GAO FOOD STAMP PROGRAM Achieving Cost Neutrality in Washington’s Family Independence Program 1 I Illlllll 141698 - ,- __- (~Ao/lt(::EI)-!m-84 Resources, Community, and Economic Development Division B-237670 June 281990 The Honorable Patrick J. Leahy Chairman, Committee on Agriculture, Nutrition, and Forestry United States Senate The Honorable E (Kika) de la Garza Chairman, Committee on Agriculture House of Representatives This report discusses the state of Washington’s progress in achieving cost neutrality in the Food Stamp Program component of its Family Independence Program, a E-year welfare reform demonstration project that the state began operating on July 1, 1988. The report recommends various actions to improve (1) the methods for establishing the amount of benefit and administrative costs that Washington State would have incurred for food stamps in the absence of the Family Independence Program and (2) the accuracy of the data and the methods used to calculate cost neutrality each calendar quarter. Copies of this report are being sent to the Secretaries of Agriculture and Health and Human Services; the Governor of Washington State; the Director, Office of Management and Budget; and other interested parties. This work was performed under the direction of John W. Harman, Director, Food and Agriculture Issues, who may be reached at (202) 2764138. Other major contributors are listed in appendix IV. / J. Dexter Peach Assistant Comptroller General Executive Summary Breaking the “cycle of poverty” of welfare recipients has long been a Purpose principal, but elusive, goal of welfare reform. To test a new approach, the Congress authorized Washington State in 1987 to conduct the Family Independence Program (FIP), a S-year demonstration project that com- bines several welfare-related programs, including the Food Stamp Pro- gram, into a single grant package for recipients. Under FIP, food assistance is provided in the form of cash instead of food stamp cou- pons. The authorizing act requires (1) the state to assure that the cash approach is not more costly to the federal government than the tradi- tional coupon program would have been and (2) GAOto report to cogni- zant congressional committees and federal agencies on whether this “cost-neutral” requirement is being met. As a result of discussions with the responsible congressional oversight committees, GAOagreed to (1) verify the reasonableness and accuracy of the state’s methodologies and calculations made to measure cost neu- trality and (2) report on problems or unresolved issues that may impede the state from complying with the authorizing act’s requirement or affect its measurability. combines three federally supported programs: Aid to Families with Background FIP Dependent Children (AFDC), Medicaid, and food stamps into one grant package for welfare recipients. The project, which provides education, training, employment opportunities, and support services, is designed to test whether a coordinated approach works better than separately administered programs to reduce recipients’ long-term dependence on welfare. FIP was authorized in the Omnibus Budget Reconciliation Act of 1987 and was implemented under special terms and conditions developed in March 1988 by Washington State and the U.S. Departments of Agricul- ture (USDA) and Health and Human Services (HHS), The program began operations in July 1988. By July 1989 the state had implemented FIP at 20 (this number varies from the state’s count of 16 because of different calculation criteria) local welfare offices. These 20 offices will be the only ones included under FYP during the &year demonstration. Originally, the state had planned to adopt the program at all 66 of its local offices by the end of the S-year period; this plan was changed by the state in June 1989. Page2 GAO/RCEDflO-84 Wa&in#t&# Fmlly Independence Program i ExecutiveSummary The 1987 act’s requirement for an assurance of cost neutrality probably Results in Brief cannot be fully satisfied. Instead, a reasonable approximation of pro- gram costs is probably the best that can be achieved. Using methodolo- gies developed with the federal agencies, the state estimates the cost of continuing to run a traditional Food Stamp Program statewide and uses the estimates as ceilings for the amount it can claim from the federal government for combined food stamp and FIP food cash reimbursement. For the first 3 quarters of FIP operation (July 1988 through March 1989 with 16 local welfare offices operating under the project), the state claimed $145.4 million for food benefit costs and administrative expenses, or about $2.5 million below the total allowed under the ceilings. GAOcited several problems with the methods that the state and the fed- eral agencies used to set the program’s benefit and administrative ceil- ings and with the cost-neutral calculations made for the first 3 quarters. For example, because FIP will no longer be implemented statewide, the state no longer needs to project the possible food stamp benefit costs that would have occurred statewide. Consequently, using alternative methodologies, as well as other actions that GAOis recommending, could help improve the accuracy of the calculations and help better ensure that the cost-neutrality requirement is being met. Principal Findings Estimating Cost Neutrality In absolute terms, the cost-neutrality requirement in the 1987 act prob- ably cannot be ensured. Where the FIP demonstration has replaced the Food Stamp Program, the state and the federal agencies cannot deter- mine with certainty what the benefit and administrative costs would have been if the change had not occurred. Given the uncertainties and risks of errors associated with estimating the costs of a replaced pro- gram, a reasonable approximation of possible program costs, using sound estimates, is probably the best that can be achieved. To determine the cost of food stamp benefits, the state provides an approximation by making statewide estimates based on costs incurred at seven “control” welfare sites. The estimates serve as ceilings for the amount of money the state can request from the federal government as reimbursement for food stamp coupon and FIP food cash costs combined. The state is using a different methodology for setting the reimbursement Page 3 GAO/RCED-90-34 Washington’s Family Independence Program Executive Summary ceiling for food stamp administrative costs, The ceiling is based on his- torical food stamp administrative costs during a fixed baseline period, as adjusted and updated for current conditions. The state calculated that for the first 3 quarters of FIP operation, a tradi- tional statewide Food Stamp Program would have cost $141.7 million in benefits and $6.2 million in administrative expenses. During this period, actual expenses for food stamp coupons and FIP food cash were $139.7 million in benefits and $5.7 million for administration, or about $2 mil- lion and $500,000, respectively, below the ceilings. Estimating Benefits GAOfound several problems with the state’s current methodology for estimating the statewide benefit cost of a traditional Food Stamp Pro- gram. For example, the state’s method was developed when plans called for all 65 local welfare offices to switch to FIP. Since the number of offices converting to FIP was limited to 20, the continued use of the statewide method increases the risk of establishing inaccurate cost ceilings. GAObelieves the state needs to consider alternate methodologies that do not establish a statewide cost ceiling. Such alternatives may enable the state to lower the risk of making inaccurate cost estimates and better determine whether the cost-neutrality requirement is being met. For example, when cost estimates are made for only 20 sites, the potential additional cost to the federal government or penalty to the state resulting from any variation or inaccuracy in the estimation technique is significantly less than when costs for the state’s entire 65 sites are esti- mated, as is done now. GAOalso found several problems with the calculations of ceiling amounts and cost neutrality for the first 3 quarters of FIP operation. The use of inconsistent data sources, data, and calculation techniques reflects the need to use the best available data and to improve internal controls that will reduce errors in data-gathering and reporting. Estimating Administrative In May 1988 the state submitted a proposed method for calculating the costs program’s administrative cost-neutral ceiling to USDAand HHSfor their approval. GAOfound that the proposed method contained a number of v questionable assumptions. It assumes, for example, that FIP will be implemented statewide and that all eligible welfare recipients will even- tually convert to FIP. These and other assumptions may be invalid and Page 4 GAO/RCEDBO-84 Washington’s Family Independence Program Executive Sumnuuy increase the risk that administrative cost neutrality will not be deter- mined as accurately as possible. In August 1989 the state submitted a revised proposal that corrected some of the weaknesses GAOidentified. As of March 1990 Washington State had not received a response from either USDAor HHSregarding its May 1988 or August 1989 proposed administrative methodology. GAOalso found problems with the calcula- tions used to determine the ceiling amount under the proposed adminis- trative cost method. FIP faces the inherent problems associated with the demonstration of Recommendations any new welfare assistance approach. While aware of these problems, GAOis making several recommendations to the Secretaries of Agriculture and Health and Human Services to improve (1) the methods for estab- lishing the amount of the state’s food stamp benefit and administrative costs in the absence of FIP and (2) the accuracy of the data and the methods used to calculate cost neutrality. Commenting on a draft of this report, USDA’SFood and Nutrition Service, Agency Comments HHS,and Washington State generally agreed with GAO’Sfindings and con- clusions. Both federal agencies and the state agreed that since there are no plans to operate FIP on a statewide basis, as originally anticipated, a review of the cost-neutrality methodology is warranted. However, all three organizations commented that it was unclear whether the several alternative methods cited by GAOfor modifying cost-neutrality calcula- tions would provide a more accurate cost ceiling than the current meth- odology. As stated in the report, the alternative methods discussed are only examples and are not intended as specific recommendations. According to Washington State, the report provides an excellent over- view of the many methodological and data issues that are connected with measuring FIP cost neutrality and is also relevant for the AFDCand Medicaid program portions of FIP. State officials pointed out, however, that two broader issues must be examined before a final consensus on cost neutrality for food stamps, AFDC,or Medicaid can be reached. The first is HHS’refusal to include the costs associated with the recent unan- ticipated caseload growth in the state’s welfare system as part of its FIP cost calculation. The second is HHS’refusal to allow the state to change the method of accounting for welfare costs of FIP participants who migrate (move permanently) from FIP sites to control sites. (See apps. I- III.) Page 5 GAO/RCED-fJO-84 Washington’s Family Independence Frogram Contents Executive Summary 2 Chapter 1 8 Introduction Washington State’s Family Independence Program 8 Food Stamp Program 10 FIP Agreements 12 Objectives, Scope, and Methodology 13 Chapter 2 17 State’s Estimating Extent of Cost Neutrality Can Only Be Estimated State’s Method for Estimating Benefit Costs 17 18 Methods Show FIP Is State’s Method for Estimating Administrative Costs 20 Cost-Neutral Comparison of Cost Ceilings and Actual Costs for First 3 22 Quarters of FIP Operation Chapter 3 26 Alternative Estimating Benefit Cost-Neutral Methodology Based on Questionable 26 Assumptions Methods May Better Alternative Methods Exist for Setting Benefit Ceilings 29 Ensure Benefit Cost State and Federal Quarterly Calculations Differed 32 Neutrality Conclusions 36 Recommendations 37 Agency and State Comments and Our Evaluation 38 Chapter 4 44 Cost Neutrality of Federal Review of State Proposal Is Slow Proposed Methodology Uses Questionable Assumptions 45 46 Administrative State’s Quarterly Calculations Need to Be Viewed With 60 Expenses Is Still Caution Unclear FNS Not Making Required Calculations 61 Conclusions 62 Recommendations 63 State Comments and Our Evaluation 63 Appendixes Appendix I: Comments From the U.S. Department of 66 Agriculture’s Food and Nutrition Service Appendix II: Comments From the Department of Health 60 Y and Human Services Appendix III: Comments From the State of Washington 63 Appendix IV: Major Contributors to This Report 76 Page 6 GAO/RCED-!XMM Washington’s Family Independence Program . Contents Tables Table 2.1: Washington State’s Cost-Neutral Calculations 22 for Food Assistance Benefit Costs Table 2.2: FNS’ Cost-Neutral Calculations for Food 22 Assistance Benefit Costs Table 2.3: Washington State’s Cost-Neutral Calculations 23 for Food Assistance Administrative Costs Table 2.4: Washington State’s Cost-Neutral Calculations 24 for Food Assistance Benefit and Administrative Costs Combined Table 3.1: Proportion of State Food Stamp Costs 27 Represented by the Seven Control Sites Table 3.2: Impact of a 1 Percentage Point Variation in 31 Costs on Cost-Ceiling Estimates Table 3.3: Comparison of the State’s and FNS’ Benefit 33 Cost-Neutral Calculations for the First 3 Quarters of FIP Operation Abbreviations Aid to Families with Dependent Children FIP Family Independence Program FNS Food and Nutrition Service GAO General Accounting Office HHS Department of Health and Human Services USDA US. Department of Agriculture Page 7 GAO/RCEDW44 WaAington’r Family Independence Program Chapter 1 Introduction The Omnibus Budget Reconciliation Act of 1987 (P.L. 100-203) author- ized the state of Washington to conduct the Family Independence Pro- gram (FIP), a 5-year welfare reform demonstration project. Under the project, the federally supported Aid to Families with Dependent Chil- dren (AFDC), Medicaid, and Food Stamp Programs are combined into a unified grant assistance package for Al%&eligible welfare recipients. As a part of the FIP grant, the recipients are provided with food assistance in the form of cash instead of food stamp coupons (a concept referred to as “food stamp cash-out”). The 1987 act amended the Food Stamp Act of 1977 (section 21(b)(4)) to require Washington State to ensure that the federal government’s cost for the cash food assistance provided under FIP does not exceed what its cost would have been under the traditional food stamp coupon program. Section 21(g) of the Food Stamp Act of 1977, as amended, requires GAO to conduct periodic audits to determine whether this “cost-neutral” requirement is being met,’ and to report the results to the Senate Com- mittee on Agriculture, Nutrition, and Forestry, the House Committee on Agriculture, and the Secretaries of Agriculture and Health and Human Services. This report discusses our observations on the extent to which Washington State is complying with the cost-neutral requirement and on issues that may impede or otherwise affect its ability to do so. purpose is to test an approach to breaking the cycle of poverty and Washington State’s FIP’S reducing recipients’ long-term dependence on welfare. To this end, it Family Independence provides current and new welfare recipients with financial incentives Program for attaining training and employment, and it increases the range of sup- port services available to them. The program, which was authorized both by the Congress and the Washington State Legislature, generally covers welfare recipients in certain areas of the state who qualify for AFLKT. Most of the recipients are members of single-parent, low-income families. Among FIP’S incentives and support services are the following: l increased program funding for training, education, and job search activities; l cash incentives (up to 35 percent of the combined AFDCgrant and food stamp benefit amounts) for recipients who are in training or working; ‘Also referred to by state and federal officials and documents as a “budget-neutral” and a “revenue- neutral” requirement. For consistency, this report uses the terms “cost-neutral” and “cost neutrality.” Page 8 GAO/RCED-SO-84 Washington’s Family Independence Program Cha9t.m 1 introduction 9 increased child care benefits; 9 extended child care and Medicaid benefits for 1 year after recipients cease to receive FIP cash benefits as a result of increased earnings; and 9 replacement of food stamp coupons with an equivalent amount of cashe2 Although the “cash-out” of food stamps is a highly visible feature of FIP, state officials do not expect it to have as significant an impact in changing the current welfare system as some of the programmatic changes made to the AF’DCportion of the project. FIP was initially implemented in 9 of the state’s 65 local welfare offices on July 1, 1988.” Seven more offices were converted to the FIP program on October 1, 1988. FIP was to have continued to be phased in at addi- tional local offices during the first 3 years until most of the 66 offices were converted to the demonstration project. However, in June 1989, concerned about the demonstration’s escalating costs, Washington State limited the number of additional offices to be converted to 4, for a total of 20 local welfare office sites. The four additional offices began FIP operations on July 1,1989, near the completion of our field workm4 After a site implements FIP, participation in the program is mandatory for all new AFDC applicants. Existing AFDC recipients have the option to convert to FIP at the time of their next annual face-to-face review with their welfare office caseworker or to remain on the regular AFDC, Medi- caid, and Food Stamp programs. As of March 31,1990, about 30,000 of the approximately 83,000 Washington families on AFDC were on PIP. At the state level, the program is administered through the Department of Social and Health Services and the Employment Security Department. Several local, regional, and state advisory groups, including a joint House-Senate legislative budget committee, also participate in aspects of FIP design and implementation. Final authority for FIP policy decisions lies with the state legislature, which receives policy recommendations ‘Although PIP participants receive food assistance in the form of cash, food stamp coupons are still used in the state for non-AFDC households and for AFDC households that do not convert to FIP. “Although 68 local welfare offices were listed in state statistical tables as of April 1989, 1 of the offices had been closed and 2 others were specialized offices that did not have Food Stamp Program activity. Thus, the state could convert 66 local welfare offices to the FIP program. %ur count of the number of local welfare offices in the state and in FIP differs from the count normally used by the state, because we counted each local office individually whereas the state counts two or more offices ln several locations as a single office. For example, as discussed in this paragraph, 16 local offices were ln FIP as of October 1,1988, and 20 in FIP as of July 1,1989. The state’s count of these FIP offices for the same periods was 11 and 16, respectively. Page 9 GAO/RCED-99-84 Washingt.on’s Family Independence program Chapter 1 Introduction from an Executive Committee comprised of representatives from the two administering state departments, the state’s Office of Financial Management, and the welfare community. Day-to-day activities of the Executive Committee, including policy coordination, fiscal, and commu- nity development functions, are carried out by an executive director and staff. At the federal level, FIP oversight is the responsibility of the Department of Health and Human Services (HHS) as the lead agency and, for the food stamp cash-out component, the U.S. Department of Agriculture’s (USDA) Food and Nutrition Service (FNS)." Key agreements between the state and the federal agencies that supplement the 1987 authorizing act, including stipulations the state had to meet to obtain final federal approval to operate the FIP demonstration, are contained in a special terms and conditions document dated March 9,1988.” The Food Stamp Program is a food assistance program available to all Food Stamp Program applicants who meet its eligibility criteria. The program is administered as a cooperative federal/state effort. It was designed to increase the food purchasing power of low-income households to permit them to buy a more nutritious, low-cost diet. To be certified as eligible for food stamps, a household must meet income and resource requirements unless all members receive AFDCor Supplemental Security Income, in which case the household is automatically eligible for food stamps. Unless exempted for reasons such as age, disability, current employ- ment, or AFDC work program registration, household members must reg- ister for work and comply with the requirements of a training and employment program. In addition, the household must (1) meet several other nonfinancial standards, including citizenship or eligible alien status; (2) provide social security numbers; and (3) if a student, meet “During the pre-implementation phase of FIP, the lead federal agency was the Interagency Low Income Opportunity Advisory Board. The Board was established in July 1987 to help carry out the President’s welfare reform initiatives by providing a focal point for coordinating state welfare reform proposals that affect more than one federal program. The Board, which is advisory only, serves aa a focal point for states in obtaining from the federal agencies the necessary statutory waivers to enable them to conduct their demonstrations. On or about July 1,1988, when FIP began to be implemented in Washington State’s local welfare offices, the lead was transferred from the Board to HHS. Several component units of HHS share responsibility for oversight of the FIP demonstration, including the Office of the Assistant Secretary for Planning and Evaluation, the Family Support Administration, and the Health Care Financing Administration. Throughout this report these components are referred to individually and collectively as “HHS.” “The special terms and conditions agreement contains 35 provisions covering, among other matters, project approval, continuation, and termination; cost neutrality; project evaluation; and reporting requirements. Page 10 GAO/RCED-90-84 Washington’s Family Independence Program chapter 1 Introduction certain criteria. Food stamp benefits are based on household size, count- able monthly income,7 and the cost of purchasing food using the Thrifty Food PlanR Benefits are issued in the form of food coupons, an average of about $139 a month per AFDC household in Washington State in fiscal year 1988, that recipients can use to purchase food to obtain a more nutritious diet. The program is administered nationally by FNS, which pays for all food stamp benefits and about half of the states’ administrative expenses. States are responsible for local administration and day-to-day operation of the program. Federal costs for food stamps in Washington State for fiscal year 1988 amounted to about $176 million in benefits and $20 mil- lion in administrative expenses.” Under the PIP demonstration, the primary difference from the Food Stamp Program is that eligible participants receive cash instead of food coupons for their food assistance benefits. With minor exceptions, changes did not occur in the rules for determining whether a household is eligible to receive food benefits or for determining the amount of monthly food benefits a household receives.1° However, under a “hold harmless” provision of the authorizing legislation, FIP participants are guaranteed to receive no less in FIP food cash assistance than they would have received under the regular Food Stamp Program. Despite the limited programmatic changes to the traditional coupon pro- gram under FIP, and in three other food stamp cash-out demonstrations 7Not all of a household’s income is actually counted. Some exclusions and deductions are allowed in determining the level of its food stamp benefits. ‘The lowest-cost USDA food plan that incorporates information on food consumption, prices, and nutrient composition of foods and on human nutritional requirements “Total federal dollars for the AFDC and Medicaid programs in Washington State for fiscal year 1988 were about $226 million and $633 million, respectively. “‘An indication of the absence of eligibility and benefit differences is also reflected in state accounting records. For example, data for June 1989 showed that each person in an AFDC household received an average of $63.26 in food stamp coupon benefits while each person in a PIP household received a virtually identical average of $63.62 in food cash benefits. Although the June 1989 data showed that per capita food cash benefits were 27 cents higher than per capita coupon benefits, this difference does not necessarily indicate a violation of the cost-neutral requirement, because cost neu- trality applies to food assistance costs statewide for benefits and administration combined for a 3- month period. As described in chapter 2, the state calculated that for the first 3 quarters of FIP operation, it was more than $2.6 million below the applicable cost-neutral ceiling. Page 11 GAO/RCEDSO-84 Washing-ton’s Family Independence Program Chapter 1 Introduction that are underway or that FNShas approved,l’ the cash-out of food stamps has always been considered controversial. Supporters contend that cash-out saves money (because printed coupons no longer need to be issued, transported, stored, safeguarded, redeemed, processed, and accounted for), simplifies program operations (because the state no longer needs to monitor the authorized use of food stamps or investigate and prosecute their misuse), and reduces the welfare stigma associated with food stamps. Opponents contend that cash-out eliminates the gov- ernment’s control over program funds to ensure that they are used only to buy food and eliminates many welfare households’ control over budgeting their funds to ensure that resources are available to buy food. They also believe that cash-out may lead to reduced nutritional levels and ultimately a decrease in benefits because policymakers could more easily cut cash benefits than food benefits. Some of these issues, as well as other nutritional and expenditure impacts of the FIP food stamp cash- out demonstration, are being studied by The Urban Institute, under con- tract with the state of Washington, as part of a comprehensive program evaluation of FIP. According to the March 1988 special terms and conditions agreement FIP Agreements governing FIP, the state is required to achieve food stamp cost neutrality each calendar quarter. Under formulas explained in more detail in chapter 2, the state estimates what the federal government’s costs in benefit payments and administrative expenses would have been for the 3-month period if the entire state had continued to operate under the traditional food stamp coupon program. For that period, the estimates become ceilings, or maximums, for the amount of reimbursement the state can receive from the federal government for the aggregate of cash food assistance for welfare recipients on FIP and food stamp coupon assistance for recipients not on FIP. FNSand HHShave approved the state’s method and data for estimating benefit payments. As of March 1990, they were still reviewing a separate method that the state pro- posed for estimating administrative costs. The requirement for the state to adhere to the food stamp cost-neutral ceiling every 3 months contrasts with the 36-month period that the spe- cial terms and conditions give the state to attain cost neutrality for the AFDCand Medicaid portions of FIP. The shorter time period was imposed I ‘Alabama, California’s San Diego County, and portions of New York State. According to an FNS official, a second food stamp cash-out demonstration in Alabama, as well as a welfare reform demon- stration program being proposed by Minnesota, do contain changes in eligibility rules and benefit levels from the regular Food Stamp Program. Page 12 GAO/RCED-90-84 Washington’s Family Independence Program Cheppter 1 Introduction for the Food Stamp Program component of FIP because of federal agency concern that a provision of the 1987 amendment to the Food Stamp Act of 1977 authorizing FIP might result in runaway costs and make it diffi- cult for the state to achieve food stamp cost neutrality once it had sig- nificantly exceeded the cost-neutral ceiling. On October 11, 1988, a technical amendment to the Food Stamp Act of 1977 removed that con- cern by modifying the language in the authorizing legislation, prompting the state to request federal approval to change the food stamp cost-neu- tral period to 36 months making it the same as the AFDCand Medicaid requirements. As of March 16, 1990, the request was still pending with FNS and HHS. If the state’s request is approved, the change in the cost-neutral attain- ment period could affect the amount of funds that FNSreleases quarterly to the state for the federal portion of food assistance payments. Under the revised arrangement, FNSwould be permitted to release any amount of food assistance funds the state requests for the first 6 quarters of PIP operation, rather than releasing no more than the cost-neutral ceiling amount. If, at the end of the 6 quarters, the state has incurred cumula- tive costs over the cost-neutral maximum for that l&month period, the state would pay back one-sixth of the overage each quarter for the next 6 quarters. Thus, not until the end of 12 quarters or 36 months of FIP operation would the state be required to attain food stamp cost neu- trality. For the final 8 quarters or 24 months of the demonstration’s 5- year period, the state would be permitted to receive only the cost-neu- tral maximum calculated for each quarter. We conducted this review to meet the periodic audit requirement of the Objectives, Scope,and 1987 act’s amendments to section 21(g) of the Food Stamp Act of 1977. Methodology As the result of discussions with the Senate Committee on Agriculture, Nutrition, and Forestry and the House Committee on Agriculture, we agreed to . verify the reasonableness of the methodologies and calculations that the state used to establish the benefit and administrative cost-neutral ceil- ings, and the accuracy of the state’s and FNS’quarterly calculations of cost neutrality; and l to the extent possible, report on any problems or unresolved issues that may impede the state in complying with the cost control provision of the authorizing legislation or affect its measurability. Page 13 GAO/RCED-SO-84 Washington’s Family Independence Program Chapter 1 Introduction To verify the reasonableness of the benefit and administrative cost-neu- tral ceilings and monitor the accuracy of the state’s and FNS’quarterly calculations of cost neutrality, we selectively reviewed the state’s cost accounting systems, cost records and reports, and internal controls. For food stamp benefit costs, we examined the reasonableness of the meth- odology agreed to by the state and the federal government for using selected “control” sites to calculate the ceiling, as well as the accuracy of the cost data reported for the control sites.‘” We also analyzed (1) the appropriateness of the assumptions underlying the methodologies, (2) whether the assumptions are still valid, and (3) several examples of alternative methodologies to determine whether they might result in greater assurances of cost neutrality. For food assistance administrative costs, we examined the state’s proposed methodologies and analyzed their underlying assumptions. For both benefit and administrative costs, we compared food stamp cost neutrality calculations made by the state and FNSfor the state as a whole for each of the first 3 calendar quarters of FIP operation (July 1,1988, through March 31, 1989), using state records and the currently agreed-upon methodologies, compared our results with calculations made by the state and FNS,and determined the reasons for any differences. Also, we discussed plans for auditing or evaluating FIP and examining food assistance cost neutrality with repre- sentatives of USDA'Sand HHS’ Offices of Inspector General, the Wash- ington State Auditor, the state agencies’ internal auditors, and FNS’ Financial Management Division and Food Stamp Program Division. To identify and track problems or unresolved issues that could impede the state in meeting the food stamp cost-neutral requirement or affect the measurability of the state’s compliance, we reviewed agencies’ files and records and interviewed state and federal agency officials and staff, principally at state offices in Olympia, Washington; FNSheadquarters in Alexandria, Virginia, and its Western Regional Office in San Francisco, California; and HHSheadquarters in Washington, DC. We examined whether necessary actions required of the state by the project’s special terms and conditions, such as providing written assurances and main- taining and providing cost data and other records, had been taken, and whether the actions were correct and/or reasonable. In addition, we interviewed advisory group representatives, welfare advocates, and other relevant organizations that are assisting in and/or monitoring the implementation of FIP, and we reviewed the progress and applicable ‘2‘Control” welfare offices or sites are also referred to by state and federal officials and documents as “comparison” offices or sites. For consistency, we use the term “control” offices or sites throughout the report. Page 14 GAO/RCED-99-94 Washington’s Family Independence FVogram Chapter 1 Introduction findings of evaluations being made of the program. We also visited five local welfare offices that had implemented FIP or were serving as control sites. Several limitations apply to our work. First, because FIP is a 5-year pro- ject whose development, implementation, and evaluation require coordi- nation among many different state and federal agencies, some of its policies and procedures were just being put into place or were being revised, and others were still pending, at the time we completed our field work. For some of the unresolved matters, the timing of their com- pletion and the nature of their resolution were still uncertain, as was their likely impact on cost neutrality. Although we continued to update this report after completing our field work to acknowledge and, to the extent possible, evaluate subsequent actions, some additional changes to the FIP program that are identified in this report as pending matters may have occurred before publication. Second, because FNSand HHShad not yet completed their review of the state’s proposed methodologies for calculating the cost neutrality of administrative costs, the state’s, FNS’ and our administrative cost calcu- lations (discussed in chs. 2 and 4) were based on the state’s proposals. The methodology could change as a result of the federal government’s review and approval process, or as a result of our findings and recom- mendations, which could in turn change the cost-neutral calculations. Finally, the legislative mandate that requires us to audit the cost neu- trality of the FIP demonstration pertains only to the food stamp cash-out portion of the project. The legislation does not require us to monitor the state’s compliance with the cost-neutral provisions of either the AFDCor the Medicaid portions of FIP, nor are we required to evaluate the overall implementation, management, or results of the FIP demonstration. Instead, the 1987 act requires the Secretary of Agriculture, in consulta- tion with the Secretary of Health and Human Services, to conduct a pro- ject evaluation. In April 1989, in accordance with FIP’S special terms and conditions, the state contracted with The Urban Institute, a research firm located in Washington, DC., for a comprehensive program evalua- tion of all three components of the FIP demonstration. The Urban Insti- tute’s evaluation was scheduled to be completed in late 1990, but according to FNSand HHSofficials in February 1990, it has been resched- uled for completion in late 1993. Although our review did not cover the state’s compliance with the AFDCand Medicaid cost-neutral require- ments, many of the findings and recommendations in this report should Page 15 GAO/RCED-90-84 Washlngton’s Family Independence Program Chapter 1 Introduction be applicable to those programs because the state’s methods for deter- mining cost neutrality are the same as, or similar to, those for the Food Stamp Program component of FIP. We conducted our review from October 1988 to July 1989 in accordance with generally accepted government auditing standards. Chapter 2 of this report describes the state’s methodologies for esti- mating what benefit and administrative costs would have been in the absence of FIP, and the resultant ceilings that were established, actual costs incurred, and the state’s and FNS’cost-neutral calculations for the first 3 calendar quarters, or 9 months, of FIP program operation. Chap- ters 3 and 4 present the results of our analyses of the methodologies and calculations for benefit costs and administrative costs, respectively. Page 16 GAO/RCED-90-94 Washington’s Family Independence Program Chapter 2 &at&s Estimating Methods Show F’IP Is Cost- Neutral In absolute terms, the 1987 authorizing act’s requirement for an assur- ance of cost neutrality probably cannot be fully satisfied. As the March 1988 special terms and conditions agreement between the state and the administering federal agencies recognizes, a reasonable approximation of likely program costs, using sound estimates, is probably the best that can be achieved. Two methods were developed for estimating what the federal govern- ment’s food assistance costs would have been if the traditional Food Stamp Program had been maintained statewide. One method pertains to the cost of benefits to welfare recipients and the other to the state’s cost of administering the program. Using these approaches, the state calcu- lated for the first 9 months of FIP operation that a statewide Food Stamp Program would have cost $147.9 million. The state reported that for this same period its actual expenses for food stamp coupons and FIP food cash were $145.4 million, or about $2.5 million below the cost-neutral maximum. The 1987 act required the state to ensure that the federal government’s Extent of Cost cost of food assistance provided under the project would not exceed the Neutrality Can Only sum of the anticipated value of the coupons that would have been dis- Be Estimated tributed, and the federal share of administrative costs that would have been incurred, under the traditional Food Stamp Program. In its strictest interpretation, this requirement probably cannot be fully guaranteed. Because the food coupon program no longer exists for those FIP welfare recipients who receive food cash benefits, the state cannot determine the exact amount of costs that would have been incurred if the tradi- tional program had not been replaced by FIP. Similarly, the use of esti- mating methodologies cannot ensure cost neutrality conclusively because the uncertainties and risks of errors associated with estimating the costs of a replaced program work against such accuracy. A reason- able approximation of probable costs, using sound methodologies that minimize the likelihood of estimation errors, is probably the best that can be achieved. In accordance with the project’s March 1988 special terms and condi- tions agreement, the state developed separate methodologies for deter- mining cost neutrality for both benefits and administration of the Food Stamp Program. Each calendar quarter the state is required to make separate calculations for each of these two cost categories to arrive at cost-neutral ceilings for the quarter, compare actual costs incurred during the quarter with the ceilings, and determine an amount over or Page 17 GAO/RCED-90234 Washington’s Family Independence Program chapter 2 State’s Estimating Methods Show FIP Is Cost- Neutral under the ceiling for each type of cost. The two amounts are then summed to determine whether the state has achieved cost neutrality (i.e., is at or below the ceilings) for the food assistance program as a whole for the quarter. In addition to the state’s calculations, FNS’ Western Regional Office in San Francisco is required to make indepen- dent quarterly assessments to determine whether the state is in compli- ance with the authorizing act’s requirement. The state and the responsible federal agencies agreed that food stamp State’s Method for benefit cost data from randomly selected local welfare offices that did Estimating Benefit not convert to FIP could be used to provide a reasonably accurate esti- costs mate of benefit costs for the state as a whole. Methodologies were there- fore established to select sites to continue to operate the traditional Food Stamp Program and to project benefit cost data from those sites to the entire state. Because the state planned to implement FIP in all of its local welfare offices during the 5-year demonstration period, it needed a method to determine, for the state as a whole, what food stamp benefits would have cost in the absence of FIP. According to the methodology agreed to by the state and the federal government during the design of the FIP program, and subsequently prescribed in the program’s March 1988, special terms and conditions agreement, the state will estimate the cost of statewide food stamp benefits by using data on food stamp costs from randomly selected local welfare offices in which FIP will not be imple- mented for at least 3 years. These “control” offices or sites serve as sur- rogates for determining what statewide food stamp costs would have been without the FIP program. Seven control sites were randomly selected, along with a matched group of sites at which FIP was to be implemented, for program comparison and evaluation purposes. The welfare offices in the second group are referred to as “treatment” offices or sites. The state and the overseeing federal agencies developed the methodology for randomly selecting the two matched sets of sites. The site selection process contained four main components: (1) the matching of welfare offices in the state on the basis of a set of 11 criteria (9 measures of program similarity and 2 geo- graphic measures); (2) the performance of a statistical analysis to deter- mine caseload sample size; (3) the random selection of matched sites to draw the caseload sample based on a probability-proportional-to-size Page 18 GAO/RCED-80434 Washington’s Family Independence Program Chapter 2 State’s E&bating Method8 Show FIP Is Cm& Neutral sampling method;’ and (4) the random assignment of the selected sites either to the control or the treatment group. The seven offices selected as control sites were Okanogan, Yakima, Yakima/Kittitas, Shelton, Pierce West, King South, and Federal Wayn2The average monthly wel- fare caseload served by the seven control offices in the fiscal year pre- ceding the sample selection was 11,486. After selecting the control sites for evaluation purposes and deciding, along with the federal agencies, to use them for cost-neutrality purposes as well,:! the state determined the proportion of its total statewide food stamp benefit costs that these seven sites represented for the agreed- upon 12-month baseline period of April 1987 through March 1988. During that period the state’s total food stamp benefit costs were $165,134,288; the seven control sites’ costs were $23,624,198, or 14.31 percent of the state’s total costs. As agreed by the state and the federal agencies, this percentage is to be used each calendar quarter during the S-year FIP project as a factor in estimating food stamp benefit expendi- tures for the state as a whole. Each quarter, total food stamp coupon expenditures in the seven control sites are to be divided by .1431 in order to estimate food stamp benefit costs for the entire state. This esti- mated figure then constitutes the maximum food assistance benefit cost (covering both food cash costs for those recipients on FIP and food stamp coupon costs for those not on FIP) for which the federal government will reimburse the state. For example, state accounting data showed that for the first quarter of FIP operation (July 1, 1988, through September 30, 1988) food stamp coupon expenditures in the seven control offices totaled $6,374,861. That benefit amount, divided by .1431, totaled $44,548,295, which became the ceiling or maximum amount that the state could claim from the federal government for food assistance benefits for the quarter. ’ A sampling method in which a unit’s chance of being randomly selected is in direct proportion to its sire or the number of its component pieces. Under this method, a welfare office with a caseload of 1,000 clients had twice the chance of being selected as a control or treatment site as an office with a caseload of 600 clients. “Welfare offices selected as treatment sites were Goldendale, Stevenson, White Salmon, Othello, Moses Lake, Spokane North, Burien, West Seattle, and Everett. “The seven control sites being used to estimate statewide food stamp benefit costs are also being used by the state to estimate what statewide AFDC and Medicaid benefit costs would have been in the absence of FIP. Those two FIP program components also have cost-neutral requirements included in the Omnibus Budget Reconciliation Act of 1987. Page 19 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 2 State’s Estimating Methods Show FIP Is Cost- Neutral Unlike the method for estimating benefit costs, the method for esti- State’s Method for mating what the state’s costs of administering the Food Stamp Program Estimating would have been in the absence of FIP was not established in advance. Administrative Costs Rather, the terms and conditions required the state to submit a plan acceptable to USDAand HHS,within 60 days of the agreement’s March 9, 1988, effective date, for estimating administrative costs. Therefore, the state submitted a methodology proposal to the two agencies in letters dated May 3 and May 5, 1988, respectively.4 The state’s administrative cost-neutral proposal differs from the method used for benefit costs in that it does not use cost data from the control sites to estimate statewide costs. Instead, the method compares histor- ical food stamp administrative costs during a fixed baseline period with actual administrative costs during the FIP period, using the following four steps: l First, an initial annual administrative cost ceiling is to be calculated by (1) determining total Food Stamp Program administrative costs for the state and the federal share thereof for the baseline period of April 1, 1987, to March 31, 1988; (2) adjusting the ceiling by annualizing certain costs, such as salary increases and inflation that occurred during the last quarter of the baseline period; and (3) determining the portion of the administrative costs that relates to the FIP target population state- wide. (As described in more detail in ch. 4, the state calculated this amount to be 42.57 percent.) l Second, the administrative cost ceiling is to be amended as necessary to account for several types of changes occurring after the baseline period, such as prior period expenditure adjustments that affect the baseline period; legislatively mandated state employees’ salary and/or benefit increases; Social Security and other employer-paid payroll tax increases; inflation; and cost increases due to federal program changes or increased federal funding levels. The result of these two steps is the amended annual administrative cost ceiling.” 41nproviding technical comments on a draft of this report, FNS headquarters officials told us that the state’s methodology proposal was not submitted to USDA until August 11,1988, rather than on May 3, 1988, as shown ln the report. That is not correct. The state’s administrative cost-neutral proposal was submitted to USDA and HHS ln attachments to its May 3,1988, and May 6,1988, letters, respec- tively, as the report indicates. The August 11, 1988, submission referred to by FNS covers the state’s benefit cost-neutral proposal, not its administrative cost-neutral proposal. “The administrative cost ceiling is sometimes referred to by federal and state officials and documents as the administrative cost “baseline.” For consistency with the terminology used for benefit costs, this report uses the term “ceiling” instead of “baseline” for the administrative cost maximum calcu- lated for each quarter. Page 20 GAO/RCED-90-84 Washington’s Family Independence Program chapter2 State’s E&mating Methods Show F’IP Is Cost- Neutral l Third, the federal portion of actual administrative expenditures for each quarter is to be determined and then adjusted to exclude certain extraordinary costs, such as salary increases applicable to all positions in a given job classification, and increased staff costs caused by caseload or work load increases not attributable to FIP. The costs are then multi- plied by the percentage that relates to the FIP target population (42.57 percent). The result of this step is the adjusted actual administrative expenditure chargeable to the federal government. . Fourth, a comparison of actual costs with the ceiling is to be made quar- terly. Each quarter’s adjusted actual expenditure is compared with one- fourth of the amended annualized ceiling, and an amount over or under the ceiling is calculated. The final cost-neutral calculation for each quarter consists of summing the results of the benefit and administrative cost-neutral calculations to determine whether the state has achieved cost neutrality (i.e., is at or below the ceilings) for the food assistance program as a whole for the quarter. FNSofficials at headquarters and the Western Regional Office in San Francisco evaluated the state’s administrative cost-neutral proposal in August and October 1988, respectively, and internally raised several questions with both the methodology and the data used by the state. Although FNS' San Francisco office agreed with its headquarters’ recom- mendation that FNSand HHSjointly respond to the state’s proposal (because the proposal contained similar methodology covering the AFDC and Medicaid components of FJP),the San Francisco office expressed its concern that a joint response could delay approval of a methodology by which to evaluate FIP cost neutrality, and it urged that a joint federal response be issued to the state as expeditiously as possible. HHSofficials subsequently told us that, because this was the first administrative cost- neutral methodology they had ever reviewed, they wanted to see what the state’s application of the methodology to the first 2 quarters of actual expenditures looked like before they completed their review and responded to the state. The state submitted data to HHSon the first 2 quarters of FIP operation, including its calculations of cost neutrality, on May 4,1989. From May through September 1989, according to FNSand HHSofficials, the agencies were discussing and reviewing the state’s May 1988 proposal. On August 11, 1989, the state submitted a revised administrative cost- neutral methodology proposal to HHS.As discussed in chapter 4, the revision’s two alternatives make several changes in the methodology Page 21 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 2 State’s Estimating Methods Show F’IP 18Cost Neutral incorporated in the first proposal. In its March 16, 1990, comments on a draft of this report (see app. III), the state indicated that it had still not received a formal response from FNSor HHSon its administrative cost proposal. Tables 2.1 through 2.4 present information on the calculated cost ceil- Comparison of Cost ings and actual costs incurred during the first 3 calendar quarters of FIP Ceilings and Actual operation (July 1988-Mar. 1989), using the methodologies developed. Costs for First According to state calculations, the state was under the overall cost-neu- tral ceiling for benefit and administrative costs combined by about $2.5 3 Quarters of FIP million. Operation Benefit Costs As shown in table 2.1, the state of Washington calculated that it was slightly more than $2 million under the cost-neutral maximum for food assistance benefit costs for the first 3 quarters of FIP operation. Table 2.1: Washington State’s Cost- Neutral Calculations for Food Assistance Costs and ceilings 1st Quarter 2nd Quarter 3rd Quarter Total Benefit Costs Food Stamp/ FIP food cash costs $43590,175 $46,900,932 $49,162,929 $139,654,036 Federal maximum based on control site projections 44,560,566 47,517,140 49,590,361 141,666,067 Amount over (under) cost- neutral maximum $(970,391) $(616,208) $(427,432) $(2,014,031) Using state financial data, FNSalso calculated that the state was under the benefit cost-neutral ceiling for the first 3 quarters of FIP, although by a slightly greater amount, as shown in table 2.2. Table 2.2: FNS’ Cost-Neutral Calculations for Food Assistance Benefit Costs and ceilings 1st Quarter 2nd Quarter 3rd Quarter Total costs Food Stamp/ FIP food cash costs $43,685,964 $46,855,589 $48,974,319 $139,515,672 Federal maximum based on control site projections 44,540,295 47504,032 49,584,284 141,636,611 Amount over (under) cost- neutral maximum $X862,331) $(648,443) $(609,965) $(2,120,739) The state’s and FNS’ calculations differ primarily because they used dif- ferent state reports and other data sources. This matter, as well as other Page 22 GAO/RCED-90-84 Washington’s Family Independence Program chapter 2 Stat& Estimating Methods Show FIP Is Cost- Neutral aspects of our evaluation of benefit costs, is described in more detail in chapter 3. Administrative Costs At the time we completed our field work, the state had made administra- tive cost-neutral calculations for the first 3 quarters of FIP operation (July 1988-Mar. 1989). The state’s calculations of administrative cost neutrality are shown in table 2.3. Those computations indicated that the actual costs chargeable to the federal government for administering the food stamp coupon and FIP food cash programs for the 3 quarters were about $536,000 below the cost-neutral ceiling the state had calculated. In addition, FNShad calculated the project’s administrative cost ceiling, with results that differed slightly from the state’s determination, but it had not yet calculated actual state costs for comparison against the ceiling. Our evaluation of both the methodology and the calculations is shown in chapter 4. Table 2.3: Washington State’s Cost- Neutral Calculations for Food Asrlstance First 3 Admlnlstratlve Costs Costs and ceilings Annual quarters Initial ceiling $8,242,105” $6j 81,579 b b Amendments to ceilina Amended ceilina $8.242,105 $6.181.579 Actual costs July-September 1988 $5,323,345 October-December 1988 4364,392 January-March 1989 5,425,722 Total actual costs 15.113.459 Adiustments to actual costs (1,852,290) Balance 13,261,169 Portion related to FIP target population 42.57% Adiusted actual costs $5.645.280 Amount over (under) cost-neutral ceiling $X536,299) aCalculated by the state according to the proposed methodology described earlier in this chapter. bin its calculations for the first 2 quarters, the state included a series of amendments that reduced the ceiling by $314,859. In its third quarter calculation documents, which show calculations cumulatively for the 3 quarters (rather than on a quarter-by-quarter basis), the state began to show the revisions as adjustments to actual costs rather than as amendments to the ceiling. This difference in treatment was necessary, according to the state, to avoid overstating some adjustments as its original methodology did. Pa@ 28 GAO/IKZD4lO44 Wa&ingt.on’r Family Independence Program Chapter 2 State’s Estimating Methods Show FIP Is Co& Neutral Eknefit and The final step in determining compliance with the food stamp cost-neu- Administrative Costs tral requirement, for both the state and FNS,is to sum the results of the separate cost-neutral calculations for benefits and administrative Combined expenditures. The state’s calculations for the first 3 quarters of the FIP project showed that it was under the overall food assistance cost-neutral ceiling by about $2.6 million, as summarized in table 2.4. Because FNSdid not calculate the state’s actual performance regarding administrative cost neutrality, it did not determine conclusively whether the state was in overall compliance with the cost-neutral requirement in the author- izing legislation. Table 2.4: Washington State’s Cost- Neutral Calculation8 for Food Assistance Costs and ceilings Benefits Administrative Total Benefit and Adminirtrative Costs Actual costs $139,654,036 $5,645,280 $145299,316 Combined Federal maximum (ceiling) 141,668,067 6,181,579 147,849,646 Amount over (under) cost-neutral ceilinaa $(2.014.0311 Sf536.299) $12.550.330) aOn August 11, 1989, the state submitted to HHS benefit cost data for the first 3 quarters of FIP opera- tion that contained two changes from the food stamp cost data that we analyzed and discussed in this report. First, benefit cost-neutral calculations for the third quarter were revised from those shown in table 2,l by the following amounts because the state used a different set of food assistance expendi- ture records that provided greater consistency with the data used for the first 2 quarters: total costs- $227,574-decrease; federal maximum-$7,604-increase; difference (under cost-neutral maximum)- $235,178increase. The totals for the 3 quarters combined changed by like amounts. Second, the state recalculated all benefit amounts to exclude one welfare location, Pasco, from the cost-neutral calcula- tions The state had proposed this change on April 13, 1989, because of the unique employment condi- tions relating to the Hanford nuclear facility near Pasco; HHS approved the exclusion on June 20, 1969. Together, the two changes resulted in the state’s reporting that for the first 3 quarters, it was $2,262,057 below the food stamp benefit cost-neutral ceiling (rather than the $2,014,031 shown in this table and discussed in the report); $439,601 below the administrative cost-neutral ceiling (rather than $536,299); and $2,701,658 below the combined benefit and administrative cost-neutral ceilings (rather than $2,550,330). Page 24 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 3 Alternative Estimating Methods May ISetter Ensure Benefit Cost Neutrality Our analysis revealed several questionable assumptions in the state’s method for estimating the anticipated value of the coupons that would have been distributed under the traditional Food Stamp Program. These assumptions may undermine the accuracy of the current method of assuring compliance with the cost-neutral requirement of FIP. In addi- tion, we found several problems with the benefit cost-neutral calcula- tions made by the state and FNSfor the first 3 quarters of FIP operation. Because all methods of cost estimation carry inherent risks of inaccu- racy, the state and federal goal should be to develop and use a method that results in the least error. The use of an alternate method for deter- mining the benefit cost ceiling and changes in procedures for making the calculations could result in more accurate cost-neutral determinations. The control sites were originally chosen, along with a matched group of Benefit Cost-Neutral treatment sites, for program evaluation purposes. Overall, the method Methodology Based on used by the state and federal agencies to randomly select the control and Questionable treatment sites for FIP program evaluation purposes was valid. We do, however, question several assumptions underlying the decision to use Assumptions cost data from the seven control sites for cost-neutral determination purposes. The following assumptions were made: . That the proportion of statewide food stamp costs represented by the seven control sites was sufficiently stable in the past to justify using those sites to accurately estimate statewide costs in the future. . That the relationship between food stamp costs in the control sites and food stamp costs in the state as a whole will remain, for the &year period of the demonstration, either constant or within a small and acceptable range of the 14.31-percent of statewide food costs incurred by the seven control sites in the 1987-88 baseline year. . That FIP would be implemented at all of the state’s local welfare offices during the demonstration, thereby establishing the need to estimate food stamp costs for the entire state. We present our analysis of these assumptions in this chapter, together with three examples of alternative estimating methods, which may pro- vide greater assurance that cost neutrality will be achieved. Page 26 GAO/RCED-90-94 Washington’s Family Independence Program Chapter 3 Altematlve Estimating Methods May Better Ensure Benefit Cost Neutrality Assumption One: Control An essential condition in selecting a small number of sites to use in accu- Sites’ Prior Costs Were rately projecting future costs at a larger number of sites is that the rela- tionship between costs at the two groups of sites be shown to be Stable in Relation to acceptably stable over an appropriate period of time. In the case of FIP, Statewide Costs the state and federal agencies agreed on a l-year pre-wp baseline period of April 1987 through March 1988 as a representative period for estab- lishing the food stamp ceiling calculation percentage. State data for that period showed that the seven control sites represented 14.31 percent of the state’s total food stamp expenditures, which then became the pro- portion used to estimate statewide food stamp expenditures in the future. In our opinion, however, a longer baseline period should have been used to better assess the stability of food stamp costs at the designated con- trol sites. FIP was designed to operate for 6 years into the future. A base- line period of 6 years prior to implementation would show whether the costs at the seven control sites were sufficiently stable over time, as a proportion of statewide costs, to justify their use in estimating statewide costs for 6 years into the future. Our review of the stability of costs over the S-year pre-FIp period from April 1983 through March 1988 showed that the proportion of the state’s food stamp costs represented by the control sites varied by slightly over 1 percentage point, from 16.76 percent to 17.86 percent (see table 3. l).’ In absolute terms this does not appear to be a large vari- ation. However, because the control sites’ food stamp expenditures are divided by .1431 to establish the statewide ceiling, a similar 1 per- centage point difference between the proportion used in the ceiling cal- culations and the actual food costs could result in an annual cost ceiling that could be $12 million too high (if the actual proportion turns out to be 16.31 percent instead of 14.31 percent) or $14 million too low (if the ‘The difference between the 16.76 to 17.86~percentrange of control site costs shown in table 3.1 for the S-year period and the 14.31 percent calculated by the state for the baseline year and discussed and analyzed in this report is due to data limitations because of a geographic boundary change. During May and June 1986, one of the seven control sites, Pierce West, was split into two offices, Pierce North and Pierce West. Because the state could not isolate for us the food stamp costs attribu- table to the Pierce North portion of Pierce West prior to the 1986 split, we had to include in our 5- year analysis the costs from Pierce North after the split in order to permit us to compare consistent data over the S-year period. Therefore, the percentages in table 3.1 for the year April 1986-March 1987 and the baseline year of April 1987-March 1988 are based on costs at eight offices rather than seven and are about 3.5percent higher than the state’s calculation. However, having to add the addi- tional welfare office to permit a consistent comparison for the &year period does not affect the results of our analysis because state records show that Pierce North’s food stamp costs were as stable as food stamp costs at all of the other control sites. Page 20 GAO/RCED-90-84 Washington’s Family Independence Program . Chapter 8 Altmnative Errtimatjng Methods May Better Enmre Beneflt Co& Neutrality actual proportion turns out to be 13.31 percent instead of 14.31 per- cent). Because of the “multiplier effect” involved, the impact of even small errors or variations in the ceiling percentage can be considerable.2 Several examples of alternative methods for setting cost-neutral ceilings that reduce the multiplier effect and could result in more accurate calcu- lations of cost neutrality are described later in this chapter. Table 3.1: Proportion of State Food Stamp Costs Represented by the Seven Control site total Control Sites Contrc&sii; as a percentage Year a State total of state total Amil1987-March 1988 $29,499,199 $165.134,288 17.86 Abril1986-March 1987 27,130,807 153;160;105 17.71 April1985-March 1986 23,885,375 141,997,645 16.82 kpril1984-March 1985 22,780,904 135,994,703 16.75 April 1983-March 1984 23,411,628 136,841,914 17.11 %cludes the Pierce North portion of the Pierce West control site, which was split off in May and June 1986. The state could not isolate food stamp costs attributable to the Pierce North portion prior to the split. Assumption Two: Control The state’s assumption that food stamp costs in the seven control sites Sites’ Future Costs Will Be will remain a stable percentage of statewide food stamp costs for the 5- year FIP period also carries inherent risks of errors. The factors that Stable in Relation to influence food stamp expenditures in any particular local welfare office Statewide Costs are numerous, complex, and often cannot be predicted. If major eco- nomic, demographic, or other changes occur in any of the control sites that do not occur in the rest of the state, or vice-versa, the proportion of food stamp costs represented by the control sites may no longer accu- rately reflect statewide costs, and these changes could affect the deter- mination of cost neutrality. For example, the opening of a large industrial plant in one of the control sites could cause food stamp expenditures in that site to decrease, while food assistance expenditures in other sites around the state may be gradually increasing. The reduc- tion of costs in the control site would lower the statewide benefit ceiling. In turn, this lower ceiling would reduce the amount of federal food assis- tance funding the state could receive, even though it may not be spending any less on food stamps and FIP food cash statewide than it would have spent on food stamps alone. “A “multiplier” of nearly 7 occurs because control site costs are divided by .1431 (i.e., multiplied by 6.988) to estimate statewide costs. The “Alternative Methods” section, discussed later in this chapter, provides additional information on the multiplier effect and its consequences on cost-ceiling accuracy. Page 27 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 3 Alternative Estimating Methods May Better Ensure Benefit Cost Neutrality It is not possible to ensure conclusively that food stamp costs in the con- trol sites will continue to represent a stable percentage of the state’s total food stamp costs in the future. Therefore, we believe it is impor- tant that, if the state continues to use a ceiling calculation percentage, such as the 14.31 percent being used at the time of our review, the per- centage needs to be adjusted whenever significant economic, demo- graphic, or other changes that affect food stamp expenditures occur in any of the control sites that do not occur in the rest of the welfare sites, or vice-versa. Currently, there are no program criteria that define what constitutes an economic or other change that is significant enough to warrant an adjustment to the cost ceiling calculation or an agreement between the state and the federal agencies regarding the circumstances under which such adjustments should be made. In our opinion, such adjustments to offset the impact of any major change in conditions affecting food stamp expenditures are necessary to help ensure that the cost-neutral ceilings continue to reflect as closely as possible what costs would have been in the absence of FIP. To facilitate making any necessary adjustments, the state and the federal agencies need to agree on the criteria for making adjustments and to be alert to major events, as well as to monitor various types of indicators over time for both control and noncontrol sites, so that they can detect any unique changes occurring at one or more sites. Examples of some possible rele- vant indicators include unemployment rates, population growth rates, and changes affecting major employers in a local area. Assumption Three: FIP As originally designed, FIP was to have been implemented in all state Will Be Implemented welfare offices, except the control offices, in the first 3 years of the demonstration, Because the converted offices would no longer have a Statewide food stamp coupon program for their FIP clients after they implemented FIP, it was necessary to develop a method to estimate what coupon costs would have been for welfare recipients in all of those sites. On June 2, 1989, however, Washington State limited the number of offices allowed to convert to FIP to only 20 of the state’s 65 local offices. With this major change in implementation, we question the need to con- tinue to estimate food stamp expenditures for the entire state. Estimates now only need to be made for the 20 offices, or approximately 30 per- cent of the state’s welfare sites, which are operating under FIP. The remainder of the state’s 45 offices, those that will not convert to FIP, do not need an estimate or approximation of the food stamp ceiling estab- lished for them. Because they will continue to operate a food stamp Page 28 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 3 Alternative Estimating Methods May Better Ensure Benefit Cost Neutrality coupon program and their actual food stamp costs will be known, the ceiling for the 45 non-FIP sites can be established “at cost” (i.e., at actual costs incurred). We believe the shift in the state’s implementation plan is sufficient reason for the state and the federal agencies to review their current methodology for calculating food stamp benefit ceilings and, if appropriate, to adopt an alternative method that results in more accu- rate cost-neutral determinations. Given the fact that FIP will no longer be implemented statewide and that Alternative Methods small variations in the proportion of the state’s costs represented by the Exist for Setting control sites can have a substantial impact upon the cost ceiling esti- Benefit Ceilings mates, we believe that alternative methods for establishing food stamp cost ceilings need to be considered. A variety of methods could be used to estimate benefit ceilings. We describe and analyze only a few in this report. The alternatives that we chose as examples are similar to the state’s current method in that they also use a ratio of costs in two groups of sites to estimate a cost ceiling. However, with our alternatives the cost ceiling would be estimated only for the sites where FIP is implemented, rather than, as presently, for the entire state, The cost ceiling for the non-Fm sites would be established “at cost.” We considered the following: l Example 1. Use costs in the 7 current control sites to establish a ceiling for the 20 FIP sites only. l Example 2. Use costs in the 19 paired non+m sites to establish a ceiling for the 20 FIP sites only.” . Example 3. Use costs in all 46 non-Fm sites to establish a ceiling for the 20 FIP sites only. Although all estimates will contain some amount of error, a significant advantage to using any of the three examples is that the estimates are made for a smaller number of sites, and thus the chances of errors are reduced. For example, any error or other cost variance in the calculation proportion is only applied to costs at 20 sites, rather than at 66 sites as in the current statewide method. Because estimates are made for a smaller number of sites, all three alternatives reduce by half or more the adverse “multiplier effect” that results when errors or other cost devia- tions occur in one group of sites that do not occur in the other sites. “The 19 non-FIP sites were paired with the 20 F’IP sites during the design of FIP, as described in chapter 2 and later in this chapter. Page 29 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 3 Altemnntive Estimating Methods May Better Ensure Beneflt Cost Neutrality Under the current method, costs that are incurred in control sites are divided by ,143l to estimate statewide costs, which results in a multi- plier effect of nearly 7. Examples 1, 2, and 3 reduce the multiplier effect to approximately 2.6, -9, and -6, respectively.4 As shown in table 3.2, these lower multipliers can significantly reduce the impact of any errors or localized changes on the overall ceiling calculations. For example, as previously discussed, there was a l-percentage point variation in the percent of statewide costs represented by the seven con- trol sites over the 5-year period immediately preceding PIP implementa- tion There is no way to determine whether a similar variation will occur in any year during the FIP demonstration period. If it does, however, under the current method the federal government could contribute $12 million more than it should under the concept of cost neutrality, or the state could unfairly be denied $14 million. Smaller or larger variances than l-percentage point would have proportionally smaller or larger dollar impacts’. Table 3.2 shows the effect of a l-percentage point varia- tion under the current method and under each of the three examples we examined. As the table also shows, when cost estimates are made for only 20 sites, the potential additional cost to the federal government or penalty to the state resulting from any variation or inaccuracy in the estimation technique are significantly less than when costs for the entire state are estimated. ‘Multiplier effects are determined by dividing 1 by the divisors shown in table 3.2, as follows: For the current method, 1 divided by .1431 = 6.988, or a multiplier of about 7. For example 1, 1 divided by .3862 = a multiplier of about 2.6; for example 2, 1 divided by 1.1438 = a multiplier of about .9; for example 3, 1 divided by 1.6997 = a multiplier of about .6. Page 30 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 3 Altemtive E&mating Methods May Better Ensure Benefit Cost Neutrality Table 3.2: impact of a 1 Percentage Point Variation in Costs on Cost-Ceiling Estimates Oollars ~. ---- in thousands ~- if baseline year percent If baseline year percent used is 1-percentage used is 1-percentage Ceiling estimate point too low point too hiah using baseline Ceiling Additional cost to Ceiling Additional cost Method of caicuiatlon year proportiona estimatea federal government estimate’ to state Current .----~-“- method Using 7 control sites’ costs to estimate statewide costs (divisor=.1431)” $186,882 $174,676 $12,206 $200,923 $14,041 Example 1 Using 7 control sites’ costs to estimate FIP sites’ costs (divisor= .3862Y’ 69,246 67,498 1.748 71.087 1,841 Example 2 Using 19 non-FIP sites’ costs to estimate FIP sates’ costs (divisor-l .1438Y’ 68.256 67.664 592 68.858 602 EXamDie 3 Usina 45 non-FIP sites’ costs to estimate FIP sites’ costs (divisor= 1.6997Y’ 67.886 67.489 397 68.287 401 aFor the current method, the ceiling estimates are statewide ceilings for all 65sites. For examples 1, 2, and 3, the ceiling estimates are for the 20FIP sites only; statewide ceilings are determined by adding actual costs incurred at the 45 non-FIP sites. bFor the current method, the state calculated the .1431 divisor, or baseline year proportion, which it currently uses to establish the food stamp ceiling each quarter, by dividing food stamp expenditures in the 7 control sites during the 1987-88 baseline year ($23,624,198) by food stamp expenditures at all 65 state welfare sites for that same period ($165,134,288). Similarly, for examples 1, 2, and 3, we calculated the divisors, or baseline year proportions, by dividing food stamp expenditures in the 7, 19, and 45 non- FIP sites, respectively, during the baseline year ($23,624,198; $69,963,897; and $103,966,834, respec- tively) by food stamp expenditures in the 20 FIP sites for that same period ($61,167,454). In addition to reducing the potential effects of errors in the estimated ceilings, it is also possible that example 2 could produce more accurate cost ceiling estimates because it uses costs from 19 non-Frp sites to esti- mate a ceiling for the 20 FIP sites with which they were matched during the FIP design period. As described in chapter 2, the state’s welfare office sites were matched in pairs by the state and the overseeing fed- eral agencies according to a set of 11 measures of program and geo- graphic similarity. Because the two sets of sites were carefully matched on a variety of programmatic factors, costs in the 19 non-HP sites might be more representative of costs in the 20 FIP sites than costs in the 7 control sites currently being used. Because we examined only 1 year’s data and limited our analysis to only three potential alternatives for more accurately determining cost-neu- tral ceilings for benefits, we cannot conclusively recommend that the Page 31 GAO/RCED-30434 Washington’s Family Independence Program Chapter 3 Alternative Estimating Methuds May Better Ensure Beneflt Cost Neutrality state and the federal agencies use example 2 or the other two alterna- tives we explored. However, we do not believe there is a need for the state to continue to use the seven control sites, or any other number of sites, to estimate costs for the entire state. Because the state has limited the number of sites that will be implementing FIP to 20 of the state’s 66 sites, actual food stamp costs in the remaining 46 sites will be known and need not be determined by an estimate. This change should allow the state and the federal agencies to limit their estimates, and therefore the impact of estimation errors or other cost variations, to only those 20 FIP sites for which food stamp coupon costs are unknown and therefore need to be approximated. In addition to examining the current method for estimating the costs of State and Federal food stamp benefits at FIP sites, we reviewed the state’s and FNS’cost- Quarterly Calculations neutral calculations for the first 3 quarters of FIP operation. Both sets of Differed calculations showed the state well below the cost ceilings for the 9- month period. The calculations did differ in nearly every respect, how- ever, as shown in table 3.3, and these differences affected the calcula- tions of the amounts by which the state was under the cost-neutral ceilings. The differences occurred because the state and FNSdid not always use the same data sources, data, or calculation techniques. Page 32 GAO/RCED-90.84 Washington’s Family Independence Program Chapter 3 Alternative EMimdng Methods May Be.tt.er Enenre Bene!lt Cost Neutrality Table 3.3: Comparison of the State’8 and FNS’ Beneflt CowNeutral Calculations Quarter ended g/30/88 state FNS bitt erence’ for the Flrbt 3 Quarters of FIP Operatlon Federal maximum $44,560,566 $44548,295 $(12,271) Food Stamps 43,105,932 43,342,647 236,715 FIP Food Cash 484,243 343,317 (140,926) Total food costs 43,590,175 431685,964 95,789 Amount over (under) ceiling (970,391) (882,331) (108,080) Quarter ended 12/31/88 Federal maximum 47.517.140 47504.032 (13.108) Food Stamps 44,410,748 ~44,626,509 215,761 FIP Food Cash 2,490,184 2,229,080 (261,104) Total food costs 46,900,932 46,055,589 (45,343) Amount over (under) ceiling (818,208) (848,443) 32,235 Quarter ended 3/31/89 Federal maximum 497590,361 49.584.284 (6,077) Food Stamps 44,210,182 44,210,182 0 FIP Food Cash 4,952,747 4,764,137 (188,610) Total food costs 49,162,929 48,974,319 (188,610) Amount over (under) ceiling 427,432) (809,985) 182,533 Total over (under) ceilina for 3 auarters $(2.014.0311 $12.120.7391 $106.708 aAmounf by which FNS’ calculation is over (under) state’s calculation. State and FNS Used First, the cost calculations differed because the state produces several Different Data Sources sets of accounting records from which food stamp and FIP food cash expenditure data can be drawn. The records differ because adjustments and corrections of errors are made to some sets of records and not to others. Because the state and FNSused different sources of state data in making their cost-neutral calculations, they arrived at different dollar results when applying the same methodology. For food stamp coupon issuances, the figures used by the state and FNSdiffered by an average of $70,000 a month; for FIP food cash expenditures, they differed by an average of $60,000 a month. Our analysis showed that the figures used by FNSfor food stamp expenditures have been adjusted for such factors as lost or stolen coupons and cashier’s issuance errors and differ from the unadjusted data used by the state. For FIP food cash expenditures, the figures used by the state differ from those used by FNSbecause they are generated by a system that takes into account terminations and new applications which occur throughout the month. To avoid inconsisten- cies in the cost-neutral determinations, it is important that the state and FNSuse the same sources of state data for their calculations and that these sources provide the most accurate data available. Page 33 GAO/RCED-90.34 Washington’s Family Independence Program Chapter 3 Alternative E&hnating Methods May Better Ensure Benefit Cost Neutrality It is equally important that the data used be internally consistent. For example, if adjusted figures are used for statewide food stamp and FIP food cash expenditures, as discussed above, adjusted data should also be used for food stamp expenditures at the control sites. During our review, neither the state nor FNSwas using adjusted data for the control sites because, according to state accounting personnel, it was not being generated. Because adjusted data tend to be higher than corresponding unadjusted data, the use of adjusted figures for statewide costs while using unadjusted figures for the control sites can work to the detriment of the state because it results in lower cost-neutral ceilings, which in turn could mean less federal funding.” Our analysis of available cost data for the first 2 quarters of FIP opera- tion showed that, if food stamp benefit costs in the control sites had been adjusted by the same percentage as the state used in adjusting statewide food stamp benefit costs (a reasonable assumption, according to state accounting personnel), the statewide benefit ceilings would have been increased by about $246,000 and $231,000, respectively. Because the state’s and FNS’ calculations for the first 2 quarters of FIP (see table 3.3) showed that the state’s actual food assistance benefit costs were below the cost-neutral maximum, the use of unadjusted cost data for the control sites did not affect the amount of federal funding it was eligible to receive for those 2 quarters. However, the difference could become important in future quarters if the state’s reported benefit costs exceed the amount of the benefit ceiling. On July 7, 1989, after we brought these calculation inconsistencies and weaknesses to JTNS attention, it formally notified the state that the two parties must use a consistent set of data to ensure consistent evaluations of cost neutrality. In its letter to the state, FNSprescribed the specific sets of state records that should be used in the future. Also, effective August 1, 1989, FNSdirected the state to submit and use adjusted food stamp cost data for the control sites to avoid inadvertently reducing the ceiling and therefore the amount of federal funding available to the state. On August 22 and 29, 1989, the state informed FNSthat although it was modifying two state financial reports to ensure more consistent and compatible data, it was unable to adjust food stamp cost data for “Because control sites’ food stampcosts are divided by .1431 to establish the statewide ceiling, the use of lower, unadjusted control site cost data results in lower ceilings. For example, if unadjusted control site costs were $1 million, the statewide ceiling would be $6,988,120 ($1 million divided by .143 1). If adjusted control site costs were, say, $1.1 million, the statewide ceiling would be $7,686,932 (b 1.1 million divided by .1431), or a ceiling increase of nearly $700,000 due to the $100,000 adjustment. Page 34 GAO/RCED-90-34 Waahingtm’s Family Independence Program chapter 8 AlterMtiveRetimadngMethodsMay Better Enemm Renedlt Cat Nemtdity the control sites. Therefore, according to the state, it intended to use different data for some cost elements than that prescribed by FNS.Our analysis of both FNS’and the state’s proposals showed that, if agreed to and used consistently by both the state and FNS,either set of records should eliminate or minimize the data inconsistency problems we found and help ensure that quarterly cost-neutral calculations are made using the best available, most internally consistent data. State Sent Some Erroneous Second, the state’s and FNS’calculation differed because some of the cost Data to FNS data sent to FNSby the state were incorrect. For the third quarter of PIP operation, the state’s social and health services agency erroneously sent FNSinformation on food stamp costs in the control sites that understated actual costs by approximately $103,000. As a result, FNS’cost-neutral calculation for the quarter showed that the state was over the cost-neu- tral ceiling by approximately $113,000. After we brought the error to FNS’attention, it obtained the correct data from the state and revised its calculations. The revision showed the state to be under the ceiling by approximately $610,000.” The large difference in FNS’ two calculations (approximately $723,000)7 that resulted from a $103,000 error was due to the multiplier effect dis- cussed earlier. Because control site cost data are divided by ,143l (i.e., multiplied by a factor of approximately seven) to arrive at an estimate of the state’s total costs, any error in control site cost data is magnified at the statewide level by a factor of nearly seven. Because of the multiplier effect, and the potentially large impact on cost neutrality of even small errors or variances in control site cost data, it is important that control site data be as accurate as possible. We believe that problems such as the error we found could be minimized, and internal control over the gathering and reporting of data could be strengthened, if FIP personnel were required to verify all FIP-related cost “In its technical comments on a draft of this report, FNS indicated that its staff had independently discovered the state’s $103,000 understatement error discussed here. That is not correct. We brought this error to FNS’ attention, and its representative told us that FNS was not aware of it. The error that FNS is referring to is another error in which the state omitted data for one control site from one of its reports. Unlike the $103,000 error, however, the latter error did not affect FNS’ calculation of the state’s cost-neutral status because FNS discovered the error before it made its calculation. 7The dollar difference between FNS’ initial calculation showing the state to be $113,000 over the ceiling and its revised calculation showing the state to be $610,000 under the ceiling is $113,000 + $610,000, or $723,000. Page 35 GAO/RCED-90-84 Washingtan’s Family Independence Program .,, ‘. chapter3 Alternative Estimating Methods May Better Ensure Benefit Cast Neutrality data before they are released from the state and forwarded to the fed- eral agencies. State and FNS Used In addition to these two problems, we found a third discrepancy that Different Calculation accounted for a small portion of the differences between the state’s and FNS’cost-neutral calculations for benefits. Although FNS,in calculating Techniques the benefit ceiling, was dividing the seven control sites’ costs by exactly 14.31 percent, the state was using a computer spreadsheet program that carried the decimal out to several more places, resulting in a benefit ceiling from $6,000 to $12,000 higher per quarter than FNS'. If the state’s reported actual costs come closer to the benefit ceiling in the future than they did for the first 3 quarters of FIP operation, the state and FNSwill need to decide which of the two calculation techniques they find most accurate and/or appropriate to use. Several actions need to be taken to better ensure that food stamp benefit Conclusions cost ceilings will more accurately represent what actual food stamp expenditures would have been if FIP had not been implemented. First, because FIP will no longer be implemented statewide, there is no longer a need for the state to project or estimate what food stamp ben- efit costs would have been statewide. Opportunities exist for exploring alternative methods for setting the ceiling that could reduce the risk of estimation errors. Because even small variations in cost data from the control sites can have substantial impacts on the cost ceiling under the current method, other methods need to be examined. This examination should include methods, such as those examples discussed in this chapter, that estimate benefit costs only for those sites at which FIP is implemented. Second, if non-FIP sites continue to be used as a benchmark to estimate costs for other sites, the cost ceiling calculation percentage needs to be adjusted as necessary to reflect significant changes that occur in one or more of the non+m sites but not in the FIP sites, or vice-versa. To do this, the state, FNS,and HHSneed to agree on criteria for making such adjust- ments, aa well as to monitor various economic and other indicators over time. These actions will allow them to be aware of events and/or condi- tions that could have an impact on food stamp expenditures in any of these sites, and thus affect cost-neutral calculations. Page 36 GAO/RCED-66434 Washington’s Family Independence Program chapter3 Alternative Estimating Methods May Better Ensure Benefit Cost Neutrality Regardless of the method used to determine the cost-neutral ceiling, cost-neutral calculations need to be made with the most current, accu- rate, and internally consistent data available. Strengthening the process by requiring FIP office personnel to verify all np-related data before they are released from the state and sent to the federal agencies could also reduce the likelihood of errors that might have significant impact on cost-neutral determinations. We recognize that FIP faces the inherent problems associated with the Recommendations demonstration of any new welfare assistance approach. However, to better ensure the program’s cost neutrality, we recommend that the Sec- retaries of Agriculture and Health and Human Services direct their responsible agencies and the state of Washington to consider alternative methods for calculating the food stamp benefit cost ceiling under the FIP demonstration. They should consider approaches such as those that esti- mate benefit costs only for sites at which FIP is implemented and adopt the method that appears to provide the greatest assurance of cost neutrality. In addition, if any number of non-FIP sites continue to be used to esti- mate benefit costs for other sites, we recommend that the Secretaries of Agriculture and Health and Human Services direct their responsible agencies and the state of Washington to . agree on the criteria for adjusting the cost-neutral calculation data for food stamp benefits when conditions affecting food stamp expenditures, and thus cost-neutral calculations, change significantly in one or more of the non-FIP sites but not in the FIP sites, or vice-versa; l monitor economic, demographic, and other indicators at both FIP and non+m sites in order to be aware of major events and/or changes in con- ditions that meet the agreed-upon criteria and warrant adjustments in the data used for calculating the benefit’s cost neutrality; and l adjust the data used to calculate the benefit’s cost neutrality to help assure accurate cost-neutral determinations, whenever conditions affecting food stamp expenditures, and thus cost-neutral calculations, change significantly. We further recommend that the Secretary of Agriculture direct the Administrator, FNS,to work with the state of Washington to l agree on the best available, most internally consistent data to be used for making quarterly cost-neutral calculations and Page 37 GAO/RCEJHO-84 Washington’s Family Independence Program chapter a AItemative JietinWing Methods May Better Eusure Benefit Coat Neutrality . require the FIP office to verify all np-related cost data before they are released from the state and sent to the federal agencies. We received written comments on a draft of this report from FNS,HHS, Agency and State and Washington State. (See apps. I, II, and III.) They generally agreed Comments and Our with our report’s findings and conclusions and indicated that these Evaluation would be helpful to them in exploring and negotiating various options and changes to FIP. Washington noted that because neither federal nor state representatives had significant experience in developing cost-neu- trality provisions, it is not surprising that even after 18 months of FTP operation, issues remain to be resolved. We agree that developing and implementing any innovative welfare assistance approach, especially one such as FIP that includes federal cost-neutrality requirements will undoubtedly raise questions and issues that take time to negotiate and resolve. FNSnoted that food stamp benefits and administrative expenses make up only a small part (less than 13 percent) of the total federal and state expenditures for PIP. Although food assistance is only one portion of FIP, our findings and conclusions should be applicable to the larger AFLE and Medicaid components of FIP because the cost-neutral requirements and calculation procedures for those programs are the same as, or similar to, those for the Food Stamp Program component. Washington State agreed, indicating in its response that many of the issues discussed in the report have relevance beyond the Food Stamp Program. We also believe that the report should be useful in developing and administering any similar welfare reform programs in other states. Comments from all three respondents relating to the clarity and tech- nical accuracy of specific statements in the draft report have been incor- porated in the report where appropriate. In addition, the three organizations provided the following comments and observations on the development and implementation of FIP’S cost-neutrality provisions. Food and Nutrition Service FNSagreed with our conclusion that a reasonable approximation of pos- and Department of Health sible program costs is probably the best that can be achieved, despite the legislative requirement for an “assurance” of cost neutrality. Both and Human Services FNSand HHS pointed out that the benefit cost-neutral methodology was Comments I developed at a time when plans called for FIP eventually to be expanded statewide and that it was reasonable for them and the state to have relied on those plans as an underlying assumption in developing the Page 88 GAO/RCED-90-84 Wa&ngt.on’s Family Independence Program Alternathe EBtLmating Method8 May Better Eneure Benefit Cmt Neutrality methodology. FNSalso believes that the use of the seven control offices for cost-neutral determinations was appropriate at the time because (1) the offices, chosen for program evaluation purposes, were representa- tive of the state as a whole, and (2) prior period cost stability could not be tested beyond 1 year because cost data could not be readily isolated for the control sites for longer than 1 year, HHS stated that the best methodology for establishing cost-neutral ceilings would have been through the use of a representative random sample of AFM: recipients throughout the state, but because this option was not available, federal and state officials instead chose a representative sample of local welfare offices to generate the statewide ceiling. While we recognize throughout the report that the cost-neutral method- ology was developed, in part, based on a “statewide implementation” assumption, we do not agree that the use of the chosen control sites was necessarily appropriate for cost-neutral determination purposes just because they had been previously selected for statewide program evalu- ation purposes. We also do not agree that food stamp costs could only be isolated for the control offices for 1 year preceding the implementation of FIP. As stated in chapter 2, an essential condition in using costs at a small number of control sites to project future costs accurately for the state as a whole is the existence of demonstrated prior-period cost sta- bility between the control sites and the state as a whole. We believe that our analysis, in which we were able to examine cost relationships for a 5-year pre-wp period, raises questions on the reasonableness of the con- trol offices selected, We found that a variation in the stability of control site costs as little as plus or minus l-percentage point could result in a statewide food stamp cost ceiling that could range from as much as $12 million above cost neutrality to $14 million below. In its response, the state of Washington indicated that in the haste to get FIP underway, neither federal nor state representatives thoroughly reviewed all of the implications of the methodology that was established, including exam- ining the stability of costs in control sites “for even 1 year prior to the beginning of FIP, let alone for 6 years as presented in the draft report.” The state therefore concluded that an actual error of more than plus or minus 1 percent would probably be expected to occur as the result of the type of sampling methodology used to establish the baseline. Despite some differences in views from ours on the reasonableness of the current methodology at the time it was developed, both FM and HHS agreed that the current methodology should be reevaluated and, if appropriate, replaced with a method that provides greater assurance of cost neutrality. However, FNSand HHS questioned whether the examples Page 39 GAO/BCED-90-94 Wasldngton’e Family Independence Progran~ Chapter 3 Alternative Estimating Methods May Better Ensure Benefit Cost Neutrality we cite in the report as calculation alternatives will result in a more accurate ceiling than that based on the current methodology. Although it did not provide any details or support for its contention, FNSsaid that the use of any of our calculation alternatives might introduce new sources of error or bias. We clearly label the examples as merely illustra- tive of a variety of methods that could be used to establish cost ceilings. We believe that FNSand HHS,in conjunction with the state, need to assess the available cost data on FIP based on nearly 2 years of program opera- tion and, by exploring various methodology options, determine whether an alternative can be developed that provides greater assurance than the current method does that food stamp benefit cost ceilings will represent what actual Food Stamp Program expenditures would have been if FIP had not been implemented. FNSagreed that it would be desirable to make adjustments for the effects of changes outside the Food Stamp Program, provided that these changes can be identified as affecting the program and the magnitude of their effects can be estimated. It said that it would entertain any specific suggestions from the state about the types of indicators and the extent of change that should be considered relevant to adjusting the method- ology. Accordingly, we believe that the state, FNS,and HHSshould seek to agree as soon as possible on criteria for monitoring and making such adjustments and that the state should adjust its calculation data when- ever the agreed-upon triggers occur. State of Washington The state’s Department of Social and Health Services and the Employ- Comments ment Security Department, in a joint response, said that for the most part they agree with the conclusions in the draft report. They believe, however, that the report does not cover broader issues that must be examined and resolved before a final consensus on cost neutrality for either the Food Stamp Program, AFDC,or Medicaid portions of FIP can be reached. These issues, which for the most part arose after we had com- pleted our field work, are summarized below and described in more detail in section A of the state’s response in appendix III. First, the state’s welfare system has experienced a significant growth in caseload. In November 1989, after reviewing the first full year of FIP operations, the state sought approval from USDAand HHSfor several changes in the methodology for determining cost neutrality. The state noted that changes were needed in the wording and interpretation of the special terms and conditions governing FIP to address several new issues that had arisen since the beginning of FIP operations and that would Page 40 GAO/RCED-90-84 Washington’s Family Independence Program Chapter I.3 Altmnative Estimating Methods May Better Ensure Benefit Cost Neutrality make the cost-neutral methodology better conform to the original intent of the Congress. The most important of these, according to the state, was a request for changes in the special terms and conditions that would require federal participation in a portion of the cost of providing wel- fare benefits to new FIP participants who, in the absence of FIP, would have been eligible to participate in the regular Food Stamp, AFDC,and Medicaid programs. The state points out that since its inception, FIP has experienced an unanticipated increase in caseload, that the great majority of these additional participants would have been eligible under the regular programs, and that both state and federal legislation require that the state serve them. The state also believes that the Omnibus Budget and Reconciliation Act of 1987, which authorized FIP and speci- fies the program’s cost-neutrality requirements, unambiguously requires federal participation in some portion of the cost of the caseload growth the state is experiencing. In January 1990, according to the state, HHSturned down its request concerning federal participation in the cost of caseload growth. The state and HHShave had some direct negotiations in an attempt to resolve the issue, but as of the state’s March 16, 1990, response to us, the issue remained open. The state noted that if it prevails in its position, either through negotiation or an appeal to the courts, the required changes to the project’s special terms and conditions would be so extensive as to require opening the discussion of cost-neutrality methods considerably beyond the parameters found in our report. The state also noted that if it does not prevail in its position, the state’s liability for the additional cases under the cost neutrality provisions may be so great as to threaten the continuation of the FIP demonstration. Second, the state has encountered problems with the accounting method for the welfare costs of FIP participants who migrate, or move perma- nently, from F'IPsites to control sites. According to the state, in December 1989 an HHSrepresentative denied a request by the state that it be allowed to include the welfare costs of FIP participants who migrate to control sites as part of the control sites’ costs for purposes of calcu- lating Food Stamp, AFIX, and Medicaid cost-neutral ceilings, The state indicates that because, as we point out in the report, the cost of each case in the control sites is multiplied by a factor of approximately seven to calculate the statewide ceiling, the exclusion of even a few of these families from the cost calculation works to the state’s financial disad- vantage. As of December 1989, according to the state, about 690 FIP cases had migrated to control sites. The state estimates that HHS' denial of its request for a change in the accounting method has resulted in Page 41 GAO/RCED-90-84 Washington’s Family Independence Program chapter3 Alternative Eetimaung Methods May Better Enaum Benefit C&et NeutraU~ accumulated costs not included in the projected cost-neutral ceiling of more than $2.9 million for the Food Stamp Program component of FTP and $14.3 million for the AFDCand Medicaid components after the first 6-plus quarters of FIP operation. The state has appealed the HHSrepre- sentative’s denial to the secretaries of HHSand USDAbut indicated that, as of March 16, 1990, it had received no formal response to its appeal. In addition, the state provided comments on several matters that were included in our report. Regarding the current method for estimating ben- efit cost neutrality, the state indicated that the several assumptions that we question in the report, including the stability of both prior and future costs in the control offices, were, in fact, assumptions. The state indicated that none of the assumptions or other implications of the methodology was thoroughly reviewed or examined in detail by either state or federal representatives involved in developing the cost-neutral methodology. The state further indicated that because the current meth- odology is designed to specifically exclude the federal government from participating in the costs of caseload growth, as discussed earlier, it believes that the methodology is more fundamentally flawed than we describe in our report. Further, the state indicated that each of the three examples we cite as methodology alternatives would increase the influ- ence of a methodological flaw related to how welfare cases migrating into FIP sites would be handled and would disadvantage the state even more than the current methodology. It did state, however, that if the problems of federal participation in the cost of caseload growth and migration are solved, then the use of some variation of the three exam- ples discussed in the report might reduce the potential cost-ceiling pro- jection error and improve the calculations. Regarding the differences in state and federal quarterly cost-neutral cal- culations for food stamp benefits, the state discussed in detail several technical points about its accounting system and data sources, and to its ongoing negotiations with the FNSWestern Regional Office on the calcu- lation of cost neutrality. It also explained why it believes that the FNS formula for releasing Food Stamp Program funding to the state is wrong, including its belief that FNS’method improperly mixes both adjusted and unadjusted data in the same computations. Because much of the negotiation between the state and FNSover the use of consistent data sources occurred after we completed our field work, as did several changes made by the state to accommodate instructions from FNSon which data sources to use, we did not examine the current status of the issue in depth. We continue to believe, however, that FNSand the state need to work together to agree on the best available, most internally Page 42 GAO/WED-SO-84 Washingtm’s Family Independence Program Chapter 8 Abmative lldlmatlng Methods May Better Jbumre Benefit Cat Neutrality consistent data to be used for making quarterly cost-neutral calculations. Finally, the state said that it would be useful for it to sit down with representatives of USDAand HHSto consider in more detail the questions that we raise in our report. We agree that such face-to-face discussions between representatives of the three parties is vital if the issues still confronting the operation of FIP are to be resolved quickly and amicably. We also believe that such discussions should occur so that the lessons learned from this demonstration project can be applied on future projects and programs. Page 43 GAO/RCEDSO-34 Washington’s Family Independence Program * Chapter 4 Cost Neutrality of A dministrative EkpensesIs’ Still Unclear Determining the costs associated with administering a new welfare pro- gram such as FIP is a complex task, and some problems in developing and implementing appropriate methods are to be expected, Our review indi- cated, however, that improvement is needed in both the methodology proposed by the state for determining administrative cost neutrality and in the timeliness of FNS’and HHS'response to the state’s proposal. Because of these problems, the extent to which food stamp administra- tive expenses are being kept within the FIP cost-neutral maximums cannot be determined conclusively. We examined the state’s May 1988 proposed method for determining the administrative cost ceiling and calculating cost neutrality and reviewed the state’s and FNS' calculations of cost neutrality for the first 3 quarters of FTPoperation.’ We found the following: l Although the state submitted its first proposed administrative cost methodology to FNSand HHSin early May 1988, it had still not received a formal response as of March 16, 1990. . The state’s proposed methodology embodies several questionable assumptions, some of which could disadvantage the state or the federal government if certain situations occur. . The state’s calculations of administrative cost neutrality were based in part on incomplete and otherwise preliminary data, and the results should be viewed with caution. l FNSwas not making certain required administrative cost-neutral calcula- tions and was therefore unable to determine conclusively whether the state was in compliance with the cost-neutral requirement of the author- izing legislation. On August 11, 1989, after we completed our field work but before FNS and HHShad completed their review of the state’s May 1988 proposal and informed the state of the problems they had identified, the state submitted a revised administrative cost-neutral methodology proposal to HHS.The proposal contains two alternate methodologies, both of which would correct some of the estimating weaknesses identified in FNS’and our reviews and would reduce the amount by which the state was under the administrative cost-neutral ceiling for the first 3 quarters of FIP operation compared with calculations made under the state’s orig- inal proposal. ‘Both the state’s method for determining administrative cost neutrality and the results of the state’s and FNS’ calculations are described in detail in chapter 2. Page 44 GAO/ItCEMJO-84 Washington’s Family Independence Program chnptm 4 Cuet Neutrality of Administrative &mwea Ia Still Unclear According to the project’s March 1988 special terms and conditions, the Federal Review of state had 60 days to prepare and submit proposals for determining food State Proposal Is Slow stamp, AFDC,and Medicaid administrative cost neutrality. The state sub- mitted the three proposals to USDAand HHSon May 3 and May 6,1988, respectively. FNSheadquarters and its Western Regional Office com- pleted their food stamp proposal analyses in August and October 1988, respectively, and they internally identified several problems with the methodology and raised several questions about the availability of data to support some of the assumptions the state used. In addition, the FNS region expressed concern to its headquarters in both October 1988 and in January 1989 that a joint response to the state from FNSand HHS, which FNSheadquarters was proposing, could delay establishment of an approved methodology by which to evaluate FIP cost neutrality, and it urged that a joint federal response be issued to the state as quickly as possible. As of March 16,1990, however, the state said that it had still not received a formal response from either FNSor HHSon its proposals (see app. III). In September 1989, FNSand HHSofficials told us that there were several reasons the two agencies had taken over 16 months to review the state’s proposal but had not yet responded to it: (1) the time it took for their program and fiscal staffs to understand and analyze the proposal, (2) a decision to wait for the results from 1 or 2 quarters of FIP operation so that the proposed methodology could be tested using actual data, and (3) the desire to coordinate the response on the food stamp proposal with their response to the state on similar methodology for the AFDCand Medicaid components of FIP. We did not determine whether FNS' and HHS'need for 16 months or longer to review the state’s three May 1988 administrative cost pro- posals was warranted for the reasons agency officials stated. However, because several questions and problems had been raised on the food stamp proposal by FNSheadquarters and its Western Regional Office as early as August and October 1988, we believe that the state should have been notified of these matters at that time. Such notification would have enabled the state and the federal agencies to work together to correct and/or clarify the food stamp proposal without a lengthy delay, and it might have facilitated an earlier completion of the federal agencies’ review and approval of all three methodologies. Because the state and FNSdid not have an approved methodology for calculating administrative cost neutrality when we completed our review, they could not formally determine whether overall food stamp Page 46 GAO/RCEIMtO-84 Washington’s Family Independence Program Chapter 4 Cast Neutrality of Administrative Expenses Is Still Unclear cost neutrality (benefit and administrative costs combined) was being achieved as the program’s special terms and conditions require. We believe that the federal agencies need to complete their reviews and resolve any differences with the state as expeditiously as possible. FNS’headquarters and Western Regional Office internal reviews, as well Proposed Methodology as our study, identified several problems or potential problems with the Uses Questionable state’s proposed administrative cost-neutral methodology, including sev- Assumptions era1 concerns about the portion of administrative costs for the FIP target population. These issues are discussed in the two following sections. FNS’ Analyses of the FNS’headquarters and regional office reviews in August and October 1988, respectively, recommended that the state make the following State’s Proposal Disclosed changes or additions to correct for methodology problems or assumption SomeProblems and errors in the state’s proposal as submitted: Questions l Exclude some cost categories (such as certain administrative require- ments, claims backlog reduction, and employment and training costs) from the cost-ceiling determination because they either were one-time- only costs that occurred during the baseline period or were otherwise inappropriate for inclusion. l Provide documentation, in support of the annualizing of certain costs in the baseline period calculation, to substantiate that 70 percent of the state’s food stamp administrative costs is for salaries and benefits and that 30 percent is for nonsalary expenses, as well as documentation to show that the 70- to 30-percentage split had been constant for a period of years. l Calculate the portion of administrative costs related to the FTP target population by using data for the full 12-month baseline period of April 1987 through March 1988, rather than by using only l-month’s data, as the state had done to arrive at the 42.57-percentage. l Treat all federal program changes, other than changes in federal finan- cial participation rates, as adjustments to actual costs rather than amendments to the ceiling. l Treat increased salary costs resulting from reclassification of positions as amendments to the ceiling rather than adjustments to actual costs. 9 Develop with FNSa standard methodology for determining when changes in administrative costs are attributable to work load or caseload changes in general, rather than specifically attributable to FIP. Page 46 GAO/RCED90-84 Washington’s Family Independence Program Chapter 4 Cost Neutrality of Administrative Expenses Ie StW Unclear In addition to these recommendations, FNSquestioned three of the pro- posal’s underlying assumptions concerning the 42.67-percent calculation of the FIP target population. First, FNSpointed out that identifying the FIP target population as the 42.57 percent of the statewide food stamp recipients served by the AFDCand the state’s General Assistance-Preg- nant Women programs assumes both that FIP will be implemented state- wide and that all those eligible to convert to FIP will do so immediately upon implementation of the program in their local welfare offices. As mentioned earlier, however, the state has limited FIP implementation to only 20 of the 66 welfare offices in the state, and even in those offices conversion of welfare recipients from the AFDC and General Assistance- Pregnant Women programs to FIP is a voluntary process that takes place over a 12-month period. Therefore, FNSpointed out, it is unlikely that the actual FIP food stamp population (and the costs associated with serving it) will ever approximate the 42.57-percent portion of food stamp recipients served by the two programs. Second, FNSindicated that the 42.57 percent used by the state represents the caseload of food stamp recipients on AFDCand General Assistance- Pregnant Women public assistance programs, not the administrative work load applicable to them. According to FNS, any use of public assis- tance caseload data as a factor in estimating food stamp administrative costs, regardless of the percentage, is likely to overstate the cost ceiling because the cost to administer food stamps to public assistance recipi- ents has historically been proportionately less than the cost to admin- ister food stamps to those not on public assistance programs. FNSstated that its acceptance of the 42.57 percent, or any other caseload-based percentage, depended on the state’s submitting data to show that, in estimating food stamp administrative costs, caseload is equivalent to work load. Third, in addition to measuring FIP food cash administrative costs against a FIP ceiling, as the state’s method proposed, FNSrecommended that total food stamp and FIP food cash administrative costs should be measured against the total food stamp ceiling to ensure that non-HP- related food stamp administrative costs do not increase disproportion- ately to FIP food cash administrative costs without being detected. Our Analysis of SheState’s We independently analyzed the state’s proposed method for determining Proposal Disclosed administrative cost neutrality under the Food Stamp Program. Our anal- ysis indicated that the most straightforward method for determining Additional Problems administrative cost neutrality would probably have been for the state to Page 47 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 4 Cost Neutrality of Admiubtrative Expeuses Ia Still Unclear have calculated a per-case cost of administration prior to FIP and then to compare that cost with a per-case cost of administration after imple- mentation of FIP. The figures could be adjusted, as appropriate, for infla- tion, salary increases, and other costs changes in the same manner as the state has suggested in its August 1988 proposal. However, state agency personnel told us that it would have been too burdensome to develop these figures and therefore proposed the alternative method described in chapter 2 and analyzed in this chapter. As with FNS' find- ings, our review showed the following: l The state needs to provide additional documentation to substantiate both the annualized costs added to the ceiling and the 42.57-percent por- tion of the administrative costs calculated by the state as related to the FrPtarget population. l Isolating administrative costs by caseload instead of work load may not be appropriate. l The state needs guidelines to help demonstrate that caseload and work load changes are not always attributable to FIP. In addition, our analysis revealed two problems not discussed by FNS. First, the 42.57 percent includes all cases for which food stamp eligi- bility determinations were made. Because some of the cases were ulti- mately determined to be ineligible for food stamp assistance and because the state has indicated to us that only about 75 percent of its food stamp administrative costs are related to the eligibility determina- tion process, the remaining 25 percent of administrative costs should be applied, in calculating actual administrative costs, only to those cases determined to be eligible. Second, although not stipulated in the state’s proposal as submitted to the federal agencies in May 1988, the state plans each quarter to recom- pute the AFDCand General Assistance-Pregnant Women proportion that it applies to actual FIP and food stamp administrative costs (calculated in the proposal at 42.57 percent), rather than using the percentage cal- culated for the baseline period as a constant percentage. In our opinion, this procedure should not be undertaken until the state and the over- seeing federal agencies (1) agree on whether the AFDCand General Assis- tance-pregnant Women caseload percentage should be recomputed each quarter and (2) develop and agree on program criteria for determining when caseload changes are to be considered FIP- or non-FIP-related. During our review, no such agreement or criteria existed. Therefore, we concur with the previously mentioned FNSrecommendation that the state and FNSneed to agree in advance on methodology and data for Page 48 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 4 Cost Neutrality of Administrative Expenses In Still Unclear determining when administrative cost changes caused by caseload or workload changes will be considered attributable to FIP- or nOrI-FIP- related causes. Also, to the extent that the state does recompute the caseload percentage each quarter, the change should be limited to caseload variations that are FM-elated. Including non-FIP-related vari- ances in the recomputation could disadvantage the state or the federal government, depending on the direction of the changea The State’s August 1989 On August 11, 1989, after our field work had been completed, the state Revised Proposal Should submitted a revised administrative cost-neutral methodology proposal to HHS.According to the state, the proposal was being revised, after Correct SomeWeaknesses experience with several quarters of FIPoperation, to better ensure con- sistency and accuracy in the calculation of administrative cost neu- trality. FNSand HHSofficials told us on September 27, 1989, that they planed to complete their evaluation of both the state’s original and revised proposals and provide a response to the state in the next several months. Although we did not review the proposal in depth, we noted that it consisted of two alternate methodologies, either of which the state recommended in lieu of its original methodology. Both alternatives provide that most quarterly revisions are to be made as adjustments to actual administrative expenditures rather than as amendments to the administrative cost ceiling. This change is necessary, according to the state, to avoid overstating some adjustments as its orig- inal methodology did. Also, both alternatives use additional state data to revise the ceiling for the percentage of expenditures related to the AFDC and General Assistance-Pregnant Women programs, but one of the two alternatives calculates the percentage on total AFDC and General Assis- tance-pregnant Women program cases for which food stamp eligibility determinations were made, while the other alternative calculates the percentage only for cases that were determined to be eligible. The effect of these proposed changes, according to the state, is to reduce the amount by which the state was under the food stamp administrative cost-neutral ceiling for the first 3 quarters of FIP operation compared with the amount based on its original proposal. As shown in table 2.3, “If the AFDC and General Assistance-Pregnant Women caseload percentage increases (from, for example, 42.67 percent to 48 percent), the inclusion of non-FIP-related changes would erroneously overstate actual costs and could make it appear that the state had exceeded the cost-neutral ceiling, when in fact it had not. Conversely, if the caseload percentage decreases (from, for example, 42.57 percent to 36 percent), the inclusion of non-FIP-related changes would erroneously understate actual costs and could make it appear that the state had not exceeded the cost-neutral ceiling, when in fact it had. Page 49 GAO/RCED-90-84 Washington’s Family Independence Program Chapter 4 Cost Neutrality of Administrative Expenses Is Still Unclear using its original methodology, the state calculated that its actual administrative expenditures were $536,299 under the ceiling. Using its two proposed alternative methodologies, the state calculates that its expenditures would have been either $473,838 or $439,601 under the ceiling, depending on the alternative used. Although the changes pro- posed by the state correct some weaknesses in the original proposal, our limited review indicates that they do not solve all of the problems raised by FNS’ and our reviews. As mentioned earlier, the state calculated that its administrative costs State’s Quarterly for food stamps fell below the cost-neutral ceiling for the first 3 quarters Calculations Need to of FIP by a substantial margin of $536,299. We traced the data the state Be Viewed With used in its calculations to their source documents (federal financial status reports and state reports created specifically for determining Caution amendments to the ceiling) and found that the calculations appeared to be appropriate. The state is uncertain about the reason it shows itself so far under administrative cost neutrality so early in the demonstration and has pointed out that the early results should be viewed with cau- tion. We agree with the state’s position for three specific reasons. First, costs reported for the first 2 quarters of FIP operation may be understated because the state’s automated cost allocation system, which is used to collect and distribute administrative costs to the various state, federal, and other programs the state operates, was not programmed to collect all FIP costs until the beginning of the third quarter. FIP adminis- trative costs for the first 2 quarters (which were used by the state and FNSto calculate cost neutrality) were determined through the use of manually prepared schedules and manual computations. The potential for error using this method of allocation is greater than when using reports generated by an automated system. For example, our detailed analysis showed that the state erroneously underreported FIP adminis- trative costs by $104,233 for the second quarter of the project’s opera- tion because it inadvertently omitted direct overhead charges attributable to the FIP headquarters office. Second, a state official told us that costs reported by any given FIP wel- fare office during its first year in the program may be inaccurate because of the difficulty and complexity of allocating overhead costs correctly on a new program when they must be divided not only among federal and state programs but also among FIP and non-FIP portions of those programs. In addition, he said that time study data, which reflect Page 50 GAO/RCED-SO-94 Washington’s Family Independence Program Chapter 4 Coet Neutrality of Administrative Expenses Is Still Unclear how much time individual welfare office workers spend on each pro- gram, and which are used to project percentages for the amount of time spent by all workers on all programs, is applied to the cost allocation process on the basis of a 2- to 3-month lag. In a start-up program such as FIP, the caseload can increase significantly each month as new applica- tions come in and as additional AFDCrecipients become eligible for con- version to FIP. Data about the FIP work load from 3 months ago are therefore not likely to be representative of what is occurring in the local welfare offices currently. The state official said that, for this reason, administrative costs reported for any welfare office during its first year under FIP should not be considered definitive, and cost-neutral calcula- tions based upon them should not be viewed as conclusive. Third, the state’s cost-neutral calculations for the first several quarters may not have been consistently applied. For example, in addition to the $104,233 omission of FIP headquarters charges discussed earlier, our analysis disclosed that the state’s food stamp administrative cost-neu- tral calculation for the first quarter of FIP operation (July 1, 1988, to September 30, 1988), which was made by one state office, omitted approximately $170,000 in FIP administrative costs that had been reported to FNSby another state office. In addition, our review showed that the state did not provide adequate back-up documentation to sup- port some of the amendments it made to the ceiling for the first 2 quarters’ calculations. Our review of FNS' food stamp administrative cost-neutral calculations F’NSNot Making revealed one major and one minor area of concern. Required Calculations Our primary concern is that FNSwas not making administrative cost- neutral calculations using actual expenditures each quarter, as required by the project’s March 1988 special terms and conditions. While FNSesti- mated administrative costs at the beginning of each quarter and com- pared them with the ceiling to determine the appropriate amount of funds to release to the state, it was not calculating the state’s actual administrative costs at the end of each quarter and adjusting them . according to the proposed methodology, or comparing actual expenses with the ceiling to determine whether the state was over or under the administrative cost-neutral ceiling for the quarter. In May 1989, an FNS official said his agency was not calculating administrative cost neu- trality on the basis of actual costs because of the need to Page 51 GAO/RCED-90-84 Washingtxm’s Family Independence Program Chapter 4 Cost Neutrality of Administrative Expenses Is Still Unclear . complete final closeout acijustments to the state’s administrative cost claims for federal fiscal year 1988 and send the adjusted data to FNS headquarters for review; . adjust the state’s administrative cost claims for the first 2 quarters of federal fiscal year 1989; and l assess the effect of the proposed change to the project’s special terms and conditions, which would permit the state to achieve food stamp cost neutrality over a 36-month rather than a 3-month perioda In addition, FNSdid not believe that such calculations were critical to make at that time because benefit cost-neutral calculations and adminis- trative cost-neutral estimates for the first 3 quarters were coming in far below the overall cost ceiling. According to the FNSofficial, the lack of an approved methodology was not why FNSwas not calculating actual administrative cost neutrality, because FNSstaff had been instructed to operate as if the state’s method- ology had been approved. We did not examine each of the cited reasons to determine whether the needed actions could or should have already been taken, or to what extent, if any, the cited reasons precluded FNS from making the required calculations, Although we did not, perform the calculations, we pointed out earlier and in chapter 2, that the lack of such calculations meant that FNSwas unable to conclusively determine whether the state was in compliance with the cost-neutral requirement of FIP'S authorizing legisla- tion. As we also noted, the state was making such calculations. A second concern is with FNS' calculation of the ceiling for food stamp administrative costs. FNS' calculation of the unadjusted ceiling differed by about $12,000 from the state’s calculation. The difference between the state’s and FNS' computations is due to the use of data from different cost categories on the FNSfinancial status reports submitted by the state. Also, FNSmade a mathematical error in its computation of the ceiling adjustments. The extent to which administrative expenses for food assistance are Conclusions being kept within the cost-neutral maximums under FIP is unclear. As of Y March 16, 1990, over 22 months after the state submitted its original administrative cost-neutral methodology proposal, FNSand HHShad not “This proposed change was discussed in ch. 2. Page 52 GAO/RCED80-84 Washington’s Family Independence Program chapter 4 Cost Neutrality of Administrative Expenses Is Still Unclear yet completed their review of the proposal or notified the state of the methodology questions and other problems they had found. Further- more, although the state’s calculations for the first 3 quarters of FIP operation showed that its administrative costs for food stamps fell below the cost-neutral ceiling by the substantial margin of $536,299, the initial calculations should be viewed with caution for the following reasons: . Some of the assumptions underlying the state’s proposed method for determining the administrative ceiling and making quarterly calcula- tions-especially those related to the use of data on the AFDCand Gen- eral Assistance-Pregnant Women programs’ caseloads-are questionable. . The state’s automated cost allocation system was not completely programmed to account for FIP administrative costs during the first 2 quarters of FIP operation. Although this situation had been partially alle- viated by the third quarter, at which time FIP administrative costs began to be allocated on an automated basis, caution still needs to be taken when viewing FIP administrative costs because of the uncertainty involved in accurately identifying administrative costs during any par- ticular local welfare office’s start-up phase. . No independent verifications of the state’s calculations were routinely occurring because FNSwas not making cost-neutral calculations on the basis of actual expenditures each quarter, as it is required to do. We recommend that the Secretaries of Agriculture and Health and Recommendations Human Services direct their agencies to expedite their response to the state of Washington regarding the state’s proposals for determining food stamp administrative cost neutrality and resolve with the state the problems and concerns raised by the agencies’ and our reviews. We also recommend that the Secretary of Agriculture direct the Admin- istrator, FNS,to ensure that food stamp administrative cost neutrality is calculated each quarter on the basis of actual food stamp and FIP food cash administrative expenditures. Neither FNSnor HHScommented on our findings or recommendations State Comments and concerning administrative cost neutrality. Washington State indicated Our Evaluation that as of March 1990, it had still not received a formal response from the two federal agencies to its May 1988 proposed methodology for Page 53 GAO/RCED-30-84 Washington’s Family Independence Program Chapter 4 Cost Neutrality of Administrative Expenses Is Still Unclear establishing administrative cost neutrality for the Food Stamp, Medi- caid, or AFJX portions of FIP. The state also noted that its next quarterly submission of FIP cost-neutrality data will contain some additional adjustments that it will have to negotiate with federal representatives. As we stated earlier, the lack of an approved administrative cost meth- odology for the Food Stamp Program portion of FIP prevents the state and FNSfrom formally determining whether overall food stamp cost neu- trality has been achieved. It also leaves in question the extent of the state’s and the federal government’s respective cost responsibilities for operation of FIP. Accordingly, we continue to believe that FNSand HHS need to complete their reviews and resolve any differences with the state on administrative costs as quickly as possible. Page 54 GAO~RCEDGO.~~ WaAingtm’r FamUy Independence Program Page 55 GAO/RCEJMO-84 Washington’s Family Independence Pmgram Comments From the U.S. Department of Agriculture’s Food and Nutrition Service Note: GAO comments supplementing those in the report text appear at the United States Food and 3101 Park Center Drive end of this appendix. Department of Nutrition Alexandria, VA 22302 Agriculture Service Mr. John W. Harman, Director Food and Agriculture Issues Resources, Community and Economic Division U.S. General Accounting Office Washington, D.C. 20548 Dear Mr. Hax-man: We have received your draft report entitled "FOOD STAMP PROGRAM: Achieving Cost Neutrality in Washington's Family Independence Procram." This report was prepared in response to the statutory requirement for the Comptroller General of the United States to conduct periodic audits to verify Washington's assurance of Federal cost neutrality for payments made with Food Stamp Program funds to support the project. The Omnibus Budget Reconciliation Act of 1987 which authorized the Family Independence Program (FIP) holds Washington to Federal cost neutrality separately for the Food Stamp, Aid to Families with Dependent Children (AFDC) and Medicaid Programs. It should be noted that food assistance costs make up only a small part of Federal expenditures under FIP. State reports indicate that less than 13 percent of total Federal and State expenditures for FIP See comment 1 are used for food stamp benefits and administrative expenses. We agree with GAO's conclusion that "a reasonable approximation of possible program costs is probably the best that can be achieved," despite the legislative requirement for "assurance." In developing and negotiating the Special Terms and Conditions under which FIP would operate, the State and Federal agencies concluded that a methodology could be developed that would provide a reasonable estimate of the aggregate value of coupons and administrative costs that would have been distributed under the Food Stamp Program if the individuals in FIP had participated instead in the traditional Food Stamp Program. While the methodology entails some potential risks for overpayment to the State, we believe these to be minimal under the circumstances. The report notes that the benefit cost neutrality methodology was developed based on the assumption that FIP would be expanded Statewide (except for the control offices) within three years. With these plans in mind, Federal and State negotiators reached agreement on a method of estimating the maximum amount of food assistance (food stamps and cash payments to FIP participants) the State would be eligible for if only the traditional Focd Stamp Program were operating in Washington. This method involves Page 66 GAO/RCED-90-84 Washington’s Family Independence Frogram Appendix I Comments From the U.S. Department of AgrIcuIture’s Food and Nutrition Service Mr. John W. Harman, Director 2 projecting Statewide costs based on actual Food Stamp Program expenditures in the control offices. The report also notes that the control offices were selected to be representative of the State as a whole. We believe this methodology was reasonable based on the information available at the time it was developed. In reaching the conclusion that "alternative estimating methods may better assure food stamp benefit cost neutrality," the report identifies several assumptions underlying the State's method for estimating the anticipated value of coupons that would have been distributed under the traditional Food Stamp Program, One of these assumptions, that FIP would be implemented Statewide, has been addressed above. GAO challenges the use of two other assumptions as potentially undermining the accuracy of the current method of assuring compliance with the cost neutrality requirement for FIP. With regard to the assumption that prior period costs for the control sites would be stable in relation to Statewide costs, GAO contends that a longer (than one year) baseline would have provided a better assessment of stability. At the time negotiations were taking place, State officials pointed out that the current configuration of Community Services Offices had been in place for just over one year. Actual food stamp cost data could not be readily isolated for the control sites for longer than a one year period. Federal officials accepted that position. The methodology also assumed that future costs in the control sites would be stable in relation to Statewide costs. GAO believes that significant economic, demographic, or other changes affecting food stamp expenditures in any of the control sites that do not occur in the rest of the State, or vice versa, should trigger an adjustment to the proportion of the State's total food stamp costs. We agree that it would be desirable to make adjustments for the effects of changes outside the Food Stamp Program, provided that these changes can be identified as affecting the program and the magnitude of their effects can be estimated. We would entertain any specific suggestions from Washington State about the types of indicators and extent of change that should be considered relevant to adjusting the See comment 2 methodology. Given that there are no plans for Statewide implementation of PIP, GAO believes that alternative methods for establishing food stamp cost neutrality ceilings need to be considered. The report expresses concern that under the current method, small variations in the proportion of the State's costs represented by the control sites can have a substantial impact on the cost ceiling estimates. Three alternatives were suggested that would address the "multiplier" effect . We have reviewed each of the suggestions and believe that in correcting for the multiplier effect they may introduce new sources of error or bias, Y Page 57 GAO/RCED-90-84 Washington’e Family Independence Program Appendix I Canmenta From the U.S. Department of Agriculture’e Food and Nutrition Service Mr. John W. Harm-an, Director 3 The reduced scope of FIP does merit reconsideration of the principles underlying the cost neutrality methodologies. State and Federal officials are currently exploring alternative methods of estimating cost neutrality ceilings. GAO's observations and findings will be helpful to us in considering various options and See comment 3. alternatives. Thank you for giving us the opportunity to review and comment on this report. We also greatly appreciate your staff's willingness to meet with us to address our questions and concerns about the draft report. Sincerely, Administrator Page 88 GAO/RCED-90.94 Washiugton’s Family Independence Program Appendix1 Commenta From the U.S. Department of Agriculture’s hod and Nutrition Service The following are GAO'S comments on USDA'S letter dated March 13, 1990. 1. We have modified our report to show the comparative size of the food GAO Comments: assistance cost to the other component costs of FIP on page 11 of the report. 2. We have included our response on pages 38 through 40 of the report. 3. We have included this information on page 40 of the report. Page 69 GAO/RCED4O-B4 Washington’s Family Independence Program Appendix II Comments From the Department of Health md Human Services Note: GAO comments supplementing those in the report text appear at the end of this appendix. DEPARTMENTOFHEALTHLHUMANSERVICES Oltice 01 Inspector General WSShlngton. D.C. 20201 MAR191990 Mr. John W. Hannan Director, Food and Agriculture Issues United States General Accounting Office Washington, D.C. 20548 Dear Mr. Hannan: Enclosed are the Department's comments on your draft report, @'Food Stamp Program: Achieving Cost Neutrality in Washington's Family Independence Program." The comments represent the tentative position of the Department and are subject to reevaluation when the final version of this report is received. The Department appreciates the opportunity to comment on this draft report before its publication. Sincerely yours, F---T\ ! Richard P. Kusserow Inspector General Enclosure Y Page60 GAO/RCED-90-84Washhgton'sFamilyIndependenceFrogram Appendix II Comments From the Department of' Health and Human Services COMMENT6OF THE DEPARTMENT OF HEALTH AND HUMAN SERVICES ON THE U.S. GENERAL ACCOUNTINQ OBFICE'B REPORT, "FOOD STZ&MP PROam . itv in ~~41~ 8 Zn$hmaden ce We have received a draft of the GAO report, FOOD 1~ Indeoendence proaram- We appreciate the opportunity you have given us to provide comments. The current methodology for determining cost neutrality was developed during a period when state plans for FIP called for statewide implementation. The best methodology for establishing a cost baseline would have been a representative randomly selected group of Aid to Families with Dependent Children (AFDC) recipients. This option was not available to us at the time. Therefore, we, together with the state, chose a representative sample of community service offices which contained a sufficient See comment 1 number of AFDC recipients to generate a statewide baseline. We recognize, along with the State, that since there are currently no plans to operate Family Independence Program on a statewide basis, a review of the cost neutrality methodology is warranted. The alternative cost neutrality proposals and recommendations put forth by GAO will be taken under considera- tion in such a review. However, it is not clear at this time that GAO's recommendations will provide a more accurate baseline than See comment 2. the current methodology. The Department appreciates the opportunity to comment on this draft report before its publication. Page 61 GAO/RCED-90-84 Washington's Family Independence Program Appendix II Comments From the Department of Health and Human Services The following are GAO'S comments ori HHS' letter‘dated March 19, 1990. 1. We have included this information and our response on pages 38 and GAO Comments: 39 of the report. 2. We have included this information and our response on page 40 of the report. Page 62 GAO/RCED-SO-84 Washington’s Family Independence Program Appendix III CommentsFrom the Stateof Washington Note: GAO comments supplementing those in the report text appear at the ‘YI end of this appendix. KKHARD I TkOMPSON g, SUWWV @ “MB STATEOF WASHINGTON DEPARTMENT OF SOCIAL AND HEALTH SERVICES Olympia. Washington 96.504~5 March 16, 1990 Mr. John W. Harman Director, Food and Agriculture Issues United States General Accounting office Resources, Community and Economic Development Division Washington, D.C. 20548 Dear Mr. Hannan: This is in response to your letters of January 25, 1990 to Commissioner Turner and me requesting a review of your draft report entitled Food Stgmp Proaram. . QBservaons on w FIP Cost r Matter& (GAO/RCED-90-84). The comments provided here constitute a joint response by the Washington State Employment Security Department and the Department of Social and Health Services. If you have questions about the details of our response, please COntaCt Barbara Flaherty, Executive Director, Washington State Family Independence Program, at (206) 753-6050. For the most part, we agree with the conclusions presented in the See comment 1 I draft report. We feel, however, that there are some broad@r issues not considered in this document which must be examined before a final consensus on cost neutrality for either Food Stamps, AFDC, or Medicaid can be reached. We would first like to address these general issues in part A below and then comment on some of the specific items included in your draft report in part B. (A) ceneral.auesu not Addressed in the Draft Rem Washington was the first state to receive specific federal legislation mandating waivers necessary for the operation of its welfare reform program. The Omnibus Budget Reconciliation Act of 1987 (OBRA) required the state to provide assurances that over the five-year life of the demonstration, the cost of FIP to the federal government would be no greater than what the cost would have been under the regular AFDC, Medicaid, and Food Stamp programs in the absence of FIP. Y Page 68 GAO/RC~~O-~~ Washington’s Family Independence Program Appendix III CommentsF'remtheBtateofW~hingten Mr. John W. Harman March 16, 1990 Page 2 The language governing cost neutrality is presented in sections 9121 and 1509 of OBRA. Section 9121 states in part that: . . . As a condition of approval of the project under this section, the state must provide assurances satisfac- tory to the Secretary that the total amount of Federal reimbursement over the period of this project will not exceed the anticipated Federal reimbursements (over that period) under the AFDC and Medicaid programs: but this paragraph shall not prevent the state from claiming reimbursement for additional persons who would qualify for assistance under the AFDC program, for costs attribu- table to increases in the state's payment standard, or any other federally-matched benefits or services . . . (Emphasis added.) Similarly, section 1509 in part requires that: . . . the cost of food assistance provided under the project will not be such that the aggregate amount of payments made under this section by the Secretary to the state over the period of the project will exceed the sum of... the anticipated aggregate value of the coupons that would have been distributed under the Food Stamp program if the individuals who participate in the program had participated instead in the Food stamp program . . . (Emphasis added.) The specific rules governing cost neutrality for the Family Independence Program are contained in Section 3 of the Special Terms and Conditions. These rules also cover a variety of aspects of FIP operation besides cost neutrality. When the cost neutrality provisions were adopted, neither federal nor state representatives had significant previous experience in this area. Therefore, it not surprising that after six quarters of FIP operation izestions continue to be raised regarding whether the methodolog; contained in the Special Terms and Conditions fully conforms to the requirements of the authorizing federal legislation and whether the original methods and data sources selected for determining cost neutrality of FIP continue to be the best and most accurate indicators of program operations and cost. After reviewing the first full year of FIP operation, we sought federal approval for several changes in the methodology for determining cost neutrality. On November 20, 1989, we wrote to the Secretaries of the U.S. Departments of Agriculture and Health and Human Services requesting corrections in wording and interpretation of the Special Terms and Conditions which we believed would address several new issues which had arisen since the beginning of FIP and Page64 GAO/RCED-90-84Washington'sFamily IndependenceProgram Appendix III Commenta From the State of Washington Mr. John W. Harman March 16, 1990 Page 3 which would make the methodology better conform to the original intent of congress (copy enclosed). In those letters we requested that: 1. [Cost] neutrality for the Washington State Family Independence Program be determined in the aggregate over the full five year life of the FIP demonstration. The state should be able to recoup any reductions in its claim for federal match resulting from expenditures in excess of [cost] neutrality out of any future savings from expenditures under budget neutrality. 2. Except as previously agreed regarding the limits on change that may occur in evaluation control sites, Washington may pass on to its public assistance recipients (including FIP enrollees) all improvements in services and benefits author- ized under federal legislation since passage of the Omnibus Budget Reconciliation Act of 1987 (OBRA), including those improvements included in the Family Support Act. 3. The U.S. Departments of Health and Human Services and Agriculture should participate fully in the costs of services to FIP enrollees, without regard to [cost] neutrality, so long as: 0 Benefits are provided to persons who would be eligible for participation in the AFDC program or any of its suc- cessors, including FSA/JOBS, and under these provisions categorically eligible for Medicaid and Food Stamp benefits, 0 Benefits and services that are provided, and the administrative costs that are incurred, would have been generally eligible for reimbursement under AFDC, Medicaid or Food Stamp programs and their successor programs. 4. The Departments of Health and Human Services and Agriculture should approve adjustments to the administrative cost method- ology which will provide for full federal participation in increases in administrative costs related to expansion of benefits or services as described above. In a letter dated January 24, 1990, Arnold R. Tompkins, Acting Assistant Secretary for Planning and Evaluation, DHHS, responded to our requests, generally approving changes relating to items one and two above. As of yet, no specific details have been negotiated regarding how any of these provisions would be implemented. Item four regarding administrative costs was not addressed in Mr. Tompkins' response. In your draft report, you indicate that after Page 65 GAO/RCED-w-84 Washington’s Family Independence Program Appendix III Commenta From the State of Washingt0n Mr. John W. Harman March 16, 1990 Page 4 more than a year we had not received any response to our proposed methodology for determining cost neutrality for administrative costs. As of March 1990, this is still the case. The most important issue raised in our appeal was item number three above, a request for changes in the Special Terms and Conditions which would require federal participation in a significant portion of the cost of providing Food Stamp, AFDC, and Medicaid benefits to persons who, in the absence of FIP, would have been eligible for participation in the regular AFDC, Medicaid, and Food Stamp programs and their successor programs. Since its inception, FIP has experienced an unanticipated increase in caseload. The great majority of these additional persons would have been eligible under the regular programs and both state and federal legislation required that we serve them. We believe that Sections 1509 and 9121 of OBRA, cited above, unambiguously require federal participation in some portion of the cost of caseload growth which we have experienced. Members of the Washington State Congressional delegation and staff, that participated in drafting this portion of OBRA, indicate that the exception language included in OBRA (e.g. I*. . . but this paragraph shall not prevent the state from claiming reimbursement for additional persons who would qualify for assistance under the AFDC program . . .I') was specifically included to cover these kinds of cost. Unless the federal government agrees to paying its fair share of the costs of serving these additional persons, our lia- bility for additional cases under the cost neutrality provisions See comment 2. may be so great as to threaten continuation of the demonstration. We have also requested review by the Secretaries of HHS and Agriculture of a denial by the FIP federal project officer of a proposal for changing how the accounting for cases that migrate from FIP to AFDC comparison sites is handled. Currently we allow FIP enrollees who move from a FIP Community Service Office (CSO) to a non-FIP CSO to continue in the program as long as there is a FIP CSO in the county where they reside. FIP enrollees move for programmatic reasons (such as to seek improved work or training opportunities not available in the FIP sites) as well as personal reasons (to obtain better or cheaper housing, to live closer to relatives, etc). For purposes of administrative efficiency, we have continued to manage these cases and account for their costs in the FIP CSO nearest their home. In this way, we do not have to provide special training for staff in those offices that might handle only a few FIP cases. Because these cases continue to be managed out of FIP Offices, they are not included in cost statistics for the control sites which are used to project the baseline or "would-have-been" cost against Page 66 GAO/RCED-90-84 Washington’s Family Independence Program . Appendix III Comment0 From the State of Washington Mr. John W. Harman March 16, 1990 Page 5 which cost neutrality is measured. This exclusion artificially reduces the baseline against which our actual expenditures are measured for cost neutrality purposes. The FIP Executive Committee, the governing body for the program, has voted to dis- continue allowing FIP enrollees to participate if they migrate from FIP sites unless they move to fulfill their training plan or to obtain work. It is expected that enrollees that have already moved would continue in the program. As indicated in the draft report, the cost of each case in the control sites is multiplied by a factor of about seven to produce the baseline. The exclusion of even a few of these families from the equation results in great disadvantage to our finances. In December of 1989, there were about 690 cases in this category. Even when the new Executive Committee policy limiting FIP participation after out-migration takes effect, there will still be a backlog of accumulated costs not included in the projected baseline of more than $2.9 million for Food Stamps and $14.3 million for AFDC and Medicaid (total cost) after the first six-plus quarters of FIP operation. (Please see the enclosed copies of the original letter to our project officer, his response, and our appeal to the Secretaries of Health and Human Services and Agriculture for more details on these calculations). Mr. Tompkins turned down our request concerning participation in the cost of caseload growth, and as yet we have had no formal response to our appeal of the migration issue. There have been some later, direct negotiations between federal representatives from DHHS and USDA and me, where all parties have sought to reach some compromise on this issue. If we prevail in our position regarding federal participation in the costs of caseload growth, either through negotiation or an appeal to the courts, then the methodology included in Section 3.06 of the Special Terms and Conditions (which was amended to include the calculation of Food Stamp benefit cost neutrality as well as AFDC and Medicaid benefits cost neutrality) would have to be modified significantly. We believe that required changes would be so extensive as to require opening the discussion of cost neutrality methods considerably Seecomment3. beyond the parameters found in the draft GAO report. IB) Specific Issues Addressed in the GAO Draft Renort cu rren t Me thod fo r Estimatinq Cost Neutralit v for Benefi t s The draft GAO report directs much attention to questions about the technical appropriateness of using the ratio of Food Stamp benefit costs in comparison sites (which were selected for use in the FIP evaluation) to total Food Stamp benefit costs in the base time Page 67 GAO/RCED-90-84 Washington’s Family Independence Program Appendix III Comments From the State of Washington Mr. John W. Harman March 16, 1990 Page 6 period (April 1, 1987 through March 31, 1988), to estimate a llwould-have-beenll cost against which benefits cost neutrality is measured in succeeding time periods. Questions are raised as to whether the following assumptions implied or contained in the existing cost neutrality methodology are correct: 1. Control sites' prior costs were stable in relation to statewide costs, 2. Control sites' future costs will be stable in relation to statewide costs, 3. FIP would be implemented statewide. The methodology which is currently used to project cost neutrality was, for the most part, a creation of the federal representatives See comment 4. involved in the negotiations of the content of the Special Terms and Conditions. All the implied assumptions listed in the draft report were in fact just that--assumptions. The comparison sites were first selected for use in the required evaluation of FIP and later adopted for use in determining cost neutrality. In the haste to get FIP underway, neither federal nor state representatives thoroughly reviewed all the implications of the IUetpJdOlOgy which was established. Similarly, neither federal nor state representa- tives examined the stability of either Food Stamp, AFDC, or Medicaid base costs in comparison sites for even one year prior to the beginning of FIP, let alone for five years as presented in the draft report. The draft report indicates that even a one percent error in the projected baseline against which cost neutrality is measured could result in a large actual swing in dollar savings or liability for the state. In fact, an actual error in the l1would-have-been'0 cost of more than plus or minus one percent would probably be expected as the result random error associated with the type of sampling methodology used to establish the baseline. Since FIP has not been implemented statewide, the mathematics of the methodology would also seem to dictate that more of the potential variance in actual total Food Stamp costs, above or below the baseline, could be the result of either random or systematic fluctuation of costs in non- FIP/non-comparison sites which are not controllable by FIP, as See comment 5. opposed to the actual effect of FIP activities. Certainly, there has been some FIP-generated increase in costs associated with the unanticipated growth in caseload which we have experienced. Using the current methodology, this increase should show up as a cost neutrality liability for the state, regardless of the problems mentioned above, but it is still difficult to sort out the true source of the differences between projected baseline Page 68 GAO/RCED-9044 Washington’s Family Independence Program Appendix III Comments From the State of Wahington Mr. John W. Harman March 16, 1990 Page 7 and the actual total costs which have been reported. We must again stress here our conviction that OBRA requires federal participation in a significant portion of the costs related to caseload growth. Since the current methodology is designed to specifically exclude these types of cost from federal participation, we find it fundamentally flawed, much beyond the aspects which you describe in your draft report. \ The report recommends that state and federal representatives explore the use of other ratios to determine FIP cost neutrality savings or liabilities, including projections based on using: 1. Costs in 7 current control sites to establish a ceiling for the 20 FIP.sites only; 2. Costs in the 19 paired non-FIP sites to establish a ceiling for the 20 FIP sites only; and 3. Use of costs in all 45 non-FIP sites to establish a ceiling for the 20 FIP sites only. In addition to the migration issues discussed above, each of the three options increases the influence of a methodological flaw related to how the benefit costs of AFDC eligibles or enrollees who migrate into FIP sites would be handled. The current methodology compares the total of actual FIP and non-FIP site costs to the projected baseline. Each of these three options compares only FIP site costs to the different bases which are indicated. This would assume that all differential increases in costs in FIP sites, over and above the increases in the projected baseline, are the result of a FIP effect. Therefore, the state would be fully responsible for these increases. Recall again, Section 1509 of OBRA requires in part that: . . . the cost of food assistance provided under the project will not be such that the aggregate amount of payments made under this section by the Secretary to the state over the period of the project will exceed the sum of . . . the anticipated aggregate value of the coupons that would have been distributed under the Food Stamp program . . . We interpret "aggregate value of coupons that would have been distributed under the Food Stamp Program" to be the total that would have been Spent for all Food Stamp benefits in the absence of FIP. Y Page 69 GAO/RCED-96-84 Washington’s Family Independence Program Appendix Ill Comments From the State of Washington Mr. John W. Harman March 16, 1990 Page 8 Under each of the three alternative projection methods, when a case eligible for AFDC or already on AFDC, migrates into a FIP site in order to receive assistance there, its cost shows up as an entirely new expense thus becoming the full responsibility of the state. In actuality this type of case constitutes no additional cost to the overall Food Stamp program. Rather, such enrollees are merely receiving their Food Stamp benefits in a different location. For these migrating cases, there may be some FIP-induced increase in Food Stamp, AFDC, or Medicaid costs associated with such factors as increased length of stay on assistance or increased payment level due to the cost of incentive payments over and above the cost of "$30 and one-third" under AFDC. These are marginal increases as compared with the additional cost of the basic benefit which would be fully attributable to PIP. If one of the three projection alternatives is seriously considered, we would have to make an accurate estimate of the rate of migration into FIP sites and propose a suitable correction. Assuming that the problems related to fair federal participation in the cost of caseload growth and migration are solved, then we agree it would be advantageous to both the federal government and ourselves to place some limit on the cost savings or liability that might result from random or systematic variation of costs unrelated to a FIP effect. (This was illustrated in the report by the relative effect of a one percent variation in cost ceiling, using the current methodology and the three scenarios listed above.) Use of some variation of the three alternatives proposed in your draft report might improve the calculation. We have not yet determined the full relative impact of adopting any of these three options on either Food Stamp, AFDC, or Medicaid costs. We do know that each of these options involves the use of a new or different sample to project a baseline against which to measure FIP experience and that the assumptions of representativeness of each should be considered Seecomment in detail before selection or substitution is made. Differences in State and Federal Cuarterlv Cost Neutralitv Calculations for Food Stamp Benefits In developing our baseline data for Food Stamp budget neutrality calculations we used Table 2K from the monthly state statistical publication entitled Income Assistance, Community Services, and Medical Assistance (commonly known as the I'Blue Booklq). Approval of the methodology for the benefits baseline calculation (including the Food Stamp calculation) came from Daniel H. Weinberg, Acting Assistant Secretary for Income Policy of DHHS, in a letter dated April 7, 1989. Our first two submissions of Food Stamp cost neutrality reports used Table 2K for coupon issuances data and Table 2N for FIP food cash, both from the Blue Book. Page 70 GAO/RCED-90-84 Wanhington’r Family Independence Frogram Appendix III Commenta From the State of Washington Mr. John W. Harman March 16, 1990 Page 9 In July of1989 we received a letter from Allen Ng, Regional Director, Food Stamp Program, USDA, San Francisco, indicating "reporting inconsistencies" in the Washington State Family Independence Program identified by GAO auditors. In particular, he was concerned about discrepancies between Blue Book tables 2N and 2s which reported similar information on FIP food cash issu- antes. He requested that we report cost neutrality information using data from the adjusted FNS-388 report as opposed to the Blue Book. In August of 1989, we responded to Mr. Ng via two separate letters, indicating the steps that would be taken to meet his concerns. First, we stated that we would improve the descriptive heading and footnotes for Table 2N in the Blue Book and would modify Food Stamp reconciliation data in Table 2s to include daily FIP cash food assistance issuances and cancelled food assistance benefits. These changes would begin with the August 1989 reporting period. Second, we said that we would continue to report cost neutrality information for food coupons from Table 2K in the Blue Book and that we would switch from Table 2N to 2s for FIP food cash reporting. We would revise all previous Food Stamp cost neutrality submissions and base them on Table 2s. Data for reporting periods after August 1989 would reflect the changes in calculating Table 2s indicated above. However, we would not be able to go back and correct Table 2s for the period prior to August 1989. Tables 2K and 25 provide information identical to that which is found in the unadiusted FNS-388 report. The final FNS-388 includes adjustments made on at the county level which are not comparable with data broken down by Community Service Office as mandated for use in the Special Terms and Conditions. By using tables 2K and 2S, we have both food cash and food coupon data from consistent sources. Because cost neutrality is figured on a ratio basis, the differences between using the adjusted and unadjusted figures were thought to be insignificant. FNS has not yet responded to our letters of last August. There is one additional issue regarding how the Food Stamp cost neutrality information is collected and used that must be dis- cussed. The FNS Regional Office in San Francisco is responsible for releasing the funds to reimburse Washington State for its Food Stamp expenditures. In his letter of July 1989, Mr. Ng indicated Page 71 GAO/RCED-90-84 Washington’s Family Independence Frogram Appendix III Commenta From the State of Wasldngton Mr. John W. Harman March 16, 1990 Page 10 that the cost neutrality baseline would be used to release Food Stamp funding according to the following equation: Baseline - Food Stamp FNS-388 Adjusted Actual Issuance = Funds Available for FIP. The FNS San Francisco Office continues to use this formula. There are several things wrong with this equation. First, the formula mixes data from differing sources, using FNS-388 adjusted data in relation to a baseline derived from Table 2K which uses unadjusted data. Second, this formula involves the use of a data set for cost neutrality calculations other than that which we have submitted or have formally proposed to submit. FNS has never responded or made any formal determination of the appropriateness of the data changes we proposed last August. Third and most important, this formula is based on requirements of Section 3.01 of the Special Terms and Conditions which established a more restrictive reimbursement and payback schedule for Food Stamp expenditures than was set up for AFDC or Medicaid cost neutrality. We believe that Section 3.01 is no longer operational. When Section 3.01 of the Special Terms and Conditions was established, USDA representatives in Washington, D.C. believed that OBRA Section 21(d)(3), which covered food cash assistance to mixed households, would open the door to much more extensive Food Stamp participation under FIP. The state of Washington agreed with this position and took immediate steps to have this portion of the Act amended or removed. In a letter dated March 9, 1980, approving the Food Stamp portion of the FIP demonstration, Richard Lyng, Secretary of Agriculture! indicated that should Section 21(d)(3) be repealed, he would discuss changes in the Special Terms and Conditions and that he . . . would expect Section 3.01 of the Special Terms and Conditions would be deleted, Section 1.02 amended to reflect any resultant changes in the Washington Administrative Code and Sections 3.06 and 3.07 [governing cost neutrality computations and payback for AFDC and Medicaid] amended to incorporate the Food Stamp Program. In September 1988, Section 21(d)(3) was repealed by the U.S. Congress. A study completed by us that fall showed that the provision had little effect on cost prior to its cancellation. In January or February 1989 we received a hand annotated copy of the Special Terms and Conditions from the Office of the Assistant Secretary for Planning and Evaluation, DHHS, incorporating the changes Secretary Lyng set out. On March 16, 1989, we wrote to Mr. Page 72 GAO/RCED-9084 Washington’s Family Independence Prt~gram , Appendix III Comments From the State of Washington Mr. John W. Harman March 16, 1990 Page 11 Ng formally requesting these changes but did not receive any response. We believe that we and the Department of Agriculture should now be at the same place regarding what are the sources of the numbers we are currently using to determine Food Stamp benefit cost neutrality and what are the sources of the numbers we should be using. In the absence of any formal notification to the contrary, we plan to continue to collect and submit Food Stamp cost neutrality data according to the formulation contained in our letters of August 1989. In addition, we consider Section 3.01 of the Special Terms and Conditions void and the annotated Special Terms and Conditions See comment 7 operative. wstrative Cost Calculations As indicated in the report, separate cost neutrality calculations are made for benefits and administrative costs. These are combined for each federal department's programs to establish a net savings or liability. Our methodology for establishing administrative cost neutrality for Food Stamps, Medicaid, and AFDC was submitted on May 3, 1988. Our submission of cost neutrality figures for the first three quarters of FIP contained calculations according to this formula. Your report'indicates that we had not received a formal response to this proposed methodology. As of March 1990, this See comment 8 continues to be the case. We anticipate that our next submission will have some additional adjustments to administrative costs not anticipated or included in the 1988 methodology. These changes will have to be negotiated with the federal representatives. (C!) Summarv Thank you for the opportunity for commenting on the GAO draft report, Food Stamp Proaram: Observations on Achievincr FIP Cost Neutrality and Other Matters (GAO/RCED-90-84). We appreciate the spirit of fairness and impartiality with which GAO staff undertook their review of FIP cost neutrality issues and complement them on the technical quality of the document. See comment 9. The draft report provides an excellent overview of many of the methodological and data issues connected with measuring Food Stamp cost neutrality for the Washington State Family Independence Program. Since we also have cost neutrality responsibilities related to changes we have made in the AFDC and Medicaid programs under FIP, many of the.issues which you discuss in your report have relevance beyond the Food Stamp program alone. Y Page 73 GAO/RCED-90-84 Washington’s Family Independence Program Appendix III CommentaPromtheStrfaofWaAngton 1 Mr. John W. Hannan March 16, 1990 Page 12 We believe that it would be useful for us to' sit down with representatives of the U.S. Departments of Health and Human Services and Agriculture to consider in more detail the quertions which you raised. Such discussions should begin with a recognition See comment 10. by federal representatives of their Departments' responsibilities for participation in the cost of caseload growth which they incurred under sections 1509 and 9121 of the Omnibus Budget Reconciliation Act of 1987. Enclosures Page74 GAO/lUXD-~Was~n*~F8mily IadepeadenceRogram Appendix Ill Comments From the State of Wahington The following are GAO'S comments on Washington State’s letter dated March 16, 1990. 1. We have included this information on pages 6 and 40 of the report. GAO Comments: 2. We have included a summary of this information on pages 40 and 41 of the report. The state’s enclosure has not been included in the report. 3. We have included a summary of this information on pages 41 and 42 of the report. 4. We have included this information on page 42 of the report. 6. We have included this information on pages 39 and 40 of the report. 6. We have included a summary of this information on pages 41 and 42 of the report. 7. We have included a summary of this information on pages 41 and 42 of the report. 8. We have included this information on pages 22,45 and 52 of the report. 9. We have included this information on pages 6 and 38 of the report. 10. We have included this information on pages 42 and 43 of the report. Page 76 GAO/RCED4044 Washington’s Family Independence Program Appendix IV Major Contributors to This Report 41111111_______‘---------- Resources, Peter M. Bramble, Jr., Assignment Manager Community, and June M. Foster, Staff Evaluator Michelle A. Gambone, Senior Social Science Analyst Economic Development Division, Washington, D.C. Alvin S. Finegold, Evaluator-in-Charge Seattle Regional Office Jane A. Dunkel, Staff Evaluator (028281) Page 76 GAO/RCED-90-84 Washington’s Family Independence Program ‘I’t~ltq,hont! 202-275-6241 ‘I’ht* first. five twpies of each report, art’ free. Aclditioml copit++ art* p1;2.00t%ictt. Ordt~rs tttttst be prepaid by cash or by check or money cwclt~r matIt* out, I.0 t.tte Stiperintettdettt~ of Docurtterrt,s.
Food Stamp Program: Achieving Cost Neutrality in Washington's Family Independence Program
Published by the Government Accountability Office on 1990-06-28.
Below is a raw (and likely hideous) rendition of the original report. (PDF)