate-aid fraud, as opposed to other types of fraud, available data did not clearly demonstrate its cost-effectiveness. 7 Contrary to Social Services’ belief, our report addresses adequately the results of other states. Specifically, on page 14 we indicate that the state of Texas estimated that in fiscal year 1999 its fingerprint imaging system generated annual savings for its Food Stamp Program ranging between $6 million and $12 million. We included data on the state of Texas because it sought guidance from the United States Department of Agriculture concerning its methodology. However, we did not include data for the state of New York because an independent evaluation of the methodology used to calculate savings for its fingerprint imaging system has never been published. 8 Social Services’ statement is inconsistent with the federal regulations relating to duplicate aid. Specifically, as we state on page 13, federal regulations require states to use names and Social Security numbers at a minimum to detect duplicate aid. Further, these regulations require states to implement other measures such as more frequent checks or increase their emphasis on prevention. Many states use computer matching to address the problem of duplicate-aid fraud. Further, the state of Maryland has been successful in addressing duplicate- aid fraud through an increased emphasis on preventing fraud in the front-end of the eligibility determination process. Thus, as we state on page 15, there are many ways to fulfill the federal requirement of detecting duplicate-aid fraud other than using fingerprint imaging technology. 9 Social Services’ recollection of past events conflicts with the historical record. The August 1998 document submitted to the governor’s office that we discuss on page 24, stated the following: “The development of a statewide fingerprint imaging system is mandated as a condition of eligibility for benefits in the 6666 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6677 CalWORKs and Food Stamp programs. The Department of Agriculture, Food and Nutrition Services (FNS) continues to limit approval to only Phase I of the SFIS project. FNS has concerns with the cost benefit of the project as it applies to the Food Stamp program. At issue is federal participation in SFIS in the amount of roughly $11.8 million. To satisfy the concerns of the FNS before proceeding would result in delays to the procurement and implementation schedules. A delayed procurement and implementation schedule would result in a delay in realizing the projected savings of $107 million for the first 3 years after full implementation. The loss of savings equate to roughly $73,377 per calendar day. To continue to wait for FNS’ approval would require the current request for proposal to be withdrawn and a new proposal to be released with changes to the implementation schedule. Such changes would be confusing to prospective bidders and the likelihood of viable bids would be reduced. The SFIS procurement is well underway. The request for proposal has been released and the initial draft bids have been received.” Thus, our report is accurate. 0 Social Services does not understand our point. We are not taking issue with the evaluation of the AFIRM system as it relates to Los Angeles County. However, there are shortcomings in that study that preclude the extrapolation of its results to the entire state. As we state on page 24, the systematic sampling design does not provide for a margin of error that would take into account any limitations in the sampling design or available data. Moreover, because a sampling design in which the variance of the estimate could not be estimated from the data used, none of the savings estimates presented in the independent evaluation reflect an upper or lower margin of error. Finally, in extrapolating Los Angeles County’s conditions when developing its statewide estimate, Social Services assumed that these conditions hold true in other counties. q Social Services is misleading the reader by citing savings of between $52.5 million and $64.6 million from the AFIRM system. Although Social Services used Los Angeles County’s recidivism data as we state on page 31, the remaining factors used to compute these amounts are not related to the methodology Social Services used to estimate savings of roughly $150 million for SFIS. Moreover, Social Services fails to mention that the independent evaluation report does not identify any of the savings it cites as attributable to the AFIRM system’s deterrence of duplicate-aid fraud. 6688 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6699 w Social Services’ statement that the adjustment of 33 percent was a reasonable assumption is unfounded. As we state on page 31, Social Services was unable to provide any data to support its theory or the use of 33 percent as opposed to some other percent. This use of a rate that is not based on sound statistical reasoning is one of the factors that causes us to question the validity of Social Services’ savings estimate. e Contrary to its belief, Social Services’ assumption regarding the time it takes county staff to print and photograph an applicant and to deny or discontinue cases when applicants fail or refuse to provide fingerprint images does not appear reasonable. Specifically, as we point out on page 29, using these assumptions Social Services estimated the total administrative costs that all counties except Los Angeles incurred for fiscal year 2000–01 would be roughly $1.8 million, yet Riverside County alone estimated its costs as amounting to 78 percent of Social Services’ estimated costs for the 57 counties. Therefore, it does not appear to us that Social Services’ assumptions are reasonable. Moreover, Social Services’ rationale for excluding investigative costs is illogical. Specifically, as we state on page 34, county staff are responsible for researching items in the SFIS queue and determining whether they are due to an administrative error or a possible fraud situation. We fail to understand why Social Services would not include an estimate for county investigative staff to resolve possible fraud items since this is such an integral part of maintaining SFIS. Further, data from Shasta County suggests that investigative costs may not be minimal. r Social Services’ estimate of the number of cases it would need to detect or deter to offset the estimated annual costs of $12 million is incorrect. First, its calculation assumes that each case would save the State $20,000 per year ($12 million/(50x12) or 600 cases per year). However, using Social Services’ monthly average savings figures on page 31, Social Services can expect to save only $580 per month per person, or $6,960 per year, which means SFIS would have to deter more than 140 duplicate-aid cases per month to achieve savings of $12 million per year. Further, as stated on page 34, between March 1, 2000, and September 30, 2002, investigators have found fraud in only 45 cases. Finally, because of its failure to capture critical data during SFIS’s implementation, as we state on page 35 Social Services cannot prove the number of cases that SFIS deters. 6688 California State Auditor Report 2001-015 California State Auditor Report 2001-015 6699 t To address Social Services’ concern, we deleted the word “annual.” y Social Services’ failure to recognize the importance of requiring counties to track their administrative costs separately causes us concern. Until Social Services understands the total cost of operating SFIS, the State cannot properly evaluate the system in terms of costs and benefits. For example, on page 29 we point out that Riverside County alone estimated its administrative costs as amounting to 78 percent of Social Services’ estimated costs for 57 counties. If Riverside County’s data is correct, Social Services’ estimate is substantially understated. Social Services is able to require counties to track their SFIS administrative costs separately from the other eligibility determination costs. Specifically, Social Services instructs counties to conduct time studies quarterly as part of its county expense claim process and assigns separate codes for each function. Currently, Social Services requires counties to code the fingerprint imaging requirement to the eligibility determination process. However, our review of a time study found that Social Services uses a separate code for county staff to track data for the Income and Eligibility Verification System, which as we discuss on page 16, is also part of the eligibility determination process. Therefore, it is feasible for counties to track their SFIS administrative costs separately using a similar mechanism. u Social Services’ statements concerning any lost savings associated with discontinuing SFIS are unreliable. Specifically, on pages 29 through 32, we identify several flaws in Social Services’ methodology for estimating savings attributable to SFIS. 7700 California State Auditor Report 2001-015 California State Auditor Report 2001-015 7711 cc: Members of the Legislature Office of the Lieutenant Governor Milton Marks Commission on California State Government Organization and Economy Department of Finance Attorney General State Controller State Treasurer Legislative Analyst Senate Office of Research California Research Bureau Capitol Press 7700 California State Auditor Report 2001-015 California State Auditor Report 2001-015 7711