Summary

Department of Social Services: For the CalWORKs and Food Stamp Programs, It Lacks Assessments of Cost‑Effectiveness and Misses Opportunities to Improve Counties’ Antifraud Efforts November 2009 Report 2009‑101 C A L I F O R N I A S T A T E A U D I T O R 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 TTY 916.445.0033 OR This report is also available on the World Wide Web http://www.bsa.ca.gov The California State Auditor is pleased to announce the availability of an on‑line subscription service. For information on how to subscribe, please contact the Information Technology Unit at 916.445.0255, ext. 456, or visit our Web site at www.bsa.ca.gov. Alternate format reports available upon request. Permission is granted to reproduce reports. For questions regarding the contents of this report, please contact Margarita Fernández, Chief of Public Affairs, at 916.445.0255. CALIFORNIA STATE AUDITOR Elaine M. Howle State Auditor Doug Cordiner B u r e a u o f S t a t e A u d i t s Chief Deputy 555 Capitol Mall, Suite 300 Sacramento, CA 95814 916.445.0255 916.327.0019 fax www.bsa.ca.gov November 3, 2009 2009-101 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 requested by the Joint Legislative Audit Committee, the California State Auditor presents its audit report concerning the Department of Social Services’ (Social Services) oversight of counties’ antifraud efforts for the California Work Opportunities and Responsibility to Kids (CalWORKs) program and the federal Supplemental Nutrition Assistance Program, known as the food stamp program in California. This report concludes that neither Social Services nor the six counties we visited have performed any meaningful analysis to determine the cost-effectiveness of counties’ antifraud efforts for the CalWORKs or food stamp programs. Therefore, we developed our own analysis and found that the measurable savings resulting from early fraud detection activities exceed the costs of such efforts for CalWORKs and approach cost neutrality for the food stamp program. Specifically, statewide projections for savings as a result of early detection showed that for every $1 spent on these activities, the State saved $1.35 for CalWORKs and 93 cents for the food stamp program. In contrast, the savings resulting from statewide ongoing investigations are typically not as cost-effective. Our calculations do not include any savings from the fraud that these activities may deter because there is no way to measure deterrence with any certainty. The cost-effectiveness of these anti-fraud efforts varies among the counties, which may be caused, in part, by differing county practices. Furthermore, we found that counties report inaccurate data on their antifraud efforts, but that Social Services has not taken sufficient steps to address this problem and passes this erroneous data to others, including internal decision makers, the federal government, and the Legislature. Social Services is also missing opportunities to improve counties’ antifraud efforts because it has not reviewed 25 of the 58 counties, including Los Angeles, over the past three years. Our review found that counties inconsistently follow up on information that may affect welfare recipients’ eligibility and they have a large backlog of unresolved cases of potential duplicate-aid fraud that the Statewide Fingerprint Imaging System (SFIS) has identified. Social Services also asserts that SFIS deters welfare fraud, yet it has not analyzed the cost-effectiveness of this system. Finally, Social Services has been delayed in taking the necessary steps to allocate $42.1 million in food stamp program overpayments the counties have collected, including $12.5 million the State is entitled to receive. Moreover, because counties currently hold the overpayment collection, the State has not had access to the funds, resulting in an estimated $1.1 million in lost interest earnings to the State. Respectfully submitted, ELAINE M. HOWLE, CPA State Auditor California State Auditor Report 2009-101 vii November 2009 Contents Summary 1 Introduction 7 Chapter 1 The Cost‑Effectiveness of Counties’ Efforts to Combat Fraud Varies, Though the Data for Such Computations Are Questionable 19 Recommendations 36 Chapter 2 The Department of Social Services and the Counties Could Improve Their Ongoing Efforts to Combat Welfare Fraud 37 Recommendations 54 Appendix Methodology for Computing the Cost‑Effectiveness of Fraud Investigation Efforts 57 Response to the Audit Department of Social Services 61 California State Auditor’s Comments on the Response From the Department of Social Services 75 California State Auditor Report 2009-101 1 November 2009 Summary Results in Brief Audit Highlights . . . The Department of Social Services (Social Services) is Our review of the Department of responsible for managing the California Work Opportunities Social Services’ (Social Services) oversight and Responsibility to Kids (CalWORKs) program and the federal of counties’ antifraud efforts related Supplemental Nutrition Assistance Program, known as the food to the California Work Opportunities stamp program in the State of California (State). Although these and Responsibility to Kids (CalWORKs) programs serve many who legitimately qualify for assistance, state program and the federal Supplemental and federal agencies generally recognize that some fraud exists, and Nutrition Assistance Program, known as federal law requires that states develop ways to detect fraud the food stamp program in California, within these programs. Although Social Services manages these found the following: programs in California, the counties are ultimately responsible for determining the eligibility of those receiving assistance, as » Although they have taken some steps, well as for detecting and investigating any fraudulent activities neither the counties nor Social Services related to these programs. Counties divide their investigative has performed any meaningful analyses efforts into early fraud detection activities (early fraud activities), to determine the cost-effectiveness of which detect potential fraud and prevent it from occurring, their efforts to detect and deter fraud in and ongoing investigations, involving cases in which counties the CalWORKs or food stamp programs. suspect ongoing fraud by persons currently receiving aid. » Our analysis of counties’ investigative Although they have taken some steps, neither the counties efforts found that, using a three-month nor Social Services has performed any meaningful analyses to projection, the measurable savings determine the cost‑effectiveness of their efforts to detect and deter resulting from early fraud activities fraud in the CalWORKs or food stamp programs. Therefore, we exceed the costs for CalWORKs and developed our own analysis. Our review of the cost‑effectiveness of approach cost neutrality for the food the counties’ investigative efforts found that, using a three‑month stamp program, assuming a three-month projection, the measurable savings resulting from early fraud projection of savings. activities exceed the costs of such efforts for CalWORKs and approach cost neutrality for the food stamp program assuming » Counties’ early fraud efforts are more a three‑month savings projection. In contrast, again using cost-effective than ongoing investigations. a three‑month savings projection, we found that ongoing investigations typically are not as cost‑effective. We measured » Neither Social Services nor the six counties cost‑effectiveness by comparing the savings resulting from efforts we visited took sufficient steps to ensure to combat fraud in the CalWORKs and food stamp programs the accuracy of the data counties report (welfare fraud)—including savings resulting from benefit denials, on their investigation activities. discontinuances, and reductions, as well as from overpayments identified in the course of investigations—to the counties’ costs » Social Services does not ensure that to perform these investigation activities. Based on an assumption counties consistently follow up on that they would have extended over three months, the statewide information it provides them that might projected savings derived from denials, discontinuances, and affect welfare recipients’ eligibility. reductions in aid payments as a result of early detection showed that for every $1 spent on early fraud activities during 2008, the » Although Social Services asserts that the State saved $1.35 for CalWORKs. For the food stamp program, Statewide Fingerprint Imaging System the 2008 return was 93 cents for every $1 spent on early fraud (SFIS) deters welfare fraud, it has not activities. On the other hand, the savings resulting from statewide assessed the cost-effectiveness of SFIS. ongoing investigations based on a three‑month projection showed that these efforts were not cost‑effective. Although these efforts 2 California State Auditor Report 2009-101 November 2009 Methods for Increasing Management likely help deter fraud, there is no way to measure this deterrence Personnel Salaries with any certainty, and therefore our calculations do not include the amount of any savings from such deterrence. • Merit salary increase program: Performance‑based salary increases funded from a merit compensation pool In large part, this difference in the cost‑effectiveness of antifraud established annually by the chancellor’s office. efforts is due to early fraud activities resulting in a much greater • Equity (market) increase program: Adjustments number of denials, discontinuances, and reductions of aid than designed to address discrepancies in pay, both within ongoing investigations produce and to early fraud activities costing and outside the university system, for comparable jobs. less. According to data that Social Services collects, the counties • Reclassification: Salary increases resulting rejected applications for aid, or they reduced or discontinued from changes in administrative classification that benefits, for about 34,700 CalWORKs cases, and the counties reflect changed assignments. identified overpayments of $19.6 million out of the approximately $3.1 billion in aid payments made during 2008. Further, the cost of investigating ongoing fraud is higher than the cost of performing early fraud activities due to the level of effort and evidence required for ongoing investigations. The statewide cost of ongoing investigations for CalWORKs during 2008 was $34 million, an amount that was $6 million (21 percent) more than the $28 million cost of early fraud activities. Our review of counties’ antifraud efforts found that the cost‑effectiveness of these efforts varied widely among the counties. For example, in 2008 Los Angeles County saved 35 cents for every dollar it spent on early fraud activities related to the food stamp program, while Orange County saved $1.82 for every dollar it spent on these activities. County practices related to early fraud referrals might partially account for the variations in cost‑effectiveness to the extent that these factors affect the number of resulting denials, discontinuances, and reductions of benefits. For example, Orange County cases represent only 3 percent of the statewide CalWORKs caseload, while Los Angeles County makes up 30 percent of the State’s CalWORKs caseload; however, Orange County referred nearly as many cases for early fraud activities as did Los Angeles County. During 2008 Orange County referred all applications meeting certain criteria for fraud review, which the county asserted often resulted in detected fraud. We also found that neither Social Services nor the six counties we visited took sufficient steps to ensure the accuracy of investigation activity reports. For example, in response to our review of their investigation activity reports submitted to Social Services, Los Angeles and Alameda counties stated that they have been unknowingly underreporting the outcomes of their investigations. Alameda County identified this problem before our review, while Los Angeles County realized the problem as a result of our inquiry. Because of a previous audit and because of its interactions with the counties, Social Services has known for several years that counties are reporting inaccurate data regarding their activities to combat welfare fraud, yet it has not taken sufficient steps to address this California State Auditor Report 2009-101 3 November 2009 problem. In addition, it uses these erroneous investigation activity reports to report to the federal government and to prepare reports submitted to internal decision makers and the Legislature. Social Services does not ensure that counties consistently follow up on information it provides that can potentially match welfare recipients to data received from various sources that might affect welfare recipients’ eligibility (match lists). One of the primary tools that Social Services distributes to counties are lists of individuals’ names that may match certain criteria that could make the individuals eligible for reduced aid amounts or ineligible for aid. Periodically, Social Services distributes to counties 10 match lists, which are generated by state and federal agencies. For six of the 10 match lists, federal regulations mandate that each aid recipient receive notification that an action will be taken on the information within 45 days. The remaining four match lists are not subject to a mandated deadline for this notification to take place. However, our review found that none of the counties we visited consistently followed up on all the match lists that they needed to complete within the 45‑day timeline, and four counties did not follow up consistently on the lists that had no set notification time frame. Such inconsistent efforts undermine the intent of the match lists, which is to provide the counties with actionable information that can prevent fraud or the continuation of fraudulent activity. We also determined that Social Services is missing opportunities to improve the counties’ follow‑up efforts on the match lists because it has not reviewed antifraud activities at 25 of the 58 counties during the three‑year period from August 2006 to August 2009. Among the counties not reviewed is Los Angeles, which helps to administer approximately 30 percent of the State’s CalWORKs cases and which Social Services last reviewed in 2005, and five small counties that Social Services’ records show have not been visited since 1995. Although Social Services indicates that it has had ongoing communications with Los Angeles County, the communications were limited to follow‑up on problems that Social Services observed in 2005 related to backlogs associated with the county’s overpayment collection efforts. These Income and Eligibility Verification System (IEVS) reviews are intended to be part of Social Services’ oversight of counties’ efforts to investigate welfare fraud. According to federal regulations, Social Services is ultimately responsible for processing matches consistently and in a timely manner. Because Social Services has not maintained adequate oversight of the counties, which conduct these efforts on its behalf, Social Services is failing to ensure that it complies with the regulations. The need for the IEVS reviews is evident, particularly because noncompliance was extensive among the counties we visited. 4 California State Auditor Report 2009-101 November 2009 Although Social Services asserts that the Statewide Fingerprint Imaging System (SFIS) is an important tool to deter welfare fraud, it has not adequately determined the cost‑effectiveness of SFIS because it believes there is no way of measuring the deterrence effect of the system. Since its implementation in 2000, counties have used SFIS to identify a total of 845 instances of fraud, of which 54 cases were identified in 2008. However, counties have a large, ongoing historical backlogs of SFIS results awaiting resolution. Indeed, as of July 31, 2009, the statewide backlog consisted of more than 13,700 unresolved cases that counties had not reviewed for more than 60 days. The backlog ranged from no cases for several counties to more than 3,600 unresolved cases for San Bernardino County. Social Services indicated that it does not follow up on counties’ reviews of SFIS cases because state laws or regulations do not mandate deadlines for such reviews. We contacted the counties we visited, as well as the counties with the highest backlogs, and several stated that they were unaware of the size of their respective backlogs. Most of the counties we contacted indicated that they did not identify fraud by using SFIS, but they indicated that they believe that SFIS is—in concept—a useful fraud deterrent. Regardless, if counties do not review their backlogs of cases, they cannot ascertain whether potential fraud is present within the backlogs. In addition, Social Services has been delayed in taking the necessary steps to allocate $17.2 million to the United States Department of Agriculture (USDA) and to claim its $12.5 million share of the $42.1 million in food stamp program overpayments that the counties have collected. Because counties currently hold these overpayments, neither the federal government nor Social Services have had access to the funds, resulting in an estimated $1.1 million in lost interest earnings to the State on its share of these funds. Moreover, the USDA has expressed long‑standing concerns about the accuracy of the information on overpayment collections reported by the counties, which Social Services does not review for this purpose. Lastly, county size, demographics, and county department staffing necessitate different approaches to investigating and prosecuting welfare fraud. In response to workload and staffing issues, counties have developed thresholds below which their district attorneys’ offices will generally not accept cases referred for prosecution. Of the more than 13,200 cases referred for prosecution that were available for counties to pursue during 2008, the counties acted on a total of 5,074, prosecuting 3,164 cases and deciding not to prosecute 1,910. Due to the low number of prosecutions, the counties’ backlog of nearly 6,400 prosecution referrals statewide at the beginning of 2008 had decreased by a mere 12 percent by the end of the year. California State Auditor Report 2009-101 5 November 2009 Recommendations To ensure that all counties consistently gauge the cost‑effectiveness of their early fraud and ongoing investigation efforts for the CalWORKs and food stamp programs, Social Services should work with the counties to develop a formula to perform cost‑effectiveness analyses using information that the counties currently submit. To make certain that counties receive the greatest benefit from the resources they spend on antifraud efforts related to CalWORKs and food stamp cases, Social Services should do the following: • Using the results from the recommended cost‑effectiveness analysis, determine why some counties’ efforts to combat welfare fraud are more cost‑effective than others. • Seek to replicate the most cost‑effective practices among all counties. To ensure the accuracy and consistency of the information on welfare fraud activities that counties report and that it subsequently submits to the federal government, the Legislature, and internal users, Social Services should take the following steps: • Remind counties that they are responsible for reviewing the accuracy and consistency of the investigation activity reports before submission. • Perform more diligent reviews of the counties’ investigation activity reports to verify the accuracy of the information submitted. • Provide counties with feedback on how to correct and prevent errors that it detects during these reviews. To ensure that counties are following up consistently on all match lists, Social Services should do the following: • Remind counties of their responsibility under state regulations to follow up diligently on all match lists. Further, it should work with counties to determine why poor follow‑up exists and then address those factors. • Perform IEVS reviews of all counties regularly. Recognizing that the deterrence effect of SFIS is difficult to measure, Social Services should develop a method that allows it to measure the benefits of this system and compare them to the cost of maintaining the system. Social Services should include in 6 California State Auditor Report 2009-101 November 2009 its cost calculations the administrative costs that counties incur for using SFIS. Based on its results, Social Services should determine whether the continued use of SFIS is justified. Social Services should continue to work with the USDA and make taking the steps necessary to distribute to the appropriate entities the $42.1 million in food stamp overpayment collections a priority. Social Services should track how counties determine prosecution thresholds and determine the effects of these thresholds on counties’ decisions to investigate potential fraud, with a focus on determining best practices and cost‑effective methods. It should then work with counties to implement the consistent use of these cost‑effective methods. Agency Comments In its response, Social Services generally agreed with the recommendations and provided additional perspective and information related to our findings. However, Social Services did not always agree with our conclusions. California State Auditor Report 2009-101 7 November 2009 Introduction Background The Department of Social Services (Social Services) is responsible for managing the California Work Opportunities and Responsibility to Kids (CalWORKs) program and the federal Supplemental Nutrition Assistance Program, known as the food stamp program in the State of California (State). These programs provide cash assistance for basic needs and food purchases to families or individuals who meet certain eligibility requirements. Due to the potential for fraud within these programs, federal law requires that states develop ways to detect such fraud. Although Social Services manages the programs in California, the counties, under Social Services’ oversight, are responsible for determining the eligibility of those receiving assistance, as well as for detecting and investigating any fraudulent activities. According to data that Social Services collects, the counties—as a result of their antifraud efforts—rejected applications for aid and reduced or discontinued benefits for about 34,700 CalWORKs cases and 52,800 food stamp cases during 2008. Also known as the welfare‑to‑work program, CalWORKs is the State’s version of the federal Temporary Assistance to Needy Families (TANF) program. CalWORKs provides cash assistance for living expenses to families with eligible children in the household. Eligibility is based on need according to age, citizenship, deprivation, income, resources, and residency. Unless they are declared exempt for such reasons as permanent or temporary disabilities, adult members of the household must meet work or vocational training requirements to maintain eligibility. In addition, individuals who have been convicted of drug‑related felonies are ineligible to receive aid under this program. The amount of cash assistance decreases as family income increases. Adults generally may not receive CalWORKs cash assistance for more than 60 months, while needy children remain eligible until they reach 18 years of age. In 2008 Social Services data shows a monthly average of 480,000 California households participated in CalWORKs, and they received approximately $3.1 billion, with an average monthly household grant of $538. Under the food stamp program, needy families and individuals receive funds that they can use only for food purchases. Families receiving cash assistance under CalWORKs are eligible for the food stamp program. In addition, families and individuals who do not qualify for CalWORKs can receive food stamp benefits based on income, asset, and resource thresholds. In 2008 a monthly 8 California State Auditor Report 2009-101 November 2009 average of 961,000 California households received food stamp assistance of approximately $3.3 billion, with an average monthly grant per household of $282. Social Services’ Role and Responsibilities Social Services is the state agency responsible for supervising the administration of the CalWORKs and food stamp programs. Through policy memos and regulations, Social Services provides guidance and oversight to counties, and it also consults with welfare advocates and the County Welfare Directors Association of California, which consists of the directors of welfare departments from the State’s 58 counties. In addition, Social Services reviews annual independent audits submitted to the State by the counties and monitors the counties’ corrective action plans. Social Services also requires counties to submit data related to their antifraud activities each month. In addition to program oversight, Social Services coordinates the counties’ efforts to combat welfare fraud by providing guidance, technical assistance, and information on fraud prevention and detection. Tools Social Services distributes to counties include lists of individuals’ names that potentially could match certain criteria that would cause the individuals’ aid amounts to be reduced or make them ineligible for aid (match lists). Federal law requires the states to help ensure that overpayments do not occur by maintaining a system to screen TANF program applicants and recipients against these lists for initial and ongoing eligibility. This system is known as the Income and Eligibility Verification System (IEVS), and federal law states that all CalWORKs applicants must provide their Social Security number to allow this screening. Although federal law does not require the State to use IEVS for food stamp applicants, state regulations require that all food stamp applicants receive IEVS screening.1 As Table 1 shows, IEVS‑related match lists can detect potential changes in recipients’ eligibility by matching welfare case information against databases from the State’s Employment Development Department and Franchise Tax Board and from the federal Internal Revenue Service (IRS) and Social Security Administration to ensure that aid recipients report all income, assets, and resources that may affect eligibility. A match occurs when there is a discrepancy between information reported by the recipient and information in these databases. Within 45 days of receiving the matches, the counties must 1 The eligibility requirements for many food stamp cases are the same as for CalWORKs cases. As a result, many IEVS‑related matches for CalWORKs cases also apply to food stamp cases. For example, during 2008 the referrals due to IEVS‑related matches for CalWORKs totaled 6,504, while food stamp referrals totaled 6,389. California State Auditor Report 2009-101 9 November 2009 follow up on matches related to recipients in their jurisdictions to determine whether recipients’ eligibility has changed. Social Services also provides the counties with four other types of match lists that identify individuals who might be deceased, incarcerated, or fleeing felons. Although the counties must follow up on these matches, there is no deadline for such follow‑up efforts for these reviews. Social Services periodically visits the counties to assess their processing of IEVS and the other match lists. Table 1 Match Lists Used by Counties to Detect Welfare Fraud MATCH LIST TYPE DESCRIPTION Match Lists That Counties Must Follow Up on Within 45 Days of Receipt Payment Verification System Received monthly based on data from the federal Social Security Administration (Social Security) and the State’s Employment Development Department. This list identifies cases in which recipients fail to report federal and state entitlement payments and individuals receiving aid in more than one state. New Hire Registry Received monthly based on data from the Employment Development Department. This list identifies recipients who were recently hired, rehired, or returned to work in California. Integrated Earnings Clearance Received quarterly based on data from the Employment Development Department. This list identifies Fraud Detection cases in which recipients fail to report or underreport employment income and those potentially receiving duplicate aid from different counties in California and the states of Arizona, Nevada, and Oregon. Beneficiary Earnings Exchange Records Received annually for all recipients and monthly for new recipients based on data from the Internal Revenue Service (IRS) and Social Security. This list identifies recipients’ out‑of‑state employment income, and income from federal, military, and self‑employment sources. Franchise Tax Board (FTB) Asset Match Received annually based on data from the FTB. This list identifies recipients’ unreported interest, dividend, and other sources of unearned income received by California entities. IRS Asset Match Received annually based on data from the IRS. This list identifies recipients’ unreported interest, dividends, lottery winnings, and other sources of unearned income reported to the IRS and not included in the FTB match. Match Lists That Counties Must Follow Up on With No Time Restrictions California Youth Authority (CYA) Match* Received monthly based on current data from the California Department of Corrections and Rehabilitation. This list identifies families that are potentially receiving aid for a minor incarcerated in a juvenile detention facility. Fleeing Felon Match Received monthly based on current data from the Department of Justice. This list identifies recipients with outstanding felony arrest warrants. Nationwide Prisoner Match Received monthly based on data from Social Security. This list identifies cases in which an adult incarcerated in a detention facility is receiving aid. Deceased Persons Match Received semiannually based on data from Social Security. This list identifies cases in which deceased individuals are being issued benefits. Source: Department of Social Services’ Income Eligibility and Verification System documentation. * The CYA is now the Division of Juvenile Facilities within the Department of Corrections and Rehabilitation. However, Social Services continues to refer to this match as the California Youth Authority Match. In addition to IEVS, in 2000 Social Services implemented the Statewide Fingerprint Imaging System (SFIS) to help prevent fraud involving duplicate aid. Because SFIS requires a fingerprint image and a photograph for each adult family member in a CalWORKs or food stamp case, Social Services asserts that the system enables 10 California State Auditor Report 2009-101 November 2009 it to identify individuals applying for and receiving aid in multiple jurisdictions and individuals using falsified or fraudulently obtained documents to assume multiple identities for the purpose of receiving aid. Social Services spent an estimated $31 million to develop SFIS, and it spends $5 million annually to maintain the system. Social Services recently entered a new, eight‑year contract at a total cost of $40 million, for ongoing system maintenance as well as to replace the equipment that counties are using for SFIS. Counties’ Responsibilities County welfare departments determine eligibility and issue CalWORKs and food stamp benefits to residents of their counties. Each county maintains staff dedicated to determining and monitoring eligibility on an ongoing basis. Because state regulations require counties to follow certain guidelines when issuing benefits and monitoring recipient eligibility, the application and eligibility determination process is similar across all counties. However, the counties have a certain amount of flexibility in how they organize their efforts to prevent and detect fraud. Therefore, some variation exists among the counties with respect to when and why cases are referred for investigation. During the application process, counties inform applicants for CalWORKs and food stamps of their rights and responsibilities as recipients of each program. For example, to ensure that their income does not exceed the level established for participation in the programs, recipients must self‑report their earnings to the county welfare department on a quarterly basis in order to continue receiving aid. In addition, recipients must Examples of Welfare Fraud inform county welfare departments of any changes in their household composition that may affect Unreported income: Individuals may fail to disclose eligibility, such as the return of an absent parent or income, earned or unearned, that may affect eligibility the departure of a child from the home. Applicants for aid. are also told that they are required to report Ineligible children: Individuals may attempt to receive truthfully or face charges of perjury as well as being benefits for children who are not eligible to receive benefits or required to pay restitution for funds they received who are already receiving benefits on another welfare case. for which they were ineligible, and that they can Absent parent in the home: Individuals may claim that a also be disqualified from receiving aid. parent who is living in the home is not living there for the purposes of receiving aid. State regulations require counties to maintain a special investigation unit to investigate potential Children not living in the home: Individuals may welfare fraud and to refer substantiated fraud either claim guardianship for children living with other adults for prosecution or for administrative settlement. or guardians. As the text box shows, welfare fraud can include a Source: The Department of Social Services’ and counties’ variety of allegations. Figure 1 shows the number Web sites. and location of the special investigation units in various counties. California State Auditor Report 2009-101 11 November 2009 Figure 1 Location and Numbers of Special Investigation Units by County DEL Counties with a special investigation unit located within: NORTE Sheriff’s office (2) SISKIYOU MODOC A county welfare department (25) The district attorney’s office (25) A county welfare department and the district attorney’s office (6) SHASTA LASSEN HUMBOLDT TRINITY TEHAMA PLUMAS GLENN BUTTE SIERRA MENDOCINO LAKE COLUSA S U TTE YUBA NEVADA PLACER R SONO M M A A RIN NAPA S Y O O L L A O NOSACRA MENTO AMA E D C L A O L D R AV O ER R A A S D T O UO A L L U P M IN N E E CONTRA JO S A A Q N UIN MONO COSTA SAN FRANCISCO SAN MATEO ALAMEDA STANISLAUS MARIPOSA SANTA CLARA MERCED MADERA SANTA CRUZ SAN FRESNO BENITO INYO TULARE MONTEREY KINGS KERN SAN LUIS OBISPO SAN BERNARDINO SANTA BARBARA VENTURA LOS ANGELES RIVERSIDE ORANGE IMPERIAL SAN DIEGO Source: California Department of Social Services. 12 California State Auditor Report 2009-101 November 2009 The actions counties undertake to prevent, detect, investigate, and prosecute welfare fraud are separated into two general categories—early fraud detection and ongoing fraud investigations. Early fraud detection activities (early fraud activities) are those initiated to substantiate the accuracy of information reported by individuals during the process of applying or reapplying for welfare benefits, adding new individuals to an existing welfare case, and determining special needs for welfare recipients. These activities are intended to prevent welfare fraud from occurring. Actions that counties can take on cases include denial, discontinuance, or reduction of aid payments. Counties use various approaches for these efforts, such as requiring that each new welfare applicant receive a visit from an early fraud detection investigator (early fraud investigator), identifying certain case characteristics that generate an automatic referral to early fraud investigators, and relying on the intake staff at the county welfare department to make a referral based on professional judgment. Investigative staff conducting these early fraud activities may or may not be sworn peace officers. State law requires that Social Services pay for all of a county’s early fraud activities with federal and state funds if Social Services approves the county’s early fraud detection program. Ongoing fraud investigations (ongoing investigations), on the other hand, involve cases in which counties suspect fraud by persons who are currently receiving aid. Typical allegations in an ongoing investigation include failure to report the presence of an absent parent in the home, a change in a child’s residence, and failure to report earned or unearned income. Counties are required to refer for welfare fraud investigation any case for which they have reasonable cause to believe that a welfare recipient has intentionally failed to disclose information that affects eligibility and subsequent receipt of benefits. These cases usually entail more complex investigations, and they could result in prosecution. Actions taken on such cases can include reduction of aid payments, denial of aid payments, or identification of overpayments. In all counties, sworn peace officers conduct ongoing investigations. Social Services recommends that counties have one sworn peace officer investigator for every 1,000 active CalWORKs cases. As Table 2 shows, three of the six counties we visited budgeted fewer sworn peace officers than the ratio recommended by Social Services. Counties must report their welfare fraud investigation and prosecution activities to Social Services each month. These reports include, among other activities, the number of early fraud and ongoing investigation referrals counties receive; the number of referrals they accept for further investigation; the number of investigations completed that resulted in denials, discontinuances, or reductions of aid; the number of cases referred for prosecution; and the results of prosecutions completed during the month. In its California State Auditor Report 2009-101 13 November 2009 instructions for the report, Social Services states that it uses the reports to evaluate the effectiveness of fraud prevention and detection programs, evaluate local agencies’ effectiveness in applying fraud prevention and detection policies, and help local agencies plan any needed changes to these efforts. Additionally, Social Services indicated that the reports provide county, state, and federal entities with information needed for budgeting, staffing, program planning, and other purposes. Table 2 Average Monthly CalWORKs and Food Stamp Caseload Compared to the Number of Sworn Peace Officer Investigators for the State and Selected Counties During 2008 AVERAGE MONTHLY NUMBER OF AVERAGE MONTHLY BUDGETED NUMBER OF INVESTIGATORS HOUSEHOLDS RECEIVING PERCENTAGE NUMBER OF HOUSEHOLDS PERCENTAGE SWORN PEACE OFFICER PER 1,000 ACTIVE ENTITY CALWORKS ASSISTANCE OF STATE CASES RECEIVING FOOD STAMPS OF STATE CASES INVESTIGATORS* CALWORKS CASES Alameda 18,684 4 35,828 4 12 0.6 Los Angeles 142,794 30 296,162 31 247 1.7 Orange 16,719 3 36,446 4 51 3.1 Riverside 24,572 5 41,762 4 22 0.9 Sacramento 31,028 6 54,310 6 29 0.9 San Diego 25,762 5 41,409 4 46 1.8 Statewide 480,207 100% 961,495 100% Sources: The Department of Social Services and county welfare departments. * These numbers are based on the budgeted positions for peace officer investigators who worked on various programs, including CalWORKs and food stamp, during fiscal year 2008–09. Scope and Methodology The Joint Legislative Audit Committee (audit committee) asked the Bureau of State Audits (bureau) to review the cost of combating fraud within the county welfare system programs. Specifically, the audit committee asked that the bureau determine the fraud prevention, detection, investigation, and prosecution structure for the CalWORKs and food stamp programs at the state and local levels and the types of early fraud detection or antifraud programs used. Further, the audit committee requested that we identify the number of special investigative units in each county and, for the counties we visited, the number of sworn peace officers employed at the units and where the unit resides in the county. We were also asked to determine how much fraud is referred or prosecuted for the two programs and the criteria used to determine when requests for investigations are referred or prosecuted. 14 California State Auditor Report 2009-101 November 2009 Additionally, the audit committee requested that the bureau determine, to the extent possible, the cost‑effectiveness of the fraud prevention efforts at the state and county levels, and to review how recovered overpayments are used. Further, we were asked to estimate, to the extent possible, the savings resulting from fraud deterred by counties’ antifraud activities and whether early fraud detection programs are more cost‑effective than ongoing investigations and prosecutions. The committee requested that we review how other states structure their antifraud efforts and identify any successes or best practices. Lastly, we were asked to assess Social Services’ justification for continuing to use both the SFIS and IEVS. Our review included six counties: Alameda, Los Angeles, Orange, Riverside, Sacramento, and San Diego. However, we did not include all six counties in all aspects of our review. Our analysis of Orange County was limited to the cost‑benefit analysis and structure because after we began our fieldwork, we observed that Orange County was reporting a high level of fraud activity in proportion to its welfare caseload. We did not review the use of CalWORKs recovery incentive funds or determine whether Orange County followed up on match lists as required. To determine the fraud prevention, detection, investigation, and prosecution structure for the CalWORKs and food stamp programs and the criteria counties use to determine when requests for investigations are referred or prosecuted, we interviewed appropriate staff at Social Services as well as staff of welfare departments and district attorneys’ offices from the six counties we visited. We also reviewed applicable laws, regulations, and policies and procedures of Social Services and the counties. To determine the number of special investigative units, as well as their location and the number of sworn peace officers at each unit, we inquired with staff at Social Services and the counties we visited. In addition, we selected a sample of cases from eight of the 10 match lists at five counties to determine whether they were appropriately following up on the information and, when applicable, doing so within specified time frames.2 We could not review the appropriateness of counties’ follow‑up efforts for the remaining two lists—the beneficiary earnings exchange records and IRS asset lists—because they contain federal tax information, and federal law expressly limits disclosure of this information. Although federal law allows disclosure of this information to state and county agencies that are responsible for administering the TANF program, it prohibits disclosure to a state audit agency, such as the bureau, except when the audit agency is auditing a state tax agency. Finally, 2 Orange County was not included in this review. California State Auditor Report 2009-101 15 November 2009 to assess Social Services’ justification for continuing to use both SFIS and IEVS, we interviewed appropriate staff at Social Services and reviewed any analyses they prepared. To ascertain how much fraud is referred or prosecuted for the CalWORKs and food stamp programs, we obtained investigation activity reports that counties submit to Social Services. To determine the cost‑effectiveness of fraud prevention efforts at selected counties for 2008, we used investigation activity reports that identified the number of cases for which these counties denied, discontinued, or reduced aid due to early fraud activities and ongoing investigations. Counties also identified in these reports the amount of overpayments identified due to fraud. We also used CalWORKs’ Summary Report of Assistance Expenditure, CalWORKs’ Cash Grant Caseload Movement Report, and the food stamp program’s Participation and Benefit Issuance Report that counties submit to Social Services to determine the statewide average monthly aid issued for a case during 2008 for CalWORKs and the food stamp program. We used these average aid figures to determine the monthly amount of aid payments avoided for denied and discontinued cases. Counties are not asked to submit any data to Social Services that identify the amount by which aid to recipients was reduced as a result of their fraud investigation efforts. Because Los Angeles County represents approximately 30 percent of the State’s CalWORKs caseload, we attempted to use its Los Angeles Eligibility Automated Determination, Evaluation, and Reporting (LEADER) system database to determine the average amount by which aid was reduced on a case due to a fraud investigation. In addition, we intended to use the LEADER system to determine the average number of months that Los Angeles County’s CalWORKs recipients had received aid at the time of a benefit reduction or discontinuance that was the result of a fraud investigation. Because an adult recipient can generally receive CalWORKs benefits for a maximum of 60 months, knowing the average number of months these recipients had already received aid for CalWORKs would have allowed us to project more accurately the amount the counties saved through their investigative efforts. However, after we obtained the LEADER database, Los Angeles County staff asserted that due to the limitations of the database and certain policies in that county, it was not feasible to perform these analyses as intended. Los Angeles County staff later identified data sets in the LEADER database that may have allowed us to compute reductions in aid resulting from early fraud and ongoing antifraud efforts. However, because of the uncertainty we had about encountering limitations with the LEADER database’s capabilities and weaknesses we identified in the county’s practices for recording fraud actions 16 California State Auditor Report 2009-101 November 2009 taken, we instead used data from the LEADER database to identify the average monthly amount by which aid was reduced on a case regardless of the reason. We also asked the five other counties that we reviewed to provide the average monthly amount by which aid was reduced using the same methodology. Only three counties were able to do so; two counties were close to the amount we calculated, the other county was significantly higher. Because the two counties’ amounts were reasonably close to the amount we calculated using the LEADER system, we used the LEADER database to determine the average monthly amount that counties saved by reducing aid for a case as a result of their investigation efforts to perform our cost‑benefit analyses. We describe our methodology for this calculation in the Appendix. The U.S. Government Accountability Office, whose standards we follow, requires us to assess the reliability of computer‑processed data. To determine the reliability of the LEADER database, we selected a random sample of records and reviewed the supporting documents, such as the case file and accounting records. We compared the information in the database to supporting documents to determine the accuracy of the information in the database. However, we did not conduct completeness testing because the source documents required for this testing are stored at 31 district offices located throughout Los Angeles County. Because of the weaknesses in the county’s practices for recording fraud actions taken and our decision not to conduct completeness testing, we concluded that the database’s information is of undetermined reliability. To determine the completeness of the data counties report to Social Services, we reviewed any supporting documents available at the six counties we visited for two months in 2006, 2007, and 2008. We compared the information on the supporting documents to the data these counties reported to Social Services for those months to determine whether the figures the counties reported matched the support. To determine the accuracy of the data counties report to Social Services, we selected a sample of cases from those reported to Social Services by the six counties for the selected months and traced them to source documents such as welfare case files or accounting records. We reviewed the source documents to determine whether the counties accurately summarized, among other things, the amount of aid, the aid program, and the disposition resulting from investigative efforts related to the case. Our review found that the six counties could not always support the data they reported. Because of these errors, we concluded that the data counties submit to Social Services on the investigation activity reports are of undetermined reliability. However, because no other data exist regarding the activities of counties to combat welfare fraud, we used the counties’ investigation activity reports in our analysis. California State Auditor Report 2009-101 17 November 2009 To evaluate how counties process the recovered overpayments and how they use CalWORKs recovery incentive funds, we interviewed staff of Social Services and at five of the counties3 we visited and reviewed their policies and procedures. Additionally, to determine whether the counties’ use of the CalWORKs recovery incentive funds was appropriate, we selected a sample of expenditures the counties incurred and for which they received reimbursements from the CalWORKs recovery incentive funds, and we compared the purposes of those expenditures with allowable activities. Based on this testing, we determined that the five counties used the CalWORKs recovery incentive funds appropriately. Further, counties receive CalWORKs and food stamp overpayment collections. Each county is responsible for returning the CalWORKs overpayment recovery funds to the State monthly, but for the food stamp program Social Services is responsible for calculating and distributing the amount of food stamp overpayments to the USDA, the counties, and itself each quarter. Although we observed that CalWORKs overpayments are processed regularly, we performed additional testing because we found that a backlog of food stamp overpayments existed. Thus, we reviewed how Social Services processes these overpayments and the reasons for the backlog. To review how other states structure their antifraud efforts, and to identify best practices and lessons learned, we identified two other states with large welfare caseloads and administrative structures similar to California’s caseload and administrative structure. Of the 10 states with the highest TANF caseloads, only California, New York, and Ohio have counties administer welfare programs as well as investigate and prosecute welfare fraud. However, staff from New York and Ohio indicated that their states have not formally studied and identified best practices. For example, each of New York’s 58 counties develops its own processes for investigating and prosecuting welfare fraud. The New York State program integrity director told us that although the state collects information regarding investigations and prosecutions and uses it to calculate the amount of aid avoided by each county and statewide, it has not performed a cost‑effectiveness analysis to determine the most cost‑effective practices used by its counties. 3 Orange County was not included in this review. 18 California State Auditor Report 2009-101 November 2009 Blank page inserted for reproduction purposes only. California State Auditor Report 2009-101 19 November 2009 Chapter 1 THE COST‑EFFECTIVENESS OF COUNTIES’ EFFORTS TO COMBAT FRAUD VARIES, THOUGH THE DATA FOR SUCH COMPUTATIONS ARE QUESTIONABLE Chapter Summary Although they have taken some steps, neither the counties nor the Department of Social Services (Social Services) have conducted meaningful analyses to determine the cost‑effectiveness to detect and deter fraud in the California Work Opportunities and Responsibility to Kids (CalWORKs) program and the federal Supplemental Nutrition Assistance Program (food stamp program). As a result, we developed our own analysis, which indicates that the cost‑effectiveness of antifraud efforts varies among the counties. Our review of the cost‑effectiveness of investigative efforts for the CalWORKs program and the food stamp program, found that the measurable savings attributable to early fraud detection activities (early fraud activities) exceed the costs of such efforts for CalWORKs and approaches cost neutrality for the food stamp program. However, ongoing fraud investigations (ongoing investigations) are typically not as cost‑effective as early fraud activities. Using an assumption that the savings would persist for three months, our calculation showed that on a statewide basis, early fraud activities were cost‑effective for CalWORKs and nearly cost‑neutral for the food stamp program. On the other hand, statewide ongoing investigations were not cost‑effective under the three‑month projection. This difference is due in large part to the fact that early fraud activities result in a much greater number of denials, discontinuances, and reductions of aid than ongoing investigations produce, and early fraud activities cost less. However, using an assumption that savings would persist for 18 months, we found that ongoing investigations were generally cost‑effective. County practices may partially account for variations in cost‑effectiveness among the counties, to the extent that these factors affect the number of resulting denials, discontinuances, and reductions. Because there is no way to measure with any certainty the extent to which antifraud efforts act as a deterrent to fraud, our calculations do not include the amount of savings attributable to deterrence. Some counties claimed that they have failed to track and report accurately all of the benefit reductions and discontinuances that result from investigations on the investigation activity report submitted to Social Services. Further, Social Services and the counties have not performed any 20 California State Auditor Report 2009-101 November 2009 meaningful analyses to compare the cost of their antifraud efforts to the savings that counties realize from aid payments avoided and overpayments identified. Lastly, we found that neither Social Services nor the six counties we visited take sufficient steps to ensure the accuracy of the counties’ investigation activity reports. Further, our review found the counties’ data to be of undetermined reliability. Social Services has known for several years that the data counties report are not always accurate. Nevertheless, Social Services uses this report, along with other information, to substantiate to the United States Department of Agriculture (USDA) the costs it expects to incur during the next year. It also uses these erroneous investigation activity reports to prepare reports submitted to internal decision makers and the Legislature. However, because no other data exist regarding the activities of counties to combat welfare fraud, we used the investigation activity reports in our analysis. Early Fraud Programs May Not Be Cost‑Effective in All Counties, but They Are Generally More Cost‑Effective Than Ongoing Investigations Our review of the cost‑effectiveness of CalWORKs and food stamp fraud investigation activities during 2008 found that the savings produced by early fraud activities using a three‑month projection exceed the costs of such efforts for CalWORKs and are almost cost‑neutral for the food stamp program, but that ongoing We measured cost‑effectiveness investigations are typically not as cost‑effective. We measured by comparing the savings cost‑effectiveness by comparing the savings resulting from efforts resulting from counties’ efforts to combat welfare fraud—including savings resulting from benefit to combat welfare fraud to denials, discontinuances, and reductions, as well as overpayments their costs to perform these identified in the course of investigations—to the counties’ costs investigation activities. to perform these investigation activities. We considered early fraud and ongoing investigation programs that achieve more than $1 in grant savings and overpayments identified for every $1 spent to be cost‑effective.4 The costs in our calculations are based on the staff time directly attributable to these early fraud activities and ongoing investigations and the related administrative costs. Fraud exists when a person knowingly, and with intent to deceive or defraud, makes a false statement or representation to obtain, continue, gain an increase in, or avoid a legitimate reduction in benefits. Failure to disclose facts that could result in a denial, discontinuance, or reduction of benefits, or the acceptance of benefits for which a person knows he or she is not eligible, also constitutes welfare fraud. Counties 4 We discuss our methodology for calculating cost‑effectiveness in more detail in the Appendix. California State Auditor Report 2009-101 21 November 2009 devote resources both to the early detection and prevention of fraud— early fraud activities—and the investigation of ongoing fraud related to cases currently or previously receiving aid— ongoing investigations. As we discuss in the Introduction, early fraud activities focus on preventing fraud from occurring Early fraud activities focus on and can result in benefit denials, discontinuances, and reductions. preventing fraud from occurring Ongoing investigations focus on identifying and prosecuting while ongoing investigations individuals who at anytime were receiving aid under fraudulent focus on identifying and circumstances, and these investigations can result in benefit prosecuting individuals who at discontinuances, and reductions as well as the identification of anytime were receiving aid under overpayments. Thus, to the extent that recipients would have fraudulent circumstances. received aid payments or continued to receive aid payments, these efforts result in savings to counties because fraudulent aid payments are avoided. As we discuss in the Appendix, we projected the savings that counties realize from aid payments avoided as a result of early fraud activities and ongoing investigations over three months and 18 months. Forming the basis of the three‑month estimate—our most conservative estimate of the savings—is the fact that recipients of both the CalWORKs and food stamp programs are required to report quarterly any changes in their eligibility, such as increased income or a child leaving the home. We also used an 18‑month projection because Social Services asserted that its ongoing analysis of historical eligibility data for CalWORKs recipients indicates that they receive aid for an average of 18 months. Although Social Services also determined that food stamp recipients receive aid for an average of 31 months, we used the shorter period in our analysis to maintain consistency between our cost‑effectiveness results for the two aid programs. Early Fraud Activities Are Generally More Cost‑Effective Than Ongoing Investigations As depicted in Table 3 on the next page, our savings calculation based on a three‑month projection showed that on a statewide basis, early fraud activities performed in 2008 were cost‑effective for CalWORKs and nearly cost‑neutral for the food stamp program. On the other hand, the 2008 statewide ongoing investigations were not cost‑effective using the three‑month projection. However, when we projected the savings over 18 months, the 2008 statewide savings due to early fraud activities for both the CalWORKs and food stamp programs were nearly eight times and just over four times greater than their respective costs, and savings from ongoing investigations also exceeded costs for both programs. 22 California State Auditor Report 2009-101 November 2009 Table 3 2008 Savings Resulting From Early Fraud Activities and Ongoing Investigations for Every $1 That Counties Spend on These Efforts THREE-MONTH PROJECTION OF SAVINGS 18-MONTH PROJECTION OF SAVINGS EARLY FRAUD ONGOING INVESTIGATIONS EARLY FRAUD ONGOING INVESTIGATIONS ENTITY CALWORKS FOOD STAMP CALWORKS FOOD STAMP CALWORKS FOOD STAMP CALWORKS FOOD STAMP Alameda $1.76 $1.74 $0.55 $0.38 $10.53 $10.46 $0.65 $0.73 Los Angeles 0.61 0.35 0.77 0.25 3.66 2.08 0.79 0.29 Orange 2.37 1.82 1.62 0.98 14.19 10.92 9.13 5.53 Riverside 1.25 1.08 0.45 0.95 7.50 6.47 0.47 0.97 Sacramento 0.53 0.52 0.64 0.88 3.15 3.10 1.50 3.29 San Diego 2.60 1.16 1.10 1.87 15.58 6.97 1.30 2.42 Statewide 2008 1.35 0.93 0.88 0.72 8.12 5.58 2.39 2.14 Statewide 2007 1.52 0.85 1.12 0.65 9.09 5.08 2.71 1.89 Statewide 2006 1.25 0.72 0.79 0.60 7.50 4.29 2.35 1.88 Sources: Bureau of State Audits’ analysis based on data that counties submit to the Department of Social Services (Social Services). Note: As we discuss elsewhere in this report, the figures that counties report to Social Services regarding investigation activities are of undetermined reliability. Of the six counties we reviewed, four had early fraud programs that were cost‑effective, but only two counties’ ongoing investigations for CalWORKs resulted in greater savings than costs when we used the three‑month projection. Because there is no way to measure with any certainty the extent to which antifraud efforts act as a deterrent to fraud, our analysis does not include any savings from fraud deterrence. However, to the extent that the measurable savings and costs in our analysis reflect actual variances in the cost‑effectiveness of counties’ efforts to combat fraud, differing policies and practices may account for some of the variation. Counties generally realize greater savings per dollar spent on early fraud activities than on ongoing investigations. This difference is due largely to the fact that according to the data that counties report, early fraud activities generally result in a much greater number of denials, discontinuances, and reductions of aid than ongoing investigations produce, and also because early fraud activities cost less. According to data that Social Services collects, the counties rejected applications for aid, or discontinued or reduced benefits, for about 34,700 CalWORKs cases during 2008 and identified overpayments of about $19.6 million out of the approximately $3.1 billion in aid payments made. However, the statewide cost of ongoing investigations for CalWORKs during 2008 was $34 million, an amount that was $6 million (21 percent) more than the $28 million cost of early fraud activities. As Table 3 shows, when we projected three months of savings, on a statewide basis for 2008 early fraud activities for the CalWORKs program resulted in $1.35 in savings California State Auditor Report 2009-101 23 November 2009 for every dollar spent, while ongoing investigations resulted in only 88 cents per dollar spent. Under the 18‑month projection, this disparity increased, with 2008 early fraud activities returning $8.12 per dollar spent, which is more than three times the ongoing investigations’ return of $2.39. Statewide, the cost‑effectiveness of early fraud activities and ongoing investigations for CalWORKs were the highest in 2007, but our calculations indicate that for the food stamp program these activities have become more cost‑effective over the past three years. Ongoing investigations generally result in fewer discontinuances or Ongoing investigations generally reductions of aid because the main purpose of these investigations result in fewer reductions or is to prove suspected fraud that may have occurred in the past and discontinuances of aid because their not to determine changes in current eligibility. In contrast, early main purpose is to prove suspected fraud activities inherently focus on a recipient’s current eligibility. fraud that may have occurred in the In addition, the backlog of ongoing investigations, which we noted past and not to determine changes exists at all of the counties we reviewed, introduces a delay between in current eligibility. the fraud referral and the actual investigation. Riverside County also stated that the information matching welfare recipients to data about their eligibility (match lists) produces information that is sometimes more than three months old and is even older by the time it is routed and reviewed. These delays decrease the likelihood that an ongoing investigation will uncover facts that affect current eligibility. Further, counties indicated that for the majority of ongoing investigation cases, the recipients are no longer receiving aid when the investigation starts. Because reductions in and discontinuances of current aid do not result from ongoing investigations of closed cases, the only measurable savings for the county result from the identification of any overpayments. In contrast, early fraud activities primarily involve efforts that result in the denial of aid to a welfare applicant who is ineligible, or a reduction in or early discontinuance of aid payments to individuals who are currently receiving aid. As a result, early fraud activities have a higher number of cases for which aid is denied, discontinued, or reduced, and these actions result in lower payments of benefits in the future. In addition, the cost of early fraud activities is generally less than the cost of ongoing investigations due to the level of evidence necessary. Typically, an early fraud activity involves reviewing the available information and interviewing or visiting the recipient to determine whether the eligibility worker’s suspicions are valid. Some counties indicated that they try to complete these types of activities within a relatively short period, resulting in less cost. For example, Orange County noted that its policy is to complete all early fraud activities on a case within 15 business days of receiving the referral, and San Diego County’s goal is to complete early fraud activities within 10 business days. 24 California State Auditor Report 2009-101 November 2009 Ongoing investigations, on the other hand, generally require investigators to establish that the recipient intended to commit fraud. Because a possible outcome of these investigations is a felony conviction, there is a greater burden on the investigators to gather the necessary evidence to build a case for prosecution. In addition, some cases can be very complex and might involve coordination among multiple jurisdictions. Los Angeles County, for example, asserted that it investigates some cases that involve organized crime. According to a lieutenant from the Los Angeles County district attorney’s office, organized crime rings involve large groups of individuals who conspire to commit fraud that spans multiple agencies and public assistance programs. Members of an organized crime ring may, for example, use falsified records or other documentation to claim a nonexistent child, a lack of assets, or an absent parent in order to collect CalWORKs, food stamp, and other public assistance benefits. According to the lieutenant, coordinating with the various agencies and counties that administer these benefits to investigate crime rings requires a significant investment of time and other resources. Because of the higher costs of performing investigations and the lower measurable savings they produce, our calculations indicate that ongoing investigations are not as cost‑effective as early fraud activities. Cost‑Effectiveness Varies Significantly Across Counties The net savings resulting from early fraud activities and ongoing investigations vary widely across the six counties we included in our analysis. In the three‑month projection for the six counties, Los Angeles County’s early fraud activities for the food stamp program resulted in the lowest savings, yielding only 35 cents for every dollar it spent. In contrast, Orange County yielded $1.82 in savings for every dollar spent in early fraud activities for the food stamp program. Similarly, in the three‑month projection, the cost‑effectiveness of the counties’ ongoing investigations related to the food stamp program ranged from as little as 25 cents for every dollar that Los Angeles County spent to $1.87 for every dollar that San Diego County spent. The results show similar variances among counties for the CalWORKs program. Differences in county practices may Differences in county practices may partially account for partially account for variations in variations in the cost‑effectiveness of early fraud activities the cost‑effectiveness of early fraud across the counties, to the extent that these practices affect the activities across the counties, to the number of resulting denials, discontinuances, and reductions. extent that these practices affect Policies that generate a large number of referrals may contribute the number of resulting denials, to greater savings. For example, the net savings for CalWORKs discontinuances, and reductions. early fraud activities in Orange and San Diego counties significantly exceed the statewide average, as well as the savings of the other four counties we reviewed. Both of these counties California State Auditor Report 2009-101 25 November 2009 reported a disproportionately high number of early fraud referrals for 2008, and both counties had policies in place in 2008 that generated mandatory early fraud referrals. According to San Diego County’s policy, it is mandatory for investigators to perform an early fraud interview for every applicant who either has not received aid in the last 12 months in the county or who has received aid within the last 12 months in the county and has an unresolved fraud suspicion against him or her.5 Until 2009 Orange County mandated early fraud referrals based on certain characteristics, such as applicant households that only listed children as eligible recipients and applicants with welfare fraud sanctions. However, due to significant budget cuts, Orange County discontinued its mandatory referrals as of February 2009. It believes that this change has resulted in a drop of more than 50 percent in early fraud referrals. Both counties indicated that their policies may account for their disproportionately high numbers of early fraud referrals. For example, although Orange and San Diego counties accounted for approximately 3 percent and 5 percent, respectively, of the State’s CalWORKs program caseload in 2008, they accounted for 11 percent and 17 percent, respectively, of all early fraud referrals for CalWORKs and each county also accounted for 4 percent of the State’s food stamp caseload and each county accounted for 11 percent of early fraud referrals for the food stamp program. Table 4 Number of Cases Referred for Early Fraud and the Results of Those Activities in 2008 CALWORKS CASES FOOD STAMP CASES PERCENTAGE PERCENTAGE OF REFERRALS OF REFERRALS CASES DENIED, RESULTING IN CASES DENIED, RESULTING IN REFERRALS DISCONTINUED, OR AID DENIED, REFERRALS DISCONTINUED, OR AID DENIED, ACCEPTED FOR REDUCED AS A RESULT OF DISCONTINUED, ACCEPTED FOR REDUCED AS A RESULT OF DISCONTINUED, ENTITY INVESTIGATION INVESTIGATION OR REDUCED INVESTIGATION INVESTIGATION OR REDUCED Los Angeles 11,832 3,548 30% 19,453 5,778 30% Orange 10,932 4,011 37 16,862 6,250 37 Riverside 23,385 1,699 7 41,137 2,685 7 Sacramento 3,778 1,394 37 3,656 1,353 37 San Diego 16,803 4,284 25 15,970 3,846 24 Statewide 101,065 27,154 27 147,081 40,965 28 Sources: Investigation activity reports that counties submitted to the Department of Social Services (Social Services) during 2008. Note: As we discuss elsewhere in the report, the figures counties report to Social Services are of undetermined reliability. We did not include Alameda County in this table because we learned that it had been reporting certain early fraud actions taken, but not all of the related early fraud referrals for these same actions. Thus, its figures would significantly overstate the percentage of referrals with actions taken on them. 5 San Diego County created Project 100 Percent to confirm and verify statements provided by CalWORKs applicants. This project is intended to be an extension of the initial eligibility determination process. 26 California State Auditor Report 2009-101 November 2009 The counties that typically generated the highest measurable net savings in 2008—Orange and San Diego—not only accepted a high number of early fraud referrals but also had a high percentage of benefit denials, discontinuances, or reductions compared to their early fraud referrals. For example, as shown in Table 4, Orange County accepted more than 10,900 CalWORKs early fraud referrals in 2008, representing 11 percent of these referrals statewide. This number is disproportionately high, given that Orange County represented about 3 percent of the CalWORKs caseload in 2008. In addition, 37 percent of these referrals resulted in denial, discontinuance, or reduction of benefits, a rate that is significantly higher than the statewide average of 27 percent. Orange County’s high number of referrals and high rate of results likely account for the county’s $2.37 net savings per $1 spent for CalWORKs using a three‑month savings projection, the second highest among the counties we reviewed. In contrast, Riverside County produced a net savings of $1.25 per $1 spent for CalWORKs early fraud activities. Although this county accepted nearly 23,400 CalWORKs early fraud referrals in 2008, representing 23 percent of all referrals accepted in the State, only about 1,700 (7 percent) of these referrals resulted in a change in eligibility. This lower percentage largely accounts for its lower calculated net three‑month savings. Among the counties we visited, the two with the highest net savings from their early fraud activities for CalWORKs in 2008 were San Diego and Orange counties. Both counties attribute their success in part to a close working relationship between investigators and eligibility workers, and both house most of their early fraud investigators with the eligibility workers, asserting that doing so allows direct access to investigators so that issues can be resolved San Diego County stated that quickly. San Diego County also stated that the consistent training the consistent training provided provided to eligibility workers has made them more diligent and to eligibility workers has made aware of fraud indicators, thus enabling them to generate a quality, them more diligent and aware of valid referral. Although sworn peace officers conduct early fraud fraud indicators, thus enabling activities in both of these counties, our analysis did not clearly them to generate a quality, valid indicate whether or how this practice affects cost‑effectiveness. For investigation referral. example, Los Angeles County also uses sworn officers to conduct early fraud activities, yet it produced one of the lowest net savings per dollar spent for CalWORKs early fraud activities among the counties we reviewed. The net savings produced by ongoing investigations also varied widely across the counties, and the savings varied more widely for the 18‑month projection than for the three‑month projection due to the disparity in the number of benefit reductions and discontinuances that counties reported as a result of these investigations. For the three‑month projection, the net savings for California State Auditor Report 2009-101 27 November 2009 the food stamp program ranged from 25 cents to $1.87 for every dollar spent. For the 18‑month projection, the net savings ranged from 29 cents to $5.53 for every dollar spent. When we discussed each county’s results with its special investigation unit managers, Alameda and Los Angeles counties acknowledged that they have been underreporting to Social Services the results from their ongoing investigations. Alameda County stated that this underreporting has occurred because it classified almost all of its ongoing investigative activities as early fraud, and it believes that as much as 30 percent of the previously reported early fraud activities were actually related to ongoing investigations. Similarly, Los Angeles County believes that the actual number of ongoing investigations that produced results exceeds the number that it reported. Because of weaknesses in their reporting practices, neither county could support its assertion of underreporting. We reviewed the data of the counties shown in tables 3 and 4, and we used these data in our analysis even though we found errors in the data that the counties reported and concluded that the data were of undetermined reliability. Unfortunately, no other data exist regarding the activities of counties to combat welfare fraud. In contrast, and as we discussed earlier in this chapter, Riverside County believes that its lower savings from ongoing investigations reflect the fact that few discontinuances and reductions occur because of ongoing investigations. Social Services and the Counties Have Not Determined Whether Their Antifraud Efforts Are Cost‑Effective Social Services has not performed any meaningful cost‑effectiveness analysis of the counties’ investigation efforts. In its role of supervising the administration of the CalWORKs and food stamp programs by the counties, Social Services should ensure that the counties efficiently deliver services to the public and also that they have effective processes in place to combat welfare fraud. Thus, for Social Services to identify best practices, all counties must consistently determine the cost‑effectiveness of their ongoing investigation activities, using a consistent period for measuring the savings. Although Social Services has developed a formula to calculate the Although Social Services has savings that counties realize as a result of their antifraud efforts—in developed a formula to calculate terms of fraudulent aid not paid and administrative savings—it does the savings that counties realize as not use this formula to evaluate the cost‑effectiveness of counties’ a result of their antifraud efforts, antifraud efforts. The formula uses the data counties report on their it does not use this formula to investigation activity reports related to denials, discontinuances, and evaluate the cost‑effectiveness of reductions of aid; the average grant amounts; and the average time counties’ antifraud efforts. period over which Social Services assumes that recipients receive 28 California State Auditor Report 2009-101 November 2009 aid. Additionally, the formula assumes that there will be a reduction in the counties’ workload when recipients are denied aid. Social Services indicated that it created this formula in 2006 to respond to a legislative request, but it could not provide any other examples of the formula’s use. Also, in 2006 Social Services compiled statewide and select county statistics on referrals for early fraud activities and ongoing investigations and any changes in aid resulting from these efforts during fiscal years 1994–95 through 2003–04, and the costs related to early fraud activities and ongoing investigations during fiscal years 1997–98 through 2003–04. These statistics describe the cost savings for early fraud activities and ongoing investigations, but only for fiscal years 2002–03 and 2003–04, reflecting that early fraud activities were more cost‑effective than ongoing investigations. Social Services did not present conclusions on the effectiveness of individual counties. This was a promising first step, but to get the most benefit Social Services should continuously compile and analyze these statistics for use in assessing and improving counties’ antifraud efforts. None of the counties we visited have performed a comprehensive analysis to compare the cost of their antifraud efforts to the savings they realize from aid payments avoided and overpayments identified. Some have completed limited studies. San Diego County identified the savings from its Project 100 Percent between 2001 and 2008, but the county did not identify the costs nor did it review the savings resulting from its other antifraud efforts. In 2002 Alameda County identified savings related to its antifraud efforts, using Social Services’ formula. The primary purpose of this analysis was to highlight the savings that its investigations generated. However, the county did not compare these savings with its costs. Los Angeles County asserted that it is not aware of a method by which it can perform a cost‑effectiveness analysis, primarily because there is no viable method to determine the benefits or the value of activities that deter fraud. While we agree that measuring the deterrence effect of counties’ efforts is difficult, figures for other measurable savings and costs are readily available, as presented earlier in this chapter. Although the counties we visited believe their efforts to avoid making aid payments to Without knowing the ineligible recipients and stopping fraud are cost‑effective, they are cost‑effectiveness of their practices, making this assertion without having performed a cost‑benefit the counties and Social Services analysis. Further, it is unclear whether the reason for the disparity cannot identify best practices that in results among the counties included in our cost‑benefit analysis can be adopted by other counties to is because some counties are more cost‑effective than others, increase the return on every dollar because the incidence of fraud is greater in some counties, or that the State spends combating fraud the counties differ in their efforts to report accurate data. Without within the CalWORKs and food knowing the cost‑effectiveness of their practices, the counties stamp programs. and Social Services cannot identify best practices that can be California State Auditor Report 2009-101 29 November 2009 adopted by other counties to increase the return on every dollar the State spends combating fraud within the CalWORKs and food stamp programs. On a promising note, a program integrity steering committee (steering committee) was formed to follow up on the results of a 10‑year statistical study on fraud prevention and detection activities in the CalWORKs and food stamp programs. Social Services completed this study, based primarily on county‑reported information from the investigation activity reports, in 2006. The steering committee, with members from the county welfare directors’ and California district attorneys’ associations, sought to identify cost‑effective approaches for improving program integrity in the CalWORKs and food stamp programs. To accomplish this goal, the steering committee surveyed seven counties about their approaches to early and ongoing fraud prevention, detection, and prosecution. From this survey, the steering committee approved recommendations in 2008 for both the counties and Social Services regarding the most promising approaches it found. Among the eight recommendations for the counties were to emphasize using early fraud activities, enable open communication among welfare fraud staff, use experienced prosecutors, report data more consistently, and provide regular training to welfare fraud staff. The 10 recommendations directed to Social Services included establishing a standard method of computing county cost savings, providing counties regular reports to enable them to monitor the cost‑effectiveness of their efforts to combat welfare fraud, maintaining a central repository of fraud training ideas, and reviewing the cost‑effectiveness of each match list, as well as soliciting feedback from the counties on the lists’ usefulness. Social Services indicated that it is addressing four of the 10 recommendations and is considering how to address the remaining six. Social Services Does Not Ensure That Counties Report Accurate Data on Their Welfare Fraud Investigations Neither Social Services nor the six counties we visited have taken sufficient steps to ensure the accuracy of investigation activity reports. These reports, which counties submit monthly to Social Services, summarize the counties’ investigative efforts and the results of those efforts. Specifically, we found that the information Social Services is aware of the these counties included on the investigation activity report is not problems with the counties’ always accurate, supported, or reported consistently. Social Services data, but it has not taken steps is aware of these problems with the data, but it has not taken steps to improve the accuracy of to improve the accuracy of the counties’ reporting. In addition, it their reporting. 30 California State Auditor Report 2009-101 November 2009 uses these erroneous investigation activity reports to populate part of a report it submits to the federal government and to prepare reports submitted to internal decision makers and the Legislature. Problems Exist in the Counties’ Reporting of Investigation Activities Social Services requires counties to submit monthly investigation activity reports to summarize their welfare fraud investigation activities for the CalWORKs and food stamp programs. Some of the activities that they must report each month include the number of fraud cases referred for investigation, the disposition of investigations that were completed, any overpayments identified as a result of completed investigations, the number of investigations referred for prosecution, and the results of prosecutions completed. The USDA requires all states to report summary investigation data on an annual basis. Our review detected numerous errors in the annual investigation activity reports for 2006 through 2008 at the six counties we visited. For example, during our review of Alameda County’s investigation activity reports, we noticed very few cases for which benefits were discontinued or reduced as a result of ongoing investigations. Alameda County indicated that it had been inaccurately reporting these cases as early fraud cases, and it believes that approximately 30 percent of the early fraud cases previously reported were actually related to ongoing investigations. Further, Alameda County noted that its system could not separately identify CalWORKs and food stamp cases, and it reported the same numbers for both programs. Alameda County indicated that it has started to revise the method it uses for preparing future investigation activity reports, but despite having previously reported inaccurate information to Social Services, the county does not plan to revise past reports, asserting it does not have the resources to do so. We also found problems with Los Angeles County’s preparation of its investigation activity report. Los Angeles County consolidates data from several sources to prepare this report. These sources include early fraud reports for the 31 district offices located in the county and ongoing investigation reports from its welfare fraud investigation headquarters. The reports are based on tally sheets prepared by investigation supervisors. However, because the tally sheets do not list specific cases, Los Angeles County could not tell us the case numbers related to the activity totals it reported on the investigation activity report. Although Los Angeles County said that it could re‑create the monthly listings, it indicated that the totals would be different due to changes that occurred subsequent to a particular monthly report. Further, Los Angeles County Due to weaknesses in Los Angeles County’s process, we could not verify the accuracy of its investigation activity reports. California State Auditor Report 2009-101 31 November 2009 noted that its staff do not consistently document the results of early fraud investigations. When an investigator concludes an early fraud investigation, he or she completes the top portion of a findings report, indicating whether fraud was detected. Los Angeles County’s procedure is for eligibility workers to complete the bottom portion of the findings report, noting the actions taken based on the investigation; enter these actions into the eligibility system; and return the findings report to the investigator. However, Los Angeles County indicated that in practice, eligibility workers do not always return the findings report, nor do they always enter the actions noted into the eligibility system. Finally, in response to our inquiries, Los Angeles County stated that it has been underreporting the results of ongoing investigations, but because of weaknesses in its reporting practices, it could not provide any data to support its belief. Due to these weaknesses, we could not verify the accuracy of the county’s investigation activity reports. The four other counties’ reports also contained errors, but to a lesser degree. For example, Sacramento County had several errors in its investigation activity reports dating back to August 2006. These errors included submitting some of the prior months’ data in the next month’s report or not reporting all required statistics. Sacramento County discovered these errors before our review started in March 2009, and in August 2009 it resubmitted corrected investigation activity reports to Social Services, which we used in our cost‑benefit analysis. Although the county’s corrected reports agreed to supporting case listings, information in the case files did not agree to the case listings in some instances. The other three counties could not support some of the information included in their reports. For example, Orange County’s investigation activity report for January 2008 included reported overpayments totaling approximately $17,200 for CalWORKs cases that were investigated for fraud. However, when asked to identify the related cases, Orange County provided a list of CalWORKs cases with overpayments that totaled $31,900, or more than $14,000 over the amount that it reported. We noted similar problems when we asked Riverside and San Diego counties for the list of cases they reported on their investigation activity reports. Based on our testing, these differences occurred because of a lack of documentation to support all numbers reported or because of clerical errors. Furthermore, the six counties we visited are inconsistently reporting the actions resulting from ongoing investigations, a situation that hinders the ability to compare the counties’ investigation activity reports. Specifically, three counties are reporting the actual actions that eligibility workers took based on the results of the investigations, while two other counties are reporting their investigators’ recommended actions. Until 2008 another county reported 32 California State Auditor Report 2009-101 November 2009 both actual actions taken and recommended actions, but now it reports only actual actions taken. The two counties reporting their investigators’ recommended actions informed us that in most but not all instances, the recommendations are the same as the actions that eligibility workers took. The counties we visited generally do little to ensure the accuracy of the investigation activity reports and most of them assign a single staff member to complete and submit these reports. These staff members prepare the reports by consolidating data provided to them from different sources. In the instructions for the investigation activity report, Social Services places the responsibility for ensuring the accuracy and completeness of the reports on the county, and it requires counties to provide contact information for the person who ensured the accuracy of reports before submittal. Most of the counties we visited acknowledged that they perform little to no review of the reports before submitting them to Social Services. Social Services’ Procedures for Reviewing Counties’ Investigation Activity Reports Are Inadequate Compounding the counties’ reporting problems is the fact that Social Services does not perform a sufficient review of the investigation activity reports that counties submit. Social Services subjects these reports to an automated review to ensure that the figures reported are reasonable. This review involves comparing the current month’s report to the report for the prior month to identify changes. If the figures in the current report differ by more than 3 percent to 20 percent from those for the previous month (depending on the size of the county), Social Services will check to see if the county submitted an explanation or if the change is the result of a seasonal variation. Social Services also may contact the counties to verify the reason for the change and, when necessary, request that the county submit a revised report. Social Services’ minimal review However, this minimal review is inadequate to detect even the is inadequate to detect even the most glaring errors in the data that counties submit. Social Services most glaring errors in the data that annually compiles the data reported by counties on their investigation counties submit. activity reports so that it can prepare a statewide investigation activity report. When creating this report, Social Services does not follow up on discrepancies, such as potential underreporting of activity by counties. For example, Los Angeles County—representing 30 percent of the State’s CalWORKs caseload—is the largest county in the State, with the next largest county having just over 7 percent of the State’s caseload. Given that its caseload is more than four times larger than that of the next largest county, it is reasonable to assume that Los Angeles County would report the highest number of cases California State Auditor Report 2009-101 33 November 2009 in most categories on the investigation activity report. However, our review of the fiscal year 2007–08 data found that this did not occur, and some discrepancies existed that Social Services should have scrutinized. For example, although the counties reported reducing benefits on a total of nearly 5,000 cases during fiscal year 2007–08 as a result of ongoing investigations, only 41 of those cases were reported by Los Angeles County, a number that seems quite low considering that the county spent over $23 million on ongoing investigations during 2008—the highest by far among the counties we reviewed. Another large county, Alameda, reported no cases in this category. As we noted previously, both counties told us that they believe they have inadvertently been underreporting the number of cases in this category, but neither could provide support for their assertions. Besides this example, we noted other instances in which the information that Los Angeles and other counties reported appeared inconsistent with the size of their caseloads. We believe that if Social Services obtains these data, it should follow up with the counties on potential reporting discrepancies to determine if the data reported are in error or if the discrepancies are the result of a county’s process that either needs improvement or might be a best practice for other counties. Social Services informed us that it reviews the investigation activity report during periodic Income and Eligibility Verification System reviews, which we discuss in Chapter 2. During these reviews, Social Services indicated that it compares the investigation activity reports with documentation located at the counties and makes any appropriate findings and recommendations, and stated that it routinely finds that counties inaccurately report data. Some of the inconsistencies we noted during our review are Some of the inconsistencies we possibly due to unclear instructions. Three of the counties we noted during our review are visited told us that they believe the instructions are unclear, possibly due to Social Services’ and they indicated that when they contacted Social Services for unclear instructions for the additional clarification, its answers were not always helpful. To investigation activity report. address these and other types of concerns, in January 2009 Social Services created an informal workgroup to work on potential revisions to the investigation activity report. The workgroup includes both Social Services’ staff and staff representing 11 counties and has met twice, in January 2009 and February 2009. According to the minutes of the last meeting, the workgroup’s role is to identify information requested on the investigation activity report that needs to be better defined, determine whether any information on the report is no longer needed by Social Services and stakeholders, and ascertain whether any additional information is needed that is not currently requested. As a result of the workgroup’s February 2009 meeting, Social Services and one county developed scenarios for each county participating on the workgroup to use in completing an investigation activity report. By a process of comparing how counties complete reports with 34 California State Auditor Report 2009-101 November 2009 identical information, Social Services expects to identify confusing areas of the investigation activity report. According to Social Services, the workgroup has not met since February 2009 because the participating counties are still responding to the scenarios. However, more frequent meetings would better ensure that the efforts of this workgroup are put to good use to improve the counties’ reporting efforts. We reported similar findings in a report issued in 1995, which found that the investigation activity reports for some counties contained errors and that Social Services did not perform sufficient review of county data to identify potential errors.6 Our prior report also found that Social Services did not provide thorough instructions to help ensure that the investigation activity reports were completed consistently, and that some counties could not provide support for the figures they reported. In a 1995 report, we recommended that Social Services clarify its instructions for completing the investigation activity report, develop an ongoing desk review process of these reports for consistency and reasonableness, and provide timely feedback to the counties when errors are noted. We further recommended that Social Services thoroughly communicate its record retention policy to the counties. It is apparent that Social Services has not adequately addressed the concerns raised in our previous report, as the problems still persist. Moreover, by not promptly addressing these issues, Social Services will continue to relay erroneous information to its management, the Legislature, and the federal government as discussed in the next section. Errors in the Counties’ Investigative Reports Are Passed on to Other Parties Social Services produces various reports based on the questionable information counties submit in their monthly investigation activity reports and provides them to the federal government, internal users, and the Legislature. The federal government requires Social Services to report on welfare fraud investigation activity related to the food stamp program but not CalWORKs. Social Services uses this report, along with other information, to substantiate to the USDA the costs it expects to incur during the next year. The USDA is authorized to reimburse states for up to 50 percent of the administrative costs involved in their operation of the food stamp program. The USDA told us that it uses the investigation activity data as reference information when making these funding decisions. For example, if Social Services tells the USDA that 6 The Bureau of State Audits’ report titled Department of Social Services: Review and Assessment of the Cost‑Effectiveness of AFDC Fraud Detection Programs (Report 94023, March 1995). California State Auditor Report 2009-101 35 November 2009 it expects an increase in costs, the USDA said it would refer to the investigation activity information to observe the trend and decide whether the increase requested is justifiable. However, because of the inaccuracies we found, the USDA might be basing Because of the inaccuracies the administrative payment for California on information that we found, the United States contains errors. Additionally, the USDA told us that it uses the Department of Agriculture might information for planning reviews, ad hoc studies and reports, and be basing the administrative formal reports, and that the information is also a component of payment for California on the data used to develop budget projections for the United States erroneous information. Congress. However, the federal Department of Health and Human Services, which administers the Temporary Assistance for Needy Families (TANF) program—the source of much of the funding for CalWORKs—does not require or use any of the information included on the investigation activity report, according to a regional TANF program manager. Social Services also uses the investigation activity reports to prepare its internal quarterly fraud investigation activity report, and for other special studies and legislative reports. The quarterly fraud investigation activity report summarizes the monthly investigation activity reports submitted by the counties, including the number and percentage of investigations received, accepted, rejected, and completed, and the results of the completed investigations. Social Services publishes this report on its Web site for the public and other interested parties. Further, Social Services uses the investigation activity report for special studies, such as the 10‑year study of welfare fraud trends that it completed in 2006. The intent of the study was to assess trends and identify best practices related to early and ongoing welfare fraud investigations, overpayment collections, and county administrative practices. According to the chief of Social Services’ Emergency Food Assistance and Fraud Bureau, the 10‑year study presented no clear trend that would help identify best practices. Social Services’ staff also indicated that Social Services provides members of the Legislature with data on welfare fraud investigations when requested. As a result, of the errors and inconsistencies we found, each of these other reports will also contain errors, which could mislead users of the data. Additionally, the instructions for the investigation activity report indicate that Social Services uses the reports to evaluate the effectiveness of fraud prevention and detection programs, to evaluate the effectiveness of local agencies’ policies, and to plan with local agencies for any needed changes. However, without reliable data from the counties, Social Services and stakeholders in the State’s CalWORKs and food stamp programs cannot make informed decisions to improve the State’s administration of these programs. 36 California State Auditor Report 2009-101 November 2009 Recommendations To ensure that all counties consistently gauge the cost‑effectiveness of their early fraud activities and ongoing investigation efforts for the CalWORKs and food stamp programs, Social Services should work with the counties to develop a formula to regularly perform a cost‑effectiveness analysis using information that the counties currently submit. To make certain that counties receive the greatest benefit from the resources they spend on antifraud efforts related to CalWORKs and food stamp cases, Social Services should do the following: • Using the results from the recommended cost‑effectiveness analysis, determine why some counties’ efforts to combat welfare fraud are more cost‑effective than others. • Seek to replicate the most cost‑effective practices among all counties. • Continue to address the recommendations of the steering committee and promptly act on the remaining recommendations. To ensure the accuracy and consistency of the information on welfare fraud activities that counties report and that Social Services subsequently reports to the federal government, the Legislature, and internal users, Social Services should take the following steps: • Remind counties that they are responsible for reviewing the accuracy and consistency of investigation activity reports before submission. • Perform more diligent reviews of the counties’ investigation activity reports to verify the accuracy of the information submitted. • Provide counties with feedback on how to correct and prevent errors that it detects during this review. • Continue with regular meetings of its workgroup to further its efforts to clarify its instructions for completing the counties’ investigation activity reports. California State Auditor Report 2009-101 37 November 2009 Chapter 2 THE DEPARTMENT OF SOCIAL SERVICES AND THE COUNTIES COULD IMPROVE THEIR ONGOING EFFORTS TO COMBAT WELFARE FRAUD Chapter Summary The Department of Social Services (Social Services) for the State of California (State) does not ensure that counties consistently follow up on lists it provides them that may match the names of welfare recipients to information received from various sources that might affect welfare recipients’ eligibility (match lists). We found that some counties did not follow up consistently on these matches as required by federal law, or the counties could not always demonstrate their follow‑up efforts. Further, some counties have noted that certain match lists are not as useful as they could be in their current format; as a result, these counties perform limited or no follow‑up on these lists. We also determined that Social Services is missing opportunities to improve the counties’ efforts because it does not visit all counties on a regular basis. For example, Social Services noted that it has not reviewed Los Angeles County’s follow‑up efforts on the match lists since 2005. Moreover, although Social Services asserted that the Statewide Fingerprint Imaging System (SFIS) deters fraud that the Income and Eligibility Verification System (IEVS) cannot identify, it has not adequately determined the cost‑effectiveness of SFIS. According to Social Services’ data, many counties are not promptly following up on matches generated by SFIS. In addition, Social Services has not taken the necessary steps to claim its share of $42.1 million in aid overpayments that the counties have collected, nor has it released the shares of these funds due to the counties and the United States Department of Agriculture (USDA). Moreover, the USDA has expressed long‑standing concerns about the accuracy of the overpayment collection information reported by the counties, which Social Services does not review for accuracy. Lastly, county size, demographics, and county department staffing necessitate different approaches to investigating and prosecuting welfare fraud. In response to workload and staffing issues, counties have developed prosecution thresholds below which the district attorney’s office will generally not accept cases referred for prosecution. Nevertheless, as of January 1, 2008, counties reported a backlog of nearly 6,400 referrals statewide, which had only decreased by 12 percent by the end of the year. Of the 13,200 cases 38 California State Auditor Report 2009-101 November 2009 referred for prosecution that were available for counties to work on during 2008, counties acted on 5,074 cases by prosecuting 3,164 cases and deciding not to prosecute 1,910 cases. Social Services Does Not Ensure That Counties Consistently Follow Up on Welfare Fraud Matches Social Services does not ensure that counties consistently follow up on information it provides them that might affect welfare recipients’ eligibility. As we discussed in the Introduction, federal and state regulations require that Social Services use IEVS and establish additional systems to generate lists of potential matches and follow up on those matches to prevent ineligible persons from receiving aid. There are 10 match lists provided by federal and state agencies, including the federal Social Security Administration (Social Security), Internal Revenue Service, and Franchise Tax Board. Social Services distributes the match lists to counties on a periodic basis—monthly, quarterly, semiannually, and annually—for follow‑up. Most of these lists are in paper form. For six of the 10 match lists, federal regulations mandate that the State must, within 45 days of receiving the match information, notify the welfare recipient of an intended action—a discontinuance of or reduction in benefits—or indicate that no action is required. In California, Social Services’ regulations require the county welfare departments to conduct this follow‑up on behalf of Social Services. For the remaining four match lists there is no mandated time period for review, and each county’s special investigative unit, located in either the welfare department or the district attorney’s office, conducts the follow‑up. As Table 5 shows, none of the counties we reviewed consistently followed up on all of the match lists that had to be completed within the 45‑day timeline. As we discussed in the Scope and Methodology, we did not review two match lists due to federal restrictions. The imposition of a 45‑day time frame to review and respond concerning these match lists implies a need to review them more quickly than match lists without a deadline. However, the results of our testing shows that counties are struggling to consistently do so, which lessens the value of these match lists as a means to detect fraud. For the four match lists without a time requirement, our testing showed that San Diego was fairly consistent in completing the matches in our sample, while the other four counties were not, as shown in Table 6 on page 40. According to Alameda County’s lieutenant of inspectors, he processes parts of all matches when time allows and consistently spends time on only one match list. Among the other three counties, Los Angeles County had no follow‑up on two of these four match lists. Los Angeles County California State Auditor Report 2009-101 39 November 2009 indicated that it does not process the nationwide prisoner match list because the matches proved to be unproductive, containing information that was too outdated and voluminous for its limited staff to handle effectively. Additionally, Los Angeles County acknowledged that it has a backlog of fleeing felon match lists dating back to 2007 due to a shortage of investigative staff. These inconsistent efforts undermine the intent of the match lists, which is to provide information to the counties that, if acted on, could affect a recipient’s eligibility or benefit amounts and, if undetected, could lead to fraud. Table 5 Status of Five Counties’ Follow‑up Efforts on Match Lists With a 45‑Day Time Requirement April 2008 Through March 2009 INTEGRATED FRANCHISE EARNINGS PAYMENT TAX BOARD CLEARANCE FRAUD NEW HIRE VERIFICATION COUNTY ASSET MATCH DETECTION REGISTRY SYSTEM Alameda  t t t Los Angeles t t  t Riverside *  t t Sacramento   t t San Diego   t t Source: Bureau of State Audits’ review of matches provided to counties during April 2008 through March 2009. Notes: As discussed in the Scope and Methodology, we did not test whether Orange County followed up on match lists. We could not review the appropriateness of counties’ follow‑up efforts for two lists—the Beneficiary Earnings Exchange Records and Internal Revenue Service asset lists—because they contain federal tax information, and federal law expressly limits disclosure of this information. * Because documentation was absent, we could not determine whether the county followed up. =The county completed all of the matches in our sample on time. =Some matches were completed late. t =Some matches were not completed and/or some were completed late. =No matches were completed. Counties Assert That the Format of Some Match Lists Could Be Improved According to representatives from the five counties whose match list follow‑up we reviewed, the format of some match lists could be improved to make them more efficient to use. For example, San Diego County indicated that the nationwide prisoner match, which Social Security produces, does not yield many positive results because the facilities entering the information report only when an individual is incarcerated and do not remove the information upon 40 California State Auditor Report 2009-101 November 2009 release, causing the individual to remain on that match list. Thus, when individuals who were incarcerated in the past apply for or receive aid, their names appear on the nationwide prisoner match list. Consequently, the county is not able to determine whether individuals appearing on the match list have been released from prison within the last month. Therefore, the county follows up only on those individuals whom the list shows were imprisoned during the last 90 days. For a similar reason, Los Angeles County told us that it does not use the nationwide prisoner match list. Table 6 Status of Five Counties’ Follow‑Up Efforts on Match Lists Without a 45‑Day Time Requirement April 2008 Through March 2009 DECEASED CALIFORNIA YOUTH FLEEING NATIONWIDE COUNTY PERSONS MATCH AUTHORITY MATCH FELON MATCH PRISONER MATCH Alameda * * * * Los Angeles     Riverside *  t t Sacramento † †  † San Diego    t Source: Bureau of State Audits’ review of matches provided to counties during April 2008 through March 2009. Note: As discussed in the Scope and Methodology, we did not determine whether Orange County followed up on match lists. * Because documentation was absent, we could not determine whether the county followed up. † During our fieldwork, Sacramento County indicated that it did not have match lists available for our review. Subsequent to the completion of our fieldwork, Sacramento County informed us that it did have these lists. =The county completed all of the matches in our sample. t=Some matches were not completed. =No matches were completed. In 2003, in response to counties’ concerns, Social Services asked the USDA to revise the nationwide prisoner match list to include prisoners’ release dates. In a letter to the USDA, Social Services indicated that the prisoner release date was the single most important data element because it would allow Social Services to eliminate superfluous information and improve the quality of the data provided to the counties. However, the USDA responded that while Social Security acknowledged the advantage of tailoring reports for specific agencies, such as Social Services, the fact that Social Security handles more than 4,000 prisoner reports makes it infeasible to accommodate individual formatting requests. Social Services indicated that it has not spoken to either federal agency since 2003 about this issue. Although Social Services stated that it California State Auditor Report 2009-101 41 November 2009 has not received any additional formal complaints from counties Although Social Services stated about this match list, two of the five counties we reviewed believe it has not received any recent that the format of the nationwide prisoner match list is not useful complaints from counties, and all the counties we reviewed make limited or no use of this list. two counties we reviewed believe that the format of the nationwide In an additional example, Sacramento County had concerns with prisoner match list is not useful and the deceased persons match, noting that the list often contains all the counties we reviewed make duplicate or mismatched Social Security numbers. Finally, limited or no use of this list. Riverside County indicated that in 2008 it received more than 4,000 matches for the beneficiary earnings exchange records match list, but only six matches resulted in identified overpayments.7 It determined that many of the matches were duplicates reported in previous months and stated that this review was not an effective use of staff time. Additionally, seven of the 10 match lists are provided to the counties as paper copies, including three that have a required 45‑day timeline. Three lists—the payment verification system, new hire registry, and integrated earnings clearance fraud detection match lists—are sent in an electronic format. Social Services indicated that four of the remaining lists are provided to some counties in electronic format depending on the county’s welfare database system. All five counties we visited told us that having all match lists in electronic form would allow them to process matches more efficiently. For example, Sacramento County indicated that the match lists received on paper require additional steps that could be avoided if they were received in electronic data files. Sacramento County believes that receiving all of these lists electronically would enable it to computerize the processing to automatically remove invalid matches and identify workable matches. Similarly, Los Angeles County asserted that processing electronic matches is more efficient than using printed copies for several reasons, such as making matches more readily available to all staff, reducing the number of lost abstracts, conducting demographic match validation, enabling supervisors to control the processing and validation of pending cases, and improving the ability to generate internal reports and reports forwarded to the State. Social Services indicated that it has a long‑term intent to modernize its business practices by moving from paper to electronic transmission of match lists. Because of the federal and state mandated follow‑up on these lists, much of which must occur within a specified time frame, Social Services should continue to explore ways to provide these reports in an electronic format so that it can help the counties follow up on them more effectively and in a timely manner. 7 We could not review the appropriateness of counties’ follow‑up efforts for two lists—the Beneficiary Earnings Exchange Records and Internal Revenue Service asset lists—because they contain federal tax information, and federal law expressly limits disclosure of this information. 42 California State Auditor Report 2009-101 November 2009 Social Services’ Monitoring of Counties’ Follow‑Up Efforts Is Weak Although Social Services has a process in place to monitor the counties’ efforts to follow up on match lists, it is missing opportunities to improve their efforts because it does not visit all counties on a regular basis and does not always enforce recommendations from the reviews that it does perform. In its communication with the counties, Social Services indicated that its periodic IEVS reviews are to determine counties’ compliance with state and federal statutes, assess the effectiveness of specific procedures, and provide counties feedback on any problems observed. However, because it asserts that it lacks resources, Social Services has not been able to review the counties’ efforts on a regular basis. Specifically, it has not reviewed 25 of the 58 counties during the three‑year period from August 2006 to August 2009. Among the 25 Social Services’ review log, unreviewed counties is Los Angeles, which represents approximately which represents approximately 30 percent of the statewide California Work Opportunities and 30 percent of the statewide Responsibility to Kids (CalWORKs) cases. Social Services’ review log CalWORKs cases, shows that dates back to 1995 and shows that Los Angeles County has not been Los Angeles County has not been reviewed since 2005 and that five small counties—Trinity, Modoc, reviewed since 2005 and that Inyo, Mono, and Alpine—have never been reviewed. Social Services five small counties have never conducted 21 IEVS on‑site reviews and three desk reviews in fiscal been reviewed. year 2008–09, but it has scheduled on‑site IEVS reviews for only 12 counties in fiscal year 2009–10. This schedule amounts to less than half of the prior year’s IEVS reviews and does not include Los Angeles County or four of the five small counties it has never reviewed. According to the chief of the Emergency Food Assistance and Fraud Bureau, Social Services has not reviewed Los Angeles County’s match procedures, but has reviewed its overpayment collection process. The IEVS reviews are intended to be part of Social Services’ oversight of the counties’ efforts to detect and prevent welfare fraud. Because Social Services is ultimately responsible—according to federal regulations—for processing matches consistently and in a timely way, and because the counties conduct these efforts on its behalf, Social Services is failing to ensure that it is in compliance with these regulations. The need for the IEVS reviews that include a review of match list follow‑up is evident, particularly given the extent of noncompliance we found among the five counties listed in tables 5 and 6. We also found that when Social Services does conduct IEVS reviews, it does not always ensure that the counties correct the problems it identifies. Within the past two years, Social Services conducted California State Auditor Report 2009-101 43 November 2009 IEVS reviews of four counties we reviewed8—all of which are considered large counties—and noted some problems, but to a more limited extent than our review found. For example, Social Services reviewed Riverside County in September 2007 and found that it had a backlog in processing the match list for the Integrated Earnings Clearance Fraud Detection System, and this backlog led to the county’s not performing follow‑up on these matches within the required 45‑day timeline. To reduce the backlog and bring the county into compliance with the 45‑day timeline, Social Services recommended several changes to Riverside County’s procedures. Social Services indicated that it would follow up with the county in six months to verify that the 45‑day requirement was being met and that the backlog had been reduced. However, as of September 2009—or nearly two years later—Social Services had not followed up or confirmed that the county had implemented its recommendations. Social Services’ chief of the Emergency Food Assistance and Fraud Bureau stated that the follow‑up was missed due to staff turnover. Riverside County asserted that although Social Services did not conduct follow‑up, the county rectified the problem by the end of Social Services’ visit. We were able to confirm that for the sample we tested of the county’s follow‑up efforts for this match list shown in Table 5 on page 39, that Riverside County completed all follow‑up within 45 days. In another instance, Social Services performed a review of San Diego County in February 2009 and found that the county was not following through on the processing of matches from the payment verification system and the new‑hire registry. It recommended that the county provide refresher training to all caseworkers in the next 30 days to help them understand the importance of completing these two matches and also that it prepare a plan of action within 30 days. In this instance, Social Services indicated that it contacted San Diego County, which asserted that the training was completed for all staff. Social Services stated that its system for assuring that counties implement any corrective measures it recommends is to review the counties’ written responses and to rely on assertions stating that the counties have made the necessary corrections in their processes. Social Services further indicated that it conducts a thorough review of the corrective action plans during its next review of affected counties. When we tested San Diego County’s follow‑up efforts using a sample containing these two match lists that were provided to the county during April 2008 through March 2009, we found that the county did not always complete these matches. 8 As noted in this report’s Scope and Methodology section, although we visited six counties, we did not review Orange County’s efforts to follow up on matches. 44 California State Auditor Report 2009-101 November 2009 Social Services Has Not Done a Cost‑Benefit Analysis of SFIS Although asserting the Statewide Social Services asserts that SFIS identifies duplicate‑aid fraud Fingerprint Imaging System (SFIS) that IEVS does not. Further, Social Services and counties deters fraud, Social Services has not told us that they believe that the awareness of the fingerprint done a cost‑benefit analysis of SFIS requirement deters individuals from fraudulently applying for aid in because it believes there is no way to multiple counties. Although it believes that SFIS plays an important measure deterrence. role in deterring fraud, Social Services has not done a cost‑benefit analysis of SFIS because it believes there is no way to measure the deterrence effect of the system. Although federal regulations do not require the use of fingerprint imaging technology, in 2000 Social Services implemented SFIS, which Social Services based on Los Angeles County’s Automated Fingerprint Image Reporting and Match system, in accordance with state law. When justifying the implementation of SFIS, Social Services did not conduct its own study; instead, it used the estimates from an evaluation Los Angeles County performed in 1997 to project statewide savings that would result from SFIS. In this evaluation, the county estimated that the overall net savings related to its Automated Fingerprint Image Reporting and Match system would be between $52.5 million and $64.6 million. Because Los Angeles County’s CalWORKs and food stamp caseload constituted approximately 40 percent of all such cases in the State at that time, Social Services believed that it was reasonable to extrapolate from Los Angeles County to the rest of the State. However, in a report we issued in 2003,9 we expressed concern that Social Services’ methodology of projecting statewide savings using Los Angeles County’s estimated savings was flawed, especially in its assumption that the incidence of duplicate‑aid fraud in Los Angeles County was representative of the incidence of this type of fraud statewide. In fact, the 2003 report found that data reported by other counties before the implementation of SFIS did not suggest that duplicate‑aid fraud was extensive enough to warrant the cost of SFIS. In that report, we recommended that Social Services fully account for the cost of SFIS by collecting data and tracking administrative costs related to SFIS to measure its cost‑effectiveness. Social Services chose not to implement our recommendation because it stated that it includes SFIS as part of the eligibility determination activities to which counties charge time and that reprogramming its system to separate SFIS activities would be too costly. 9 Bureau of State Audits’ report titled Statewide Fingerprint Imaging System: The State Must Weigh Factors Other Than Need and Cost‑Effectiveness When Determining Future Funding for the System (Report 2001‑015, January 2003). California State Auditor Report 2009-101 45 November 2009 Studies that Social Services conducted in 2005 and 2009 have Studies that Social Services concluded that SFIS identifies fraud that IEVS and other eligibility conducted in 2005 and 2009 have determination procedures do not. Social Services provided us concluded that SFIS identifies with a summary of an analysis that it performed in 2005 of 28 fraud that IEVS and other eligibility welfare applications that were rejected because SFIS identified that determination procedures do not. the applicants were attempting to receive aid under two separate identities. According to the summary, IEVS did not detect 24 of the 28 fraudulent applications. The summary indicated that IEVS failed to detect these cases because the individuals applying for aid used legitimate names, Social Security numbers, and dates of birth. Social Services concluded that IEVS would detect only cases involving unsophisticated fraud. In early 2009 Social Services compared SFIS to another process, known as file clearance, to determine whether it could replace SFIS. File clearance is a process that counties use to determine whether an applicant for aid has received or is currently receiving CalWORKs, food stamp, or benefits for the Medical Assistance Program (Medi‑Cal), and involves checking the State’s Medi‑Cal Eligibility Data System. In early 2009 Social Services surveyed the 19 largest counties to ask whether they believe that file clearance could be used in lieu of SFIS to prevent duplicate aid fraud. Ten counties responded to the survey and nine of the 10 counties indicated that both SFIS and file clearance were necessary. Social Services reviewed the matches generated by SFIS between December 2008 and January 2009. Because the majority of these cases occurred in Los Angeles and Sacramento counties, Social Services reviewed 65 cases from these counties and found that only five related to fraud. The remaining cases were not fraudulent; instead, they had experienced administrative errors caused by multiple or incorrect client identification numbers. Both Los Angeles and Sacramento counties indicated that they use SFIS to identify and correct mistakes associated with client identification numbers. As a result of this comparison, Social Services determined that counties see file clearance as an eligibility check and SFIS as a means to verify identification and prevent fraud. Some states, including Texas, Connecticut, Arizona, and New York, use fingerprint imaging systems similar to SFIS to detect duplicate aid fraud. In 1999 Texas completed an evaluation of the deterrent effect of its fingerprint imaging system and projected a savings of between $5.9 million and $11.6 million per year. Although Texas, Connecticut, and New York have not evaluated the ongoing effects and savings from their fingerprint imaging systems, Arizona is required by state law to conduct a yearly cost analysis of its fingerprint imaging program. To do so, Arizona calculates the value of costs avoided as a result of attempts by applicants to receive duplicate aid as identified by its fingerprint imaging system, and it also tracks and calculates the costs avoided for the number of cases 46 California State Auditor Report 2009-101 November 2009 closed because applicants refused to undergo fingerprint imaging. Arizona then compares this total benefit with the annual cost of the contract to operate the system to identify potential savings during Arizona calculates the value of costs the year. In fiscal year 2007–08, Arizona had an annual cost of avoided as a result of attempts by $874,000 for its fingerprint imaging program; however, this state applicants to receive duplicate aid calculated that it deterred individuals in approximately 2,200 cases and the number of cases closed of potential fraud, for a projected savings of $10 million, because because applicants refused to the individuals refused to be fingerprinted. In addition, Arizona undergo fingerprint imaging, which identified 10 cases of duplicate‑aid fraud, for a savings of $46,000. it then compares to the annual cost When comparing these savings to its contract cost of $874,000, to operate its system. Arizona concluded its net savings was $9.2 million. However, the large and ongoing historical backlog of SFIS results awaiting resolution by county staff raises questions of how counties are using SFIS in deterring fraud. Social Services tracks the number of cases that have generated an unexpected SFIS result—indicating potential duplicate‑aid fraud—that have awaited resolution by county staff for more than 60 days. As of July 31, 2009, the statewide backlog was more than 13,700 unresolved cases. The backlog per county ranged from no unresolved cases to almost 3,700. We asked Social Services what actions it takes to ensure that counties are addressing the backlog. Social Services told us that it had previously monitored the backlog and sent backlog reports to the counties. However, according to Social Services, it has discontinued this practice because of staffing and workload issues within Social Services and because of the fact that state law and regulations do not require counties to process the backlog. Social Services also noted that the counties have access to backlog information through SFIS. Table 7 shows backlog information for the six counties we reviewed as well as for those counties with the highest backlogs in the State. We contacted the six counties we reviewed and three additional counties with the highest backlogs—San Bernardino, Tulare, and Santa Clara—to understand how they approach processing SFIS results. Several were not aware of the size of their respective backlogs. Others, when asked why they had not followed up on their backlogs, indicated that staff turnover or increased welfare caseloads may have limited the time available for county staff to follow up on SFIS results. One county indicated that county staff charged with SFIS processing might not have received sufficient training. In addition, Alameda and Orange counties suggested that a large portion of their backlog was likely due to clerical errors rather than potential duplicate‑aid fraud. Interestingly, Los Angeles County had just 33 unresolved cases, which it attributes to extensive staff training and the fact that a welfare fraud investigator must clear an unexpected SFIS result before benefits can be approved. San Diego County had reduced its backlog from a high of almost 4,800 to 587 over the previous year, which it California State Auditor Report 2009-101 47 November 2009 Table 7 Backlog of Unresolved Statewide Fingerprint Imaging System Matches as of July 31, 2009, for Select Counties BACKLOG AS OF COUNTY JULY 31, 2009 San Bernardino 3,686 Alameda 2,597 Tulare 1,266 Santa Clara 1,161 San Diego 587 Orange 511 Riverside 287 Sacramento 189 Los Angeles 33 Subtotal 10,317 All Other Counties 3,399 Statewide Total 13,716 Sources: Aging resolution reports from the Department of Social Services’ Statewide Fingerprint Imaging System. attributed to the efforts of a dedicated staff member. However, San Diego County did not identify any fraud as a result of these efforts, and most of the counties we contacted also indicated that they did not identify any fraud as a result of SFIS. Most of the counties we contacted indicated that, in concept, they believe SFIS is a deterrent to duplicate aid fraud. Nevertheless, if counties do not review the backlog, there is no way to know whether it contains potential fraudulent cases. Despite the size of the backlog, the number of duplicate‑aid The number of duplicate‑aid cases cases SFIS has detected is fairly low, given its cost. In 2008 Social SFIS has detected is fairly low; Services data show that statewide the counties used SFIS to in 2008 the counties identified identify 54 cases of duplicate‑aid fraud, and they have identified a 54 cases of duplicate‑aid fraud total of 845 instances of fraud through SFIS since its implementation and they have identified a total of in 2000. Social Services asserted that SFIS does not identify many 845 instances of fraud through SFIS cases because it deters people from applying for duplicate aid. It since its implementation in 2000. noted that it has not performed a cost‑effectiveness analysis because the chief benefit of SFIS is that it keeps people from applying for aid fraudulently, a benefit that it asserts cannot be measured. We acknowledge that fraud deterrence is difficult to measure. However, because the State is spending approximately $5 million per year to maintain SFIS, Social Services has an obligation to justify whether the continued use of SFIS is cost‑beneficial to the State. 48 California State Auditor Report 2009-101 November 2009 Social Services Has Not Taken the Necessary Steps to Claim Its Share of $42.1 Million in Food Stamp Overpayment Collections Since December 2003 counties have collected and deposited into trust funds more than $42.1 million in overpayments recovered from food stamp recipients. However, Social Services has been delayed in taking the steps needed to claim its share of these overpayments or to distribute the shares of these funds due to the administering federal agency, the USDA, and release the share due the counties. Overpayments to food stamp recipients can result from administrative errors by counties or inadvertent errors or fraud by recipients. Once counties identify the errors or fraud, they calculate the amount overpaid to recipients and set up a collection account. Counties collect the overpayments from recipients through various means, including tax refunds intercepted and held by the federal government. However, the counties we reviewed deposited the actual cash they collect in their bank accounts and receive the interest earnings on these collections until Social Services claims its and the federal government’s share. Counties report the collections to Social Services every quarter on a USDA Status of Claims Against Households report (collections report), and Social Services is responsible for calculating the state, federal, and county shares of the overpayments collected. To calculate the respective shares, Social Services must work with the USDA to reconcile funds collected through federal tax refund intercepts, which the USDA retains, with the funds counties have reported to Social Services on the collections report. Once Social Services and the USDA reconcile the total funds collected, Social Services calculates the USDA’s share10 of the total collections, with the remainder of the funds split evenly between the State and the county. To accomplish the distribution, Social Services, with USDA approval, offsets each county’s claim for reimbursement of administrative expenses in the following month by the state and federal share of the collections. Similarly, Social Services gives the federal share to the USDA by reducing its own future claims for food stamp administrative funds by the federal share. Social Services asserted that several problems have delayed its efforts to distribute the $42.1 million in overpayments that have accumulated. According to a manager in Social Services’ federal reporting section, turnover of staff assigned to this task in Social Services has delayed its efforts to reconcile the tax intercepts and county collections for the food stamp program overpayments 10 The federal share is 65 percent for fraud claims and for claims due to inadvertent errors that are offset against a person’s unemployment compensation, and it is 80 percent for all other claims due to inadvertent errors. California State Auditor Report 2009-101 49 November 2009 for the past six years. A USDA financial management specialist and the Social Services manager stated that they began to work on the delayed reconciliations in early 2008; however, the USDA does not expect this process to be completed until September 2010. In August 2009 the USDA directed Social Services to halt the reconciliations so they can coordinate the application of these funds. Another problem is caused by the inaccuracies in the counties’ collection reports. The USDA has a long‑standing concern regarding the accuracy of the collection reports submitted by the counties. In March 2003, at the direction of the USDA, Social Services requested that each county welfare administrator certify that the collections report was accurate and properly completed. Subsequently, Social Services notified the counties that in January 2006, the USDA reviewed these monthly collection reports and raised serious concerns about the accuracy of the information reported, due to the inconsistencies among counties in completing the collection reports. In response to the USDA’s concerns, in July 2007, Social Services requested that the 19 largest counties verify the accuracy of the information reported for one quarter—April to June 2007. However, most of these 19 counties were unable to verify the information contained in their reports, according to a June 2008 statewide letter to counties from Social Services. Consequently, Social Services, at the USDA’s direction, required counties to perform additional verification of the reports. This verification included having the reports reviewed by a county or independent auditor and documenting the review. Additionally, Social Services required the counties to make adjustments for any errors detected on their next collection report and to explain the nature of the errors found. Counties had to complete this verification and notify Social Services of the results by August 31, 2008. Even so, the completion of this verification process took until April 2009, when the last county successfully validated its report. Social Services asserted that it plans to incorporate the validation of collection reports into future IEVS reviews, but it has not determined yet how it will accomplish this validation. However, as we noted previously, Social Services is not performing IEVS reviews on a regular basis. Because Social Services has been delayed in addressing this issue, Because the counties place the cash the USDA, counties, and Social Services have not had access to collected in their bank accounts these funds. Social Services records show that of the $42.1 million and earn interest on the cash, balance, $17.2 million would go to the USDA, with the remaining we estimate that Social Services $24.9 million split between Social Services and the counties may have lost approximately (almost $12.5 million each). Moreover, because the counties place $1.1 million in interest during the cash collected in their bank accounts and earn interest on the the six‑year delay for its share cash, we estimate that Social Services may have lost approximately of the unclaimed funds. $1.1 million in interest during the six‑year delay on its share of the 50 California State Auditor Report 2009-101 November 2009 unclaimed funds. Similarly, the USDA also was not able to earn interest on these funds. Given the difficult budgetary issues that all governments are currently facing, it is imperative that Social Services obtain all funds due it and ensure that it and the USDA promptly receive their share of these funds. Because counties are holding the cash collected in their bank accounts, neither Social Services nor USDA have received any of the interest earned on the collections. Investigation and Prosecution Efforts Vary by County County size, demographics, and county department staffing necessitate different approaches to investigating and prosecuting welfare fraud. Counties are required to maintain a special investigation unit with staffing based on the size of their CalWORKs caseload. State regulations recommend that the counties have one sworn investigator for every 1,000 CalWORKs cases they handle, although as shown earlier in Table 2 on page 13, half of the six counties we visited have a higher ratio and half have a lower ratio. Although the counties appear to have similar criteria for investigations, their procedures for conducting investigations and their criteria for prosecution and imposing administrative sanctions vary. As we discuss in the Introduction, the special investigation unit can be located within the county welfare department, the district attorney’s office, or the sheriff’s office. The six counties we visited all classify their antifraud efforts as either early fraud activities or ongoing investigations. Although these six counties use sworn peace officers to conduct ongoing investigations, only Riverside and Sacramento counties use investigative staff who are not sworn peace officers to conduct early fraud activities. Riverside County told us that using non‑sworn investigative staff to handle early fraud referrals enables sworn investigators to focus on investigating potential ongoing fraud. Sacramento County told us that it uses non‑sworn staff to conduct early fraud investigations because these staff are more knowledgeable about eligibility issues and are less expensive than sworn peace officers. In addition, Sacramento County believes that welfare applicants tend to feel more relaxed and cooperative when investigators who are not peace officers conduct home visits. The other four counties we reviewed use sworn investigative staff to conduct all investigations, regardless of the type of referral. To explain why it uses sworn investigators only, Orange County stressed legal advantages, stating that the presence of sworn officers places more emphasis on the client’s need to be truthful and that the officers can testify to hearsay, make arrests, and take immediate action in abuse and neglect situations. California State Auditor Report 2009-101 51 November 2009 California law allows prosecution of welfare fraud as a felony when the potential monetary loss is at least $400. The counties have developed guidelines that set monetary thresholds below which the district attorney generally does not prosecute fraud. These thresholds vary among the counties we visited and can be as high as $10,000, depending on the type of offense. These variances may affect the number of cases referred and successfully prosecuted in each county, which are shown in Table 8. The counties we visited stressed that their thresholds are flexible and indicated that they refer cases for prosecution that fall below the threshold if warranted by the specific characteristics of the case. For example, two counties told us that they would investigate repeat offenders even if the monetary loss was below their prosecution thresholds. Table 8 Prosecution Activities Related to CalWORKs and Food Stamp Cases in 2008 BACKLOG OF NEW COUNTIES’ NET BACKLOG OF REFERRALS FOR REFERRALS FOR ACTIONS PROSECUTIONS ADJUSTMENTS REFERRALS FOR PROSECUTION NOT PROSECUTION TAKEN ON THE REFERRALS FILED AS A TO REFERRAL PROSECUTION NOT ACTED UPON AS OF RECEIVED REFERRALS FOR PROSECUTIONS NOT PERCENTAGE OF BACKLOG ACTED UPON AS OF ENTITY JANUARY 1, 2008 DURING 2008 PROSECUTION FILED PROSECUTED ACTIONS TAKEN DURING 2008* DECEMBER 31, 2008 Alameda 790 76 60 58 2 97% (13) 793 Los Angeles 1,589 811 895 769 126 86 (452) 1,053 Orange 111 175 170 140 30 82 ‑ 116 Riverside 48 351 348 332 16 95 ‑ 51 Sacramento 16 150 147 125 22 85 ‑ 19 San Diego 64 57 62 62 ‑ 100 (7) 66 Statewide 6,381 6,858 5,074 3,164 1,910 62 (2,566) 5,599 Sources: Investigation activity reports submitted by counties for 2008 to the Department of Social Services (Social Services). Note: As we discuss elsewhere in this report, the figures counties reported to Social Services regarding their investigation activities are of undetermined reliability. Because counties are to report referrals for prosecution by program, and many food stamp recipients also receive CalWORKs benefits, these figures double count cases for which recipients receive assistance under both programs. * When counties find that the previous month’s data was inaccurate, they report adjustments in the following month. The data reported by counties statewide show variances in the number of referrals for prosecution of CalWORKs and food stamp fraud and in the outcomes of the prosecutions filed, as shown in Table 8. As of January 1, 2008, there was a statewide backlog of 6,381 referrals, based on data that counties reported to Social Services, and 6,858 new cases were referred for prosecution during 2008. This amounted to 13,239 cases, of which the counties took action on 5,074 cases. Of these cases, the district attorneys chose to prosecute 3,164 (62 percent) and decided not to prosecute the remainder. As a result, the backlog of prosecution referrals decreased by about 12 percent during 2008, with 5,599 referrals for prosecution reported by counties as the backlog on December 31, 2008. 52 California State Auditor Report 2009-101 November 2009 Although 62 percent of the actions taken statewide were to file prosecutions, the six counties we reviewed filed prosecutions at a higher rate. Alameda, Riverside, and San Diego counties filed prosecutions for over 90 percent of the cases they acted on during 2008, while the other three counties prosecuted more than 80 percent. Despite these high rates of prosecution, the number of cases remaining in the referral backlog varies significantly by county. Alameda and Los Angeles counties reported high backlogs relative to the other counties we reviewed. Alameda County told us that the referral backlog includes referrals made to the district attorney investigative staff for investigations, and also stated that the two inspectors performing this work handle other public assistance fraud investigations. Such large prosecution backlogs create the risk that, by the time these counties prosecute these cases, the cases may be too old to achieve worthwhile results. Table 9 shows the prosecution outcomes that counties reported for cases during 2008.11 Of the cases referred for prosecution statewide, a successful outcome—a conviction or pleading of charges—was achieved 83 percent of the time. Most of these outcomes were convictions for welfare fraud. In only 12 percent of the prosecutions did the outcome result in a dismissal, indicating that generally once a case is prosecuted it will likely result in a conviction or pleading. Five of the six counties we reviewed were more successful in obtaining convictions or pleadings than the statewide average, with Alameda County recording the lowest success rate in obtaining convictions at 65 percent of its cases. Further, Alameda County, with 33 percent of the cases ending in dismissals, had more than twice the statewide average rate for dismissals, with most of the counties we reviewed ranging from 1 percent to 6 percent of their cases resulting in dismissals. Subsequent to receiving excerpts of our draft report on September 30, 2009, the Alameda County District Attorney’s Office (district attorney), sent us a letter dated October 12, 2009, disputing the number of dismissals that Alameda County’s welfare department reported. The district attorney believes that the number of dismissals should be 36 not 35; that only 23 of the 36 dismissals relate to the CalWORKs and food stamp programs; and that of the 23 dismissals only seven were granted on a motion by the prosecution while the remaining dismissals were the result of the normal operations of the judicial system. The district attorney believes that Alameda County’s reported data is inaccurate. Nevertheless, we display in Table 9 the prosecution data that the county welfare departments, including Alameda County, reported to Social Services in the investigation activity reports. 11 Because of the length of time needed for criminal prosecutions of CalWORKs and food stamp cases, the cases included in Table 8 are not necessarily the same as those included in Table 9. California State Auditor Report 2009-101 53 November 2009 Table 9 Prosecution Outcomes Related to CalWORKs and Food Stamp Fraud Cases During 2008 CASES IN WHICH CASES IN WHICH STATUTES OF PROSECUTION CASES RESULTED CHARGES WERE CASES DISMISSED LIMITATIONS CASES IN CONVICTIONS PLEADED BY JUDGE CASES DECLINED EXPIRED OTHER* CONCLUDED ENTITY DURING THE YEAR AMOUNT PERCENT AMOUNT PERCENT AMOUNT PERCENT AMOUNT PERCENT AMOUNT PERCENT AMOUNT PERCENT Alameda 106 69 65% 0 0% 35 33% 2 2% 0 0% 0 0% Los Angeles 849 747 88 0 0 102 12 0 0 0 0 0 0 Orange 126 53 42 60 48 7 5 0 0 0 0 6 5 Riverside 722 349 48 360 50 4 1 9 1 0 0 0 0 Sacramento 121 113 93 2 2 6 5 0 0 0 0 0 0 San Diego 102 98 96 0 0 4 4 0 0 0 0 0 0 Statewide 5,295 3,278 62 1,090 21 634 12 59 1 16 0 218 4 Sources: Investigation activity reports submitted by counties to the Department of Social Services (Social Services) for 2008. Note: As we discuss elsewhere in this report, the figures that counties report to Social Services regarding their investigation activities are of undetermined reliability. * Other includes such actions as administrative actions, disqualification consent agreements, and acquittals. The results from the counties we reviewed are especially notable when compared to the overall fraud referral caseload that these counties represent. The data suggest that other counties have higher percentages of dismissals and that the counties we reviewed, if their data are accurate, may use best practices that could be of benefit to other counties. As a result, it is in the best interest of Social Services to track these variances, as well as study the counties’ prosecution practices to determine whether other counties could become more effective in their efforts by emulating the successful prosecution practices used elsewhere. Once they are referred for prosecution, CalWORKs and food stamp fraud cases follow a standard process with few variations. If the case is worthy of prosecution as determined by a district attorney, the attorney typically files felony welfare fraud and perjury counts for all individuals involved in the case. Upon partial or full repayment, some district attorneys may reduce the charges to a misdemeanor welfare fraud count and may expunge the defendant’s record of the felony charges. State regulations require counties to conduct administrative disqualification hearings for CalWORKs and food stamp fraud cases for which the facts do not warrant prosecution or cases that have been referred for prosecution and subsequently declined. An impartial administrative law judge conducts these hearings. If the administrative law judge finds that a welfare recipient intentionally committed welfare fraud, the individual is barred 54 California State Auditor Report 2009-101 November 2009 from receiving welfare benefits for a specified period of time or permanently, depending on the nature of the violation. Of the counties we reviewed, only Sacramento County extensively used the Despite a requirement in state administrative disqualification hearing process during 2008. Despite regulations, of the counties we the requirement in state regulations, Social Services told us that reviewed, only Sacramento County many counties have stopped using the administrative disqualification extensively used the administrative hearing process, which it attributes to county investigative staff disqualification hearing process believing that the administrative disqualification hearing standard of during 2008. proof is higher than in criminal cases. In addition, Orange County told us that it discontinued the administrative disqualification hearings in July 2004 because it determined that, in the two and half years the program was in operation in the county, the cost outweighed the benefit by $57,000. Orange County also provided us a December 2005 letter from Social Services’ chief administrative law judge indicating that participation in the administrative disqualification hearing process was optional for CalWORKs. Social Services told us that it has convened a workgroup with the State’s presiding administrative law judge to discuss county concerns and clarify the appropriate application of the administrative hearing process. When the workgroup has completed its efforts, Social Services will issue a letter to counties that will explicitly address the administrative disqualification hearing requirements. Recommendations To ensure that counties are consistently following up on all match lists, Social Services should do the following: • Remind counties of their responsibility under state regulations to follow up diligently on all match lists. Further, it should work with counties to determine why poor follow‑up exists and address those reasons. • Revive its efforts to work with counties and federal agencies to address the counties’ concerns about match‑list formats and criteria. • Perform IEVS reviews of all counties regularly and better enforce the counties’ implementation of its recommendations to correct any findings and verify implementation of the corrective action plans submitted. Recognizing that the deterrence effect is difficult to measure, Social Services should develop a method that allows it to gauge the cost‑effectiveness of SFIS. Social Services should include in its efforts to measure cost‑effectiveness the administrative cost that counties incur for using SFIS. Based on its results, Social Services should determine whether the continued use of SFIS is justified. California State Auditor Report 2009-101 55 November 2009 To expedite the distribution of the $42.1 million in food stamp overpayment collections to the appropriate entities, Social Services should continue to work with the USDA and make its reconciliation of the backlog of overpayments a priority. Further, it should develop procedures to ensure that it promptly reconciles future overpayments. Additionally, Social Services should continue to monitor the counties’ collection reports to ensure that counties are reporting accurate information. Social Services should track how counties determine prosecution thresholds for welfare fraud cases and determine the effects of these thresholds on counties’ decisions to investigate potential fraud, with a focus on determining best practices and cost‑effective methods. It should then work with counties to implement the consistent use of these cost‑effective methods. Social Services should either ensure that counties follow state regulations regarding the use of administrative disqualification hearings or pursue changing the regulations. 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 the report. Respectfully submitted, ELAINE M. HOWLE, CPA State Auditor Date: November 3, 2009 Staff: John Baier, CPA, Audit Principal Kris D. Patel Christina Animo Nicholas D. Cline Meghann K. Leonard, MPPA Whitney M. Smith Maya Wallace, MPPA Benjamin Ward, CISA, ACDA Legal: Scott A. Baxter, JD For questions regarding the contents of this report, please contact Margarita Fernández, Chief of Public Affairs, at 916.445.0255. 56 California State Auditor Report 2009-101 November 2009 Blank page inserted for reproduction purposes only. California State Auditor Report 2009-101 57 November 2009 Appendix METHODOLOGY FOR COMPUTING THE COST‑EFFECTIVENESS OF FRAUD INVESTIGATION EFFORTS To calculate the cost‑effectiveness of both early fraud detection activities and ongoing fraud investigations, which we show in Table 3 on page 22, we applied the following formulas at the statewide level and for the six counties we visited. Specifically, we calculated the measurable costs and savings of these efforts during 2008 with respect to the California Work Opportunities and Responsibility to Kids (CalWORKs) program and the federal Supplemental Nutrition Assistance Program, known as the food stamp program in California. The following methodology describes our formula and sources of information, using our calculation of the cost‑effectiveness of Los Angeles County’s early fraud detection activities (early fraud activities) and ongoing investigations for CalWORKs as an illustrative example. Cost‑Effectiveness Formulas Early Fraud Activities (Monthly welfare benefits saved as a result of early fraud activities × Projected period of savings) ÷ Cost of performing early fraud activities = Savings achieved for every $1 spent. Ongoing Investigations [(Monthly welfare benefits saved as a result of ongoing investigations × Projected period of savings) + Overpayments identified] ÷ Cost of performing ongoing investigations = Savings achieved for every $1 spent. Monthly Welfare Benefits Saved as a Result of Early Fraud Activities and Ongoing Investigations The denial, discontinuance, and reduction of aid payments due to investigative efforts result in a savings to the State of California (State) by avoiding future aid payments for potentially fraudulent activities. Counties submit to the Department of Social Services (Social Services) a monthly report indicating the number of cases for which they denied, discontinued, or reduced aid due to early fraud activities, ongoing investigations, and certain matches identified by the State for the CalWORKs and food stamp 58 California State Auditor Report 2009-101 November 2009 programs. Counties also report to Social Services the number of CalWORKs and food stamp cases receiving aid and the total amount of aid, which allowed us to calculate a 2008 statewide average monthly amount of aid per case of $538.40 and $281.97 for the CalWORKs and food stamp programs, respectively. Because the counties avoided the entire aid payment when denying or discontinuing a case, we calculated the savings resulting from these actions by multiplying the number of cases for which aid was denied or discontinued by the average 2008 monthly grant amount. To determine the savings due to reductions in aid, we multiplied the number of cases for which the county reduced aid due to investigative efforts by the 2008 average amount of reductions for all cases—$174.89 for CalWORKs and $65.55 for food stamp programs—in Los Angeles County’s welfare database. The sum of these savings represents the total savings resulting from early fraud activities and ongoing investigations. Three counties provided their average reduction per case during 2008. Two counties’ figures were reasonably comparable to Los Angeles County’s average reductions. However, Orange County told us that it had an average reduction of $422 for CalWORKs and $132 for food stamp programs. If we use Orange County’s reduction amounts, its cost‑benefit would be higher. However, even with using Los Angeles County’s reduction amounts, Orange County’s savings are still the highest or second highest among the six counties. Nevertheless, to be consistent, we used Los Angeles County’s average reductions in our computation because it represents the largest caseload in the State for both the CalWORKs and Food Stamp programs. Projected Period of Savings Forming the basis of the three‑month estimate—our most conservative estimate of the savings—is the fact that recipients of both the CalWORKs and food stamp programs are required to report quarterly any changes in their eligibility, such as increased income or a child leaving the home. We also used an 18‑month projection because Social Services asserted that its ongoing analysis of historical eligibility data for CalWORKs recipients indicates that they receive aid for an average of 18 months. Although Social Services also determined that food stamp recipients receive aid for an average of 31 months, we used the shorter period in our analysis to maintain consistency between our cost‑effectiveness results for the two aid programs. California State Auditor Report 2009-101 59 November 2009 Overpayments Identified Through their ongoing investigations, counties identify overpayments to recipients because of welfare fraud. Although counties might identify such overpayments through early fraud activities, they indicated that most, if not all, of the overpayments are identified through ongoing investigations. Therefore, we include any overpayments counties reported as a one‑time savings attributable to ongoing investigations. Cost of Performing Early Fraud Activities and Ongoing Investigations The counties’ investigative costs are largely reimbursed with federal and state funds. Counties submit quarterly expense claims to Social Services for reimbursement of administrative costs. These expense claims separately identify the costs of early fraud activities and ongoing investigations for both CalWORKs and food stamp programs, by using different codes. These costs generally include personnel costs and a related share of administrative costs associated with performing early fraud and ongoing investigations. Cost‑Effectiveness Calculations for Los Angeles County Los Angeles County reported that during 2008 it denied or discontinued 2,329 CalWORKs cases and reduced aid in 1,219 CalWORKs cases through its early fraud activities. We calculated that these actions resulted in a monthly savings of $1,467,125 for early fraud, using the method described previously for early fraud. Further, Los Angeles County reported that during 2008 it denied or discontinued 13 CalWORKs cases and reduced aid in 21 CalWORKs cases through ongoing investigations. We again calculated the resulting monthly savings to be $10,672, using the method described earlier. In addition, Los Angeles County identified $7,778,234 in CalWORKs aid overpayments during 2008. In its expense claims submitted to Social Services, the county claimed $7,211,996 for costs related to CalWORKs early fraud activities and $10,130,254 for costs related to CalWORKs ongoing investigations. Using the cost‑effectiveness formulas noted earlier, we calculated the cost‑effectiveness of Los Angeles County’s early fraud activities and ongoing investigations related to CalWORKs cases using the three‑month projection as shown on page 60, to arrive at the results included in Table 3 on page 22. 60 California State Auditor Report 2009-101 November 2009 Cost‑Effectiveness of Early Fraud Activities for CalWORKs According to a Three‑Month Projection ($1,467,125 savings × 3 months) ÷ $7,211,996 costs = $0.61 in savings for every $1 spent Cost‑Effectiveness of Ongoing Investigations for CalWORKs According to a Three‑Month Projection [($10,672 savings × 3 months) + $7,778,234 overpayments identified] ÷ $10,130,254 costs = $0.77 in savings for every $1 spent California State Auditor Report 2009-101 61 November 2009 (Agency response provided as text only.) Department of Social Services 744 P Street Sacramento, CA 95814 October 13, 2009 Ms. Elaine M. Howle* State Auditor 555 Capitol Mall, Suite #300 Sacramento, CA 95814 Dear Ms. Howle: SUBJECT: BUREAU OF STATE AUDITS REPORT: DEPARTMENT OF SOCIAL SERVICES: It Lacks Assessments of Cost Effectiveness and Is Missing Opportunities to Improve Counties’ Antifraud Efforts in the CalWORKS and Food Stamp Programs (Report# 2009-101) The California Department of Social Services (CDSS) appreciates the opportunity to respond to the draft Bureau of State Audits (BSA) report entitled Department of Social Services: It Lacks Assessments of Cost‑Effectiveness and Is Missing Opportunities to Improve Counties’ Antifraud Efforts in the CalWORKS and Food Stamp Programs. The enclosed response addresses the CDSS concerns and efforts to implement the recommendations. Further, the CDSS will endeavor to work with the BSA auditors on a cost-effective fraud deterrence methodology. If you have any questions concerning the enclosed CDSS response, please contact me at (916) 657-2598 or Karen Ruiz, Deputy Director, Information Systems Division, at (916) 654-1039. Sincerely, (Signed by: John A. Wagner) JOHN A. WAGNER Director Enclosure * California State Auditor’s comments begin on page 75. 62 California State Auditor Report 2009-101 November 2009 CDSS Response to the BSA Draft Report 2009-101 Recommendations for Social Services: Introduction‑Background BSA Statement: “Federal and state laws require counties to maintain a special investigation unit to investigate potential welfare fraud and refer substantiated fraud either for prosecution or for administrative settlement. Social Services recommends that counties have one sworn peace officer investigator for every 1,000 active CalWORKs cases.” Response: Social Services acknowledges that the counties are not consistent in maintaining the required ratio of staff in Special Investigations Units (SIUs). All County Information Notice I-18-09, dated April 7, 2009, was issued to remind the counties that state regulations that (Manual of Policies and Procedures (MPP) 1 Section 20-007.113) require a ratio of one sworn peace officer for every 1000 active CalWORKs cases applies to both the CalWORKs and CalWORKs Stage One Child Care caseloads. It further reminds counties that in 7 CFR 272.4(g), the Food Stamp Program is required to operate a fraud detection unit in areas where 5000 or more households participate in the program. While MPP 20-007.113 does not specifically address the federal regulation, counties also need to consider this workload when determining appropriate staffing levels for their SIUs. Chapter #1 The Cost Effectiveness of Counties’ Efforts to Combat Fraud Varies, Though the Data for Such Computations are Questionable Sub Chapter #4: Social Services and the Counties Have Not Determined Whether Antifraud Efforts Are Cost‑Effective BSA Statement: “Although Social Services has developed a formula to calculate the savings that counties realize as a result of their antifraud efforts— in terms of fraudulent aid not paid and administrative savings—Social Services doesn’t use this formula to evaluate the cost effectiveness of counties’ antifraud efforts.” Response: In response to a legislative request in 2006, Social Services provided information regarding CalWORKs Fraud Costs and Collections/Savings for Fiscal Year (FY) 2003-04 (this information was provided to BSA on September 8, 2009). The estimated savings represents avoided grant/administrative costs associated with early and ongoing fraud activities that resulted in benefit reductions, denials and discontinuances. The estimate assumes that absent the program integrity efforts, cases would have otherwise come on to or stayed on aid for a period of time. The fraud costs represent administrative costs as reported by the counties for activities related to early and ongoing fraud activities as identified by specific CalWORKs Program Codes (description of codes provided to BSA on October 6, 2009). Social Services is in the process of updating this information with FY 2007-08 data and will provide the analysis to the BSA as soon as possible. Other Comment The BSA has developed their own cost effectiveness methodology of comparing the savings resulting from efforts to combat welfare fraud to the counties’ costs to perform investigation activities. Social Services has not had time to fully evaluate the BSA methodology. However, upon preliminary review it appears that the Draft BSA Report 2009-101 1 10/13/09 California State Auditor Report 2009-101 63 November 2009 CDSS Response to the BSA Draft Report 2009-101 BSA’s conservative estimate of assuming savings for only a three-month period is inappropriate. 2 The average length of stay on the CalWORKs program is approximately 18 months. Although some cases may have been on aid for some number of months prior to the fraud determination (in the case of discontinuances and benefit reductions), it seems reasonable to assume they would have otherwise stayed on for more than just an additional three months. The BSA indicates that their three-month assumption is based on the fact that recipients of both the CalWORKs and food stamp programs are required to report quarterly any changes in their eligibility. However, the information reported on the quarterly report does not necessarily have a direct relationship to the finding of fraud and if something fraudulent was reported it may not be identified immediately (i.e., may be reported for several quarters before action taken). Sub Chapter #7: Social Services’ Procedures for Reviewing Counties’ Investigation Activity Reports Are Inadequate BSA Statement: “When creating this report, Social Services does not follow up on discrepancies, such as potential underreporting of activity by counties. For example, Los Angeles County – representing 30 percent of the State’s CalWORKs caseload – is the largest county in the state, with the next largest county having just over 7 percent of the State’s caseload. Given that its caseload is more than four times larger than that of the next largest county, it is reasonable to assume that Los Angeles County would report the highest number of cases in most categories on the investigation report. However, our review of the fiscal year 2007‑08 data found that this did not occur . . . . For example, although the counties reported reducing a total of nearly 5,000 cases during fiscal year 2007‑08 as a result of ongoing investigations, only 41 of those were reported by Los Angeles County, a number that seems quite low considering that the county spent over $23 million on ongoing investigations during 2008 – the highest by far among the counties we reviewed.” Response: Generalizing an expectation that activity numbers will be relative to caseload counts is an unproven 3 assumption. This is especially true when the measurement is based on a very complex report containing 382 cells, reflecting activity in a program area that historically has large fluctuations depending on applicant and recipient characteristics, county staffing, and changing investigative emphasis within counties. The Los Angeles County example cited by the audit to support the contention is a good example of the difficulties. The 41 cases of benefit reduction the audit mentions were for the ‘non early fraud’ subcategory. When the 3,502 cases in the ‘early fraud’ subcategory are counted, Los Angeles County has the highest number of benefit reductions for the combined subcategories. 4 Social Services is very interested in improving its data review approach, but unfortunately, the audit does not 5 offer specific suggestions in this area beyond the belief that Los Angeles should usually report the highest number in each category. Although Social Services staffing limitations preclude more on site review than that provided by the IEVS reviews, we are open to concrete suggestions for items meriting follow up. Draft BSA Report 2009-101 2 10/13/09 64 California State Auditor Report 2009-101 November 2009 CDSS Response to the BSA Draft Report 2009-101 Sub Chapter #8: Errors in the Counties Investigation Reports Are Passed On to Other Parties BSA statement: 6 “Further, Social Services uses this report for special studies, such as a 10 year study of welfare fraud trends that it completed in 2006. The intent of the study was to assess trends and identify best practices related to early and ongoing welfare fraud investigations, overpayment collections, and county administrative practices.” Response: Social Services did not use the fraud investigation quarterly report for the 10-Year Study because there were several revisions made to the monthly report after the quarterly report was published. Social Services uses current data provided from the monthly reports to reflect the most current information available during that time for special requests. Recommendation #1: “To ensure that all counties consistently gauge the cost-effectiveness of their early fraud activities and ongoing investigation efforts for the CalWORKs and food stamp programs, Social Services should work with the counties to develop a formula to perform a cost-effective analysis using information that the counties currently submit.” Response: Social Services has been working with its internal stakeholders to develop a formula to evaluate the cost effectiveness of county fraud operations. We hope to have this formula completed soon, as resources permit. Additionally, Social Services has established a workgroup, with county participation, to improve the accuracy of the data collected in the Investigative Activity Report (DPA-266). This data would be used in conducting any cost-effective analysis of the counties’ fraud operations. However, due to limited state and county resources, it may take quite some time to complete this process. Recommendation #2: “To ensure that counties are getting the most benefit from the resources they spend on antifraud efforts related to CalWORKs and food stamp cases, Social Services should do the following:” • “Using the results from this cost-effectiveness analysis, determine why some counties’ efforts to combat welfare fraud are more cost-effective than others.” Response: Social Services conducted an analysis of the cost-effectiveness of county fraud operations in 2006, when it completed a study of ten years of county data. Currently, the Fraud Bureau needs more resources to continue these efforts. As described above, Social Services believes the focus should be first to ensure the accuracy of the counties’ report data before developing a formula for determining the cost-effectiveness of these operations. Social Services continues to work on this effort, as resources permit. • “Seek to replicate the most cost-effective practices among all counties.” Draft BSA Report 2009-101 3 10/13/09 California State Auditor Report 2009-101 65 November 2009 CDSS Response to the BSA Draft Report 2009-101 Response: One thing that Social Services learned from its work with the Program Integrity Steering Committee (PISC) is that what might be a best practice in one county may not work in another county for a variety of reasons. 7 Social Services has already shared statewide potential “promising approaches” that were developed by the PISC Peer Review Team. • “Continue to work on the recommendations of the steering committee that it is already addressing and more promptly act on the remaining recommendations.” Response: Social Services is continuing to work on the remaining five recommendations of the PISC, as resources permit. Recommendation #3: “To ensure the accuracy and consistency of the information on welfare fraud activity that counties report and that Social Services is subsequently reports to the federal government, the Legislature, and internal users, Social Services should:” • “Remind counties that they are responsible for reviewing the accuracy and consistency of the investigation activity report before submission.” Response: Social Services has established a workgroup to clarify the instructions and data that are to be entered on the DPA-266. The efforts of this workgroup will continue, as state and county resources become available. Once the report has been revised, CDSS will provide technical assistance to the counties on how to complete the report accurately. • “Perform a more diligent review of the counties’ investigation activity reports to verify the accuracy of the information submitted.” Response: Social Services reviews these reports during its IEVS Review process. Inaccuracies in the reports are discussed 8 with county staff during the county review. • “Provide counties feedback on how to correct and prevent errors that it detects during this review.” Response: During county IEVS Reviews, Social Services staff provide county staff directions on how to correct 8 inaccuracies in the DAP-266 report. • “Continue with its committee’s efforts to clarify instructions to counties for completing the investigation activity report.” Draft BSA Report 2009-101 4 10/13/09 66 California State Auditor Report 2009-101 November 2009 CDSS Response to the BSA Draft Report 2009-101 Response: Social Services continues its workgroup efforts. However, the limited state and county resources make the outcome of this workgroup a long-term goal. Chapter #2: The Department of Social Services and the Counties Could Improve Their Ongoing Efforts to Combat Welfare Fraud Subchapter #2: Counties Assert That the Format of Some Match Lists Could Be Improved BSA statement: “Additionally, seven of the 10 match lists are provided to the counties as paper copies, including three that have a 45‑day timeline. Three lists‑the payment verification system, new hire registry, and integrated earning clearance fraud detection match lists – are sent in an electronic format. Social Services indicated that four of the remaining lists are provided to some counties in electronic format depending on the county’s welfare database system. All five counties we visited told us that having all match lists in electronic form would allow them to process matches more efficiently.” Response: Social Services is moving towards paperless transmission of matches to counties. The Payment Verification System (PVS), Integrated Earnings Clearance Fraud Detection (IFD), and New Hire Registry (NHR) matches, which used to represent the bulk of paper matches sent to counties, are now sent electronically. The PVS match converted to paperless transmission in May 2007, and the IFD and NHR matches started paperless transmission in June 2009. Although Social Services has limited resources, other matches will convert to paperless transmission when feasible. Subchapter #3: Social Services Monitoring of Counties’ Follow‑Up Efforts Is Weak BSA statement: “Although Social Services has a process in place to monitor the counties’ efforts to follow up on match lists, it is missing opportunities to improve their efforts because it does not visit all counties on a regular basis and does not always enforce recommendations from the reviews it does perform. . . Specifically, it has not reviewed 26 of the 58 counties during the past three years.” Response: 9 Social Services has not physically reviewed 23 of the 58 counties in the last three years due to limited staff 10 and resources. However, self-assessment surveys were sent out to all 58 counties in 2006-2007. Social Services received prior approval from FNS to use this alternative review process. In 2008 and 2009, Social Services conducted physical reviews in more counties, and, in just the last year, have added detailed desk audits as an alternative. Draft BSA Report 2009-101 5 10/13/09 California State Auditor Report 2009-101 67 November 2009 CDSS Response to the BSA Draft Report 2009-101 In 2006-2007, Social Services sent self-assessment surveys to all 58 counties and conducted physical IEVS reviews on the following 9 counties: Del Norte Placer Humboldt Shasta Kern Sutter Orange Yolo Yuba In 2007-08, Social Services conducted physical IEVS reviews on 6 counties: Fresno Nevada Orange Riverside Santa Clara Ventura In 2008-09, Social Services conducted IEVS reviews on 27 counties: 11 Napa El Dorado Solano Yuba San Joaquin Ventura Tulare Santa Barbara San Bernardino Marin Sacramento San Diego Merced San Francisco Colusa (Desk Review) Sonoma Del Norte Kern (Desk Review) Lassen (Desk Review) Butte Kings Stanislaus Orange Madera San Mateo Alameda Contra Costa In total, Social Services completed IEVS reviews on 35 counties during the last three years, in addition to quarterly visits to Los Angeles County to follow up on a focused IEVS review of notices of action and collections issues. BSA statement: “Among the counties not reviewed is Los Angeles, which represents approximately 30 percent of the State’s California Work Opportunities and Responsibility to Kids (CalWORKs) cases.” Response: The 1995 IVES review of Los Angeles County indicated serious problems and backlogs with the Los Angeles Department of Public Social Services (LA-DPSS) processing of overpayments. Based on a review of LA County reports, which indicated a significant decrease in collection activity, Social Services conducted a focused IEVS review in 2007. Since that review, Social Services management and staff have been working 12 and meeting regularly with senior executive and management staff of LA-DPSS to resolve these issues. Since July 2008, Social Services has required a formal quarterly report from LA-DPSS detailing their progress in this endeavor. LA-DPSS has made significant progress: - In early 2008, they had a backlog of 92,752 closed food stamp claims needing resolution. Their most recent quarterly report indicates that 73,120 of these claims have been resolved to date. - In early 2008, they had a backlog of 15,149 open claims that required adequate notices of action be sent out. Their most recent quarterly report indicates 11,618 of these claims have been resolved to date. Draft BSA Report 2009-101 6 10/13/09 68 California State Auditor Report 2009-101 November 2009 CDSS Response to the BSA Draft Report 2009-101 During the past two years, Social Services staff and management have met with LA-DPSS on seven occasions and will be meeting with them again on October 27, 2009. LA-DPSS has dedicated over 80 of their staff to this effort and have had to train over 4,000 of their staff in the handling of payments to remedy these problems. Both LA-DPSS and Social Services believe that it is more appropriate to resolve these issues before implementing a new full IEVS review. BSA statement: “Social Services’ review log dates back to 1995 and shows that . . . five small counties –Trinity, Modoc, Inyo, Mono, and Alpine – have never been reviewed.” Response: With USDA concurrence, Social Services has conducted self-assessment surveys of many smaller counties due to staffing shortages and where it has not been economically feasible to physically travel to perform a site review. Recently, a detailed desk review was completed on Lassen County, and a desk review will soon be completed on Alpine County. The remaining counties—Inyo, Modoc, Mono and Trinity—will be reviewed either in-person or via desk review during FY 2009-10. Subchapter #4: Social Services Has Not Done a Cost‑Benefit Analysis of the Statewide Fingerprint Imaging System BSA statement: “Although it believes that SFIS plays an important role in deterring fraud, Social Services has not done a cost‑benefit analysis of SFIS, because it believes there is no way to measure the deterrence effect of the system.” Response: Social Services acknowledges that it has not performed a new cost-benefit analysis of SFIS since it first used 13 Los Angeles County’s AFFIRM data results, but disagrees with the BSA conclusion that such an analysis is necessary to justify the ongoing value of SFIS. BSA acknowledges the difficulty in accurately measuring the deterrence effect of SFIS and notes that, of the five states using systems such as SFIS, only Arizona measures the benefits in strict monetary terms. As noted in the report, Social Services has established the deterrence value of SFIS in 1997 (based on data 14 from Los Angeles’ AFIRM project). Additionally, in 2005, CDSS conducted its own study of the effectiveness of the IEVS system in identifying duplicate aid in comparison to SFIS. Additionally, in 2009, Social Services compared the capability of the county file clearance process in identifying potential duplicate aid to SFIS. In both the 2005 and 2009 studies, Social Services found that SFIS identified fraudulent duplicate aid that was not detected by either IEVS or the file clearance processes. The BSA noted that Arizona reports annual deterrent savings of $10 million. California’s combined CalWORKs 15 (TANF) / Food Stamps (SNAP) caseload is three times the size of Arizona’s caseload. Accordingly, it is reasonable to estimate that the deterrent value of California’s SFIS is at least $10 million annually which, when weighed against its current $5 million annual maintenance and operations (M&O) cost, yields at least a $2 savings for each dollar spent. Draft BSA Report 2009-101 7 10/13/09 California State Auditor Report 2009-101 69 November 2009 CDSS Response to the BSA Draft Report 2009-101 The $5 million cost applies only to the current fiscal year. The federal government began sharing the project 16 costs on September 1, 2009, with the execution of a new 8-year contract. Therefore, the annual cost to the state during the remaining seven years on the contract will be $4 million. Social Services believes that the studies it has conducted, coupled with the information available from other states, more than justifies the deterrent value of SFIS. Further, Social Services believes it would not 17 be beneficial to spend additional limited state resources to collect and analyze the requisite data for an independent cost-benefit analysis of SFIS. Social Services acknowledges that, in some counties, a significant backlog of SFIS match results await resolution. Social Services will issue an All County Information Notice to remind the counties of their responsibility to complete these matches. Social Services has developed a SFIS Best Practices Handbook that includes guidance to the counties on how to resolve SFIS results and the importance of working the Resolution Queues in a timely manner. The release of the SFIS Best Practices Handbook is pending final review and comment. Along with the SFIS Best Practices Handbook, Social Services will be issuing a County Assessment Report to each county. It provides information on how to improve photo and fingerprint image quality and stresses the importance of processing the resolution queues in a timely manner. Finally, Social Services offers a SFIS training program to counties, which includes training tools and materials that counties can use and web-based training classes. The information can be found at www.sfis.ca.gov/training_page.html. Counties can enroll in classes directly on the SFIS website. Subchapter #5: Social Services Has Not Taken the Necessary Steps to Claim Its Share of $42.1 Million in Food Stamp Overpayment Collections BSA statement: “However, Social Services has not taken the steps needed to claim its share of these overpayments or to distribute the funds due to the administering federal agency, the USDA.” Response: Social Services has taken the steps needed to claim its share of these overpayments or to distribute the 18 funds due to the USDA-FNS. Social Services has been working closely with USDA-FNS since March 2008 to complete Tax Offset Program (TOP) reconciliation, and for FNS to provide approval before any offsets can be processed. Social Services has provided all back-up documentation and awaits FNS’ completion of their Letter of Credit (LOC) Adjustment in the Automated Standard Application for Payments (ASAP) before Social Services can process the offsets against the counties. The counties report the collections to Social Services quarterly on a USDA Status of Claims Against Households report (FNS 209). However, counties directly input their collection data into the Statewide Automated Reconciliation System (SARS). SARS automatically calculates the Federal, State, and County Draft BSA Report 2009-101 8 10/13/09 70 California State Auditor Report 2009-101 November 2009 CDSS Response to the BSA Draft Report 2009-101 shares based on the information input into the SARS by the counties. After federal tax refund intercepts are reconciled, Social Services reduces the adjusted total of collections by the amount retained. Additionally, line 24, not the remainder of funds as noted in the report, of the FNS 209 is split evenly between State 19 and County. 18 Social Services is not solely responsible for holding up on the overpayment collection reconciliation. The reconciliation process is a joint effort between USDA and Social Services to reconcile the overpayment collection on a quarterly basis. Due to the staff turnover and lack of process documentation in both Social Services and FNS, the reconciliation from the FY 2003-04 to current FY overpayment collection have not been completed timely since January 2004. The reconciliation process was reinitiated by FNS effective March, 2008. Social Services has been working with FNS since that time. At that time the outstanding overpayment periods were March 2004 quarter through December 2007 quarter. Starting in March 2008, Social Services has reconciled some FY 2006-07 and 2007-08 overpayment collection with FNS. However, due to FNS workload priorities, they have asked Social Services via email to hold on subsequent reconciliations until they have cleared some of the quarter reconciliations that were previously sent from FRU. Social Services has provided FNS with all documents needed to complete the reconciliations. Social Services has to obtain FNS’s approval of the collection amount prior to recouping from the counties. Social Services cannot process the adjustments without USDA approval. Social Services has not delayed addressing this issue, the USDA, counties, and Social Services have not had access to these funds. This is a joint effort with FNS to complete the reconciliations in order for Social Services to process the offsets. However, Social Services was not aware counties have earned interest when they deposited the overpayment collection into their bank accounts until this audit occurred. Per 45 CFR 92.21 and the Cash Management Improvement Act (CMIA), Social Services has authority to require the counties to remit the interest earned to the Federal Agency and the State Agency. 18 Social Services disagrees with the statement that Social Services is solely responsible for holding up on the overpayment collection reconciliation. The reconciliation process is a joint effort between USDA and Social Services to reconcile the overpayment collection on a quarterly basis. Due to the staff turnover and lack of process documentation in both CDSS and FNS, the reconciliation from the FY 2003-04 to current FY overpayment collection have not been completed timely since January 2004. BSA Statement: “Social Services has had several problems that delayed its efforts to distribute the $42.1 million in overpayments that have accumulated.” Response: The reconciliation process was reinitiated by FNS effective March, 2008. Social Services has been working with FNS since that time. At that time the outstanding overpayment periods were March 2004 quarter through December 2007 quarter. Starting in March 2008, Social Services has reconciled some FY 2006-07 and 2007-08 overpayment collection with FNS. However, due to FNS workload priorities, they have asked Social Services via email to hold on subsequent reconciliations until they have cleared some of the quarter reconciliations that were previously sent from FRU. Social Services has provided FNS with all documents needed to complete the reconciliations. Social Services has to obtain FNS’s approval of the collection amount prior to recouping from the counties. Social Services cannot process the adjustments without USDA approval. Draft BSA Report 2009-101 9 10/13/09 California State Auditor Report 2009-101 71 November 2009 CDSS Response to the BSA Draft Report 2009-101 BSA Statement: “Because Social Services has delayed addressing this issue, the USDA, counties, and Social Services have not had access to these funds.” Response: Social Services disagrees with the statement that Social Services has delayed addressing this issue, the USDA, counties, and Social Services have not had access to these funds. This is a joint effort with FNS to complete 18 the reconciliations in order for Social Services to process the offsets. However, Social Services was not aware counties have earned interest when they deposited the overpayment collection into their bank accounts until this audit incurred. Per 45 CFR 92.21 and the Cash Management Improvement Act (CMIA), Social Services has authority to require the counties to remit the interest earned to the Federal Agency and the State Agency. Subchapter #6: Investigation and Prosecution Efforts Vary by County BSA Statement: “Counties are required to maintain a special investigation unit with staffing based on the size of their food stamp and CalWORKs caseload. State regulations recommend the counties to have one sworn investigator for every 1,000 CalWORKs cases it handles.” Response: See Social Services’ prior response in the Introduction-Background section. Recommendation #1: “To ensure that counties are consistently following up on all match lists, Social Services should:” • “Remind counties of their responsibility under federal and state regulations to diligently follow up on all match lists. Further, it should work with counties to determine why poor follow‑up exists and address those reasons.” Response: Social Services agrees with this recommendation and will issue an All County Information Notice to remind counties of their obligation. • “Revive its efforts to work with counties and federal agencies to address the counties’ concerns regarding match list formats and criteria.” Response: Social Services will consider this recommendation as resources permit. • “Perform IEVS reviews of all counties regularly and better enforce the counties’ implementation of its recommendations to correct any findings and verify implementation of the corrective action plans submitted.” Draft BSA Report 2009-101 10 10/13/09 72 California State Auditor Report 2009-101 November 2009 CDSS Response to the BSA Draft Report 2009-101 Response: Social Services agrees with this recommendation and has redirected staff to fully staff the IEVS Review team. Social Services plans to conduct regular reviews on a three-year cycle with regular follow ups on county corrective actions. Recommendation #2: “Recognizing that the deterrence effect is difficult to measure, Social Services should develop a method that allows it to measure the cost‑effectiveness of SFIS. Social Services should include in its efforts to measure cost‑effectiveness the administrative cost that counties incur for using SFIS. Based on its results, Social Services should determine whether the continued use of SFIS is justified.” Response: 20 As BSA describes, it is impossible to accurately measure the deterrence effect of SFIS. As previously stated to BSA, Social Services is at a loss for how to develop a reasonable cost-effectiveness measurement without this critical component. However, a new independent cost-benefit analysis would not be beneficial, 13 for the reasons previously stated in Chapter #2, Subchapter #4. Recommendation #3: “To expedite the distribution of the $42.1 million in food stamp overpayment collections to the appropriate entities, Social Services should make its reconciliation of the backlog of overpayments a priority. Further, it should develop procedures to ensure that it promptly reconciles future overpayments. Additionally, Social Services should continue to monitor the counties’ collection reports to ensure that they are reporting accurate information.” Response: Social Services has been in regular communication with USDA on developing a process to review the accuracy of the FNS-209 collection report data. With the concurrence of USDA, Social Services staff will be focusing this upcoming year on county FNS-209 validation report issues and corrective action statuses that Social Services and USDA have identified. Recommendation #4: “Social Services should track how counties determine prosecution thresholds and determine the effects of these thresholds on counties’ decisions to investigate potential fraud, with a focus on determining best practices and cost‑effective methods. It should then work with counties to implement the consistent use of the more effective methods.” Response: See Chapter #1, Subchapter 7 and 8 responses. Recommendation #5: “Social Services should either ensure that counties follow state regulations regarding the use of administrative disqualification hearings or pursue changing the regulations.” Draft BSA Report 2009-101 11 10/13/09 California State Auditor Report 2009-101 73 November 2009 CDSS Response to the BSA Draft Report 2009-101 Response: Social Services has established a workgroup to look at making the administrative disqualification hearing (ADH) process work more smoothly. The efforts of this workgroup continue as both state and county resources permit. Additionally, Social Services is now completing for final departmental review an All County Information Notice that details the county responsibilities for both the Food Stamp and CalWORKs ADH processes. Draft BSA Report 2009-101 12 10/13/09 74 California State Auditor Report 2009-101 November 2009 Blank page inserted for reproduction purposes only. California State Auditor Report 2009-101 75 November 2009 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 to our audit from the Department of Social Services (Social Services). The numbers below correspond to the numbers we placed in the margins of Social Services’ response. Our legal counsel believes that Social Services misstates that the 1 state regulation contained in Section 20‑007.113 of the Manual of Policies and Procedures requires counties to meet this ratio. The regulation reads “A ratio of at least one investigator for every 1,000 California Work Opportunities and Responsibility to Kids (CalWORKs) program cases or major fraction thereof is recommended,” which our legal counsel believes should be interpreted as a suggested but not a mandatory ratio. We disagree; we believe using more than one projection period is 2 appropriate to provide differing views on the potential savings. As we describe on page 21, we used the three‑month projection as a conservative estimate of the savings, due to the fact that recipients of both the CalWORKs program and the federal Supplemental Nutrition Assistance Program, known as the food stamp program in California, are required to report quarterly any changes in their eligibility, such as increased income or a child leaving the home. These changes, when reported quarterly, should result in prompt changes to aid payments. We also used an 18‑month projection based on Social Services’ analysis. It is surprising to us that Social Services now expresses concerns with the three‑month projection of savings. Over a two‑month period, we met formally with Social Services on five occasions, as well as made numerous telephone and e‑mail contacts, to share our methodology and drafts of cost‑effectiveness calculations. Despite what we believe was ample time to review our methodology, at no time did Social Services’ staff express this concern. Social Services misses our point. We highlight Los Angeles County’s 3 underreporting of the number of ongoing investigations as an example of a glaring error that Social Services could have readily identified if it had more thoroughly reviewed the county’s data for reasonableness. As we note on page 32, given its size, Los Angeles County reported very few results for its ongoing investigations compared to other counties. We are not suggesting that Los Angeles County should always report high numbers of results in all categories on the investigation activity report. However, it is reasonable for Social Services to question why Los Angeles County reported such low numbers, especially given that its CalWORKs 76 California State Auditor Report 2009-101 November 2009 caseload is more than four times larger than the next largest county and that it spent over $23 million performing ongoing investigations during 2008. In fact, as we note on pages 2 and 31, Los Angeles County acknowledged to us that it significantly underreported the number of ongoing investigations it performed but only realized this problem after we questioned the numbers. Had Social Services better scrutinized its data, it may have discovered this discrepancy. Simple follow‑up on such discrepancies would not only assist Social Services in identifying reporting errors, but it would also assist in evaluating the performance of counties’ antifraud efforts. Moreover, as we note on pages 29 and 34, Social Services has long been aware of errors in counties’ reporting, but it continues to compile and pass on this information to stakeholders without subjecting the data to more than a cursory review. 4 Social Services is incorrect in stating that “Los Angeles County has the highest number of benefit reductions for the combined [early fraud and ongoing investigations] subcategories.” According to Social Services’ compilation of county data for fiscal year 2007–08, Orange County reported 1,361 cases more than Los Angeles County for the total of these two subcategories. Specifically, Los Angeles County reported a total of 3,543 cases with benefits reduced (3,502 cases for early fraud and 41 cases for ongoing investigations), while Orange County reported 4,904 cases for these two subcategories (3,341 and 1,563 cases), or 1,361 cases more. 5 It is disappointing that Social Services believes we did not offer any concrete suggestions for improving its review procedures. We clearly indicate that a more thorough review of counties’ reported data is a starting place. This approach coupled with its knowledge of counties’ practices could dramatically improve the review process. 6 Social Services misunderstood which report we were referring to. Thus, we clarified the text on page 35 to read “Further, Social Services uses the investigation activity report for special studies, such as . . . ”. 7 While we might agree that a certain best practice used by one county may not be suitable for use by all counties, we still believe it would be beneficial for Social Services to identify and share with other counties those practices that have proven to be cost‑effective. 8 We do not believe that the periodic Income and Eligibility Verification System (IEVS) reviews are a sufficient process for validating the accuracy of counties’ investigation activity reports without also performing other activities such as following up on obvious discrepancies. We acknowledge on page 33 that Social Services reviews the investigation activity report during the IEVS California State Auditor Report 2009-101 77 November 2009 reviews, but as we also note on page 42, it has not reviewed 25 of the 58 counties during the past three years. Moreover, it already collects and compiles the counties’ data from the investigation activity reports, and we believe that a more thorough and rigorous review of the compiled county data rather than the minimal review that we describe on page 32 could identify potential county reporting errors. We are uncertain which time period Social Services refers to 9 when it acknowledges having “not physically reviewed 23 of the 58 counties in the last three years.” The time period we used for our review was the three‑year period from August 2006 to August 2009, thus it is likely that our time period is different. To avoid confusion, we clarified the text on page 42 to specifically state the time period we used. We also revised the number of counties that Social Services did not review during this period from 26 to 25 counties. Until its response, Social Services did not inform us of the 10 self‑assessment surveys that counties performed during fiscal year 2006–07. However, considering the extent to which we found the counties we reviewed did not appropriately follow up on match lists, we do not have a basis to believe the self‑assessments improved counties processing of information matching welfare recipients to data about their eligibility. Social Services asserts it conducted 27 IEVS reviews in fiscal 11 year 2008–09, however the review log it provided us indicates that Alameda, Contra Costa, and Santa Barbara counties were conducted or were to be conducted in fiscal year 2009–10, not 2008–09. The draft audit report we provided Social Services reflected 25 on‑site IEVS reviews based on Social Services’ planned review schedule. We revised the text on page 42 to reflect that 21 on‑site reviews and three desk reviews were performed in fiscal year 2008–09 based on the review log it provided us. We believe Social Services cited an incorrect year. Social Services’ 12 review log indicates that the last IEVS review of Los Angeles County was performed in 2005, not 2007. We stand by our conclusion that a cost‑benefit analysis of the 13 Statewide Fingerprint Imaging System (SFIS) is necessary. With any expenditure of public funds, a cost‑benefit analysis is prudent. There are indications that SFIS may not be as beneficial as Social Services asserts, specifically, the large backlog of unresolved cases of potential duplicate aid fraud that have been allowed to languish as well as the relatively small number of instances of duplicate aid fraud that it has identified—54 cases in 2008 and 845 since 78 California State Auditor Report 2009-101 November 2009 its implementation in 2000. Therefore, a cost‑benefit analysis is warranted to continue the ongoing expenditure of $5 million for SFIS. 14 As noted on page 44, in our 2003 audit, we concluded that Social Services’ methodology of projecting statewide savings using Los Angeles County’s estimated savings was flawed, especially its assumption that the incidence of duplicate‑aid fraud in Los Angeles County was representative of the incidence of this type of fraud statewide. 15 Rather than making a high‑level comparison to Arizona’s results, Social Services should review Arizona’s methodology and determine whether parts or all of the methodology would work for California, and if so, what data needs to be collected to conduct such a cost‑benefit analysis. 16 Social Services’ distinction between the State and federal share of funding for SFIS is dubious. Whether funding comes from the State or the federal government is irrelevant because when managing state and federal programs, Social Services has a responsibility to ensure that public funds are spent wisely. 17 We are unclear how Social Services believes that the actions it describes “more than justifies the deterrent value of SFIS.” We saw no indications that Social Services has any data—internal or from other states—of SFIS deterrent value. We acknowledge on page 45 that Social Services has performed limited studies on SFIS, which indicate that SFIS identifies fraud, which other systems do not, but these studies did not analyze the costs compared to the benefits SFIS provides or measure deterrence. 18 Social Services mischaracterizes our report. We do not attribute the reasons for the delay entirely to Social Services. Rather, as we state on page 49, a United States Department of Agriculture (USDA) financial management specialist and the Social Services manager indicated that they began to work on the delayed reconciliations in early 2008. Further, we acknowledge that in August 2009 the USDA directed Social Services to halt the reconciliations so they can coordinate the application of these funds and take other actions. 19 Social Services incorrectly implies we miscalculated the amount of the remaining funds to distribute. On page 49, we identify that the remaining $24.9 million is split between Social Services and the counties. When reporting this figure, we did not use the federal report that Social Services refers to, rather, we used the amounts Social Services identified on a worksheet it provided to us. California State Auditor Report 2009-101 79 November 2009 Social Services misquotes our report. We do not say that “it is 20 impossible to measure the deterrence effect of SFIS.” Rather, we state that the deterrence effect is difficult to measure. In fact, as we state on pages 45 and 46, Arizona developed a method to calculate the projected deterrence effect of its fingerprint imaging system. 80 California State Auditor Report 2009-101 November 2009 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