New York State Department of Social Services, DAB No. 1358 (1992) Department of Health and Human Services DEPARTMENTAL APPEALS BOARD Appellate Division SUBJECT: New York State DATE: September 30, 1992 Department of Social Services Docket No. 91-14 Audit Control No. A-02-87-02016 (A-02-90-02019) Decision No. 1358 DECISION The New York State Department of Social Services (State) appealed a determination by the Administration for Children and Families (ACF). ACF disallowed $92,115,289 in federal financial participation (FFP) claimed by the State under Title IV-E of the Social Security Act (Act). The disallowance was based on an audit of State claims for FFP in foster care maintenance payments made by the New York City Department of Social Services (City) during the period October 1, 1983 through September 30, 1985. The audit was performed by this Department's Office of Inspector General using statistical sampling methods. After considering the State's response to the audit, ACF found that the City had made errors in determining Title IV-E eligibility, or had failed to document eligibility, in 186 of 257 sample cases. ACF determined the amount of maintenance payments associated with each finding and then projected the sample results to calculate a disallowance amount of $64,123,732 in FFP in maintenance payments. ACF also disallowed $27,991,567 in FFP which ACF found was claimed for administrative costs associated with unallowable maintenance payments. The State requested that we address the threshold legal issues in the case, before addressing ACF's findings for individual sample cases, and we granted that request. Thus, we address in this decision general arguments raised by the State challenging ACF's authority to disallow maintenance payments for children who were ineligible under Title IV-E (or not documented as eligible); ACF's authority to base a disallowance on statistical sampling methods; and the method ACF used to calculate the disallowance of administrative costs. In support of these challenges, the State raised numerous complicated and sophisticated arguments. We conclude that there is some merit to the State's arguments on the administrative costs. On the other hand, the State's arguments on the maintenance payments issues appear designed primarily to obfuscate, through convoluted reasoning and mistated or incomplete quotations, what is at bottom a fundamental proposition. In a grant program such as Title IV-E, the State's entitlement to funds is limited to payments authorized by statute, which meet the conditions established by the statute and implementing regulations. If the State has claimed FFP in excess of the amount to which the State is entitled, it must repay the funds. In determining the amount a state has been overpaid, ACF is reasonable in using any audit technique, consistent with its own policies, which produces reliable evidence of the overpayment amount, including valid statistical sampling. Below, we first summarize our conclusions on the major issues raised by the State. We then provide a detailed analysis of the State's arguments. Summary of our decision We uphold the disallowance of foster care maintenance payments, subject to downward adjustment based on our further review of ACF's findings in individual sample cases, for the following reasons. o Section 474(a)(1) of the Act authorizes federal funding for foster care maintenance payments only if those payments meet certain conditions specified in the Act. ACF regulations make clear that FFP is available only if those conditions and ACF regulatory requirements are met. FFP paid to the State for maintenance payments which do not meet statutory and regulatory conditions constitutes an overpayment to the State to be adjusted under section 474(d) of the Act. ACF's authority to disallow is thus derived from the Act. Furthermore, the implementing regulations contemplate that this enforcement mechanism will be used. o Section 471(b) (which authorizes the Secretary to prospectively withhold part or all Title IV-E funding based on a finding of substantial failure to comply with a state plan) is not the only mechanism Congress established to enable the Secretary to enforce Title IV-E requirements. Funding limitations are another enforcement mechanism available to the Secretary under Title IV-E, as the Supreme Court has recognized. o The State's position that it is entitled to FFP in any foster care payment it makes, so long as it is substantially complying with its approved Title IV-E state plan, is contrary to the plain language of the Act and well-established grants law. Having an approved plan makes a state eligible for payments, but a state's entitlement to payments is limited under section 474(a) of the Act. Adopting the State's position would frustrate congressional intent in Title IV-E. Congress placed restrictions on Title IV-E funding for maintenance payments as part of a scheme to prevent foster care "drift" and to ensure quality care for foster children. o Limits on payments established by the Act and regulations must be met by the State as a condition for FFP, even if they can be characterized as "technical" as the State alleged. In any event, the requirements for the most part are directly related to accomplishing the Act's purposes. The State's allegations here indicate that the auditors may have applied a few of the requirements in a manner inconsistent with ACF policy. We know of no basis, however, for excusing a state's failure to meet clear requirements, properly applied. o Statistical sampling performed in accordance with scientifically accepted rules and conventions produces reliable evidence of the amount of unallowable costs claimed by a state. Use of statistical sampling as a basis for disallowing costs has been consistently upheld by this Board and the courts as appropriate where, as here, a case-by-case review would be impracticable. o The State had timely and adequate notice that ACF might disallow Title IV-E costs based on statistical sampling. ACF was not required to use notice and comment rulemaking to promulgate its policy on statistical sampling. The policy does not affect the State's entitlement to funding, nor its obligation to meet applicable requirements. It simply relates to an audit technique used to generate evidence on the amount of unallowable costs claimed by a state. The State has had ample opportunity to challenge the statistical methods used and will have a further opportunity to contest the findings in individual sample cases. o ACF presented evidence to show that its sampling was performed in accordance with scientifically accepted rules and conventions, and that it could say with 95% confidence that the true value of unallowable maintenance payments was at least the disallowed amount. The State presented no expert testimony or other evidence to show that the methods were invalid, but challenged instead whether the methods were consistent with ACF and OIG sampling policies -- in particular, those relating to sample size. The State, however, misinterpreted the scope and effect of those policies. Moreover, ACF bore the burden of any increase in potential sampling error from use of a smaller sample size. o ACF was not required to permit a tolerance for unallowable payments in Title IV-E, simply because such a tolerance is permitted under a system applicable to Title IV-A of the Act which involves ongoing sampling of a state's entire IV-A caseload. Even if ACF had the authority to establish such a tolerance by regulation, it would not be appropriate for us to do so by adjudication. Nor do contract principles apply here to prevent ACF from disallowing the full amount of unallowable claims. ACF properly disallowed the full amount since the State had clear notice that FFP was available only in payments which met applicable requirements. We reverse the disallowance of administrative costs for the following reasons. o The method ACF used to calculate the disallowance of administrative costs is flawed. It incorrectly assumes a proportionate relationship between the allowability of maintenance payments and the allowability of administrative costs. o ACF's method superimposes a further allocation of administrative costs on top of the allocation method approved for the State (at least with respect to part of its costs). ACF did not show that compelling circumstances exist which would warrant retroactively changing the approved allocation method. ACF is not precluded from further examining the allocation method in fact used by the State for administrative costs and devising a reasonable method of calculating the extent to which the State's overstatement of its maintenance payments claims also resulted in an overstatement of administrative cost claims. Analysis I. ACF has authority to disallow payments which did not meet federal requirements. ACF found, after examining individual sample cases, that the State had claimed FFP for foster care payments for 186 children determined ineligible for 285 separate reasons. Specifically, ACF found: o 76 children were not removed from the home of a specified relative as a result of a judicial determination or were no longer under the jurisdiction of the court or did not have a timely judicial determination, in violation of Section 472(a)(1) and (a)(2) of the [Act] and regulations at 45 CFR 1356.2(a) and (b). o 65 children had judicial determinations which did not contain evidence that the court had reached a decision to the effect that continuation in the home would be contrary to the welfare of the child, or that reasonable efforts were made to prevent removal or made it possible for the child to return home, in violation of Section 472(a)(1) of the Act and regulations at 45 CFR 1356.21(a) and (b). o 58 children were not eligible for AFDC either initially or during their placement in foster care, in violation of Section 472 of the Act and regulations at 1356.21(a). o 22 children had neither a Social Security number nor applications for a number prior to April 1, 1985, in violation of Section 402(a)(25) of the Act. o 5 case folders were not found, in violation of regulations at 45 CFR 74.21. o 6 case folders were inadequately documented to support a determination, in violation of regulations at 45 CFR 74.21. o 15 children were ineligible because responsibility for care and placement was either not imposed on NYC or was transferred to a private agency, in violation of Section 472(a)(2) of the Act and regulations at 1356.21(a). o 15 children were not removed from the home of a specified relative or were not physically removed from the home, in violation of Section 472(a) of the Act and regulations at 45 CFR 1356.21(a). o 16 children were no longer in foster care yet payments continued, in violation of Section 472(a) of the Act and regulations at 45 CFR 1356.21a and 1356.60(a)(i). o 4 children were voluntarily placed into care without a signed voluntary agreement in violation of Section 472(a)(3) of the Act and regulations at 45 CFR 1356.30(a). o 3 children were ineligible because they were not United States citizens or legal aliens in violation of Section 402(a)(33)(B) of the Act. Disallowance letter at 3-4, State Exhibit (Ex.) 1. These findings were based on requirements set out in section 472 of the Act, or on requirements for Title IV-A eligibility, incorporated by reference into section 472. The payments were made in fiscal year (FY) 1984 or FY 1985. The State argued that Title IV-E does not expressly authorize a retroactive disallowance of FFP based on a failure to meet the requirements of section 472. Rather, the State maintained, ACF's authority was limited to withholding FFP as an adjustment of quarterly advances of federal money under a national allocation formula, or to withholding future payments to a state after finding that a state had substantially failed to comply with provisions of the state plan or that the state plan no longer complied with the Act. State Brief (Br.) at 34-35. The State noted that ACF had not questioned the State's compliance with its State plan in general or with any particular plan provisions. Consequently, the State argued, it was entitled, under section 474(a) of the Act, to its share of Title IV-E funding. The State argued that Title IV-E was different from the programs under Titles IV-A and XIX since Congress had provided mechanisms in these programs which allowed recovery from beneficiaries or from providers of services. According to the State, no such parallel recovery system is available under Title IV-E. Thus, the State reasoned, since Congress did not intend that states could recoup misspent federal funds from foster care children, it consequently could not have authorized this disallowance. State Br. at 35-37. For reasons explained in this section, we conclude that these arguments have no merit. These arguments incorrectly assume that, absent specific reference to a "disallowance authority" in Title IV-E, the State may claim FFP in any foster care payments (up to the amount allocated to the State), so long as it has an approved State plan and is not substantially failing to comply with that plan. In establishing a grant program like Title IV-E, however, Congress is exercising its spending authority under Article 1, section 8, of the United States Constitution. Congress appropriates funds only for authorized purposes. A grantee's entitlement to federal funds does not extend beyond FFP authorized in the grant statute. Essentially, a federal agency's disallowance of costs is based on a determination that the grantee has claimed funds to which it is not entitled under the grant statute, implementing regulations, and other applicable grant conditions. If the grantee has already received the funds in excess of the amount to which it is entitled, a disallowance is necessarily retroactive. Thus, the key question to ask is whether the State may retain FFP even if it did not meet section 472 requirements. We conclude that Congress clearly intended section 472 to act as a limit on the State's entitlement to FFP. We further conclude that the Act gives ACF the authority to determine the amount of FFP paid to the State in excess of the State's entitlement and to reduce the grant award by the amount of such overpayment. Below, we first discuss the history, statutory scheme, and language of Title IV-E. We then address ACF's regulations. A. Title IV-E of the Act establishes a grant program with specific purposes. The Adoption Assistance and Child Welfare Act of 1980, Public Law 96-272, was enacted on June 17, 1980, as part of a comprehensive restructuring of certain Social Security Act programs. This restructuring was based on the belief that the public child welfare system responsible for serving dependent and neglected children had become a holding system for children living away from their parents. Congress intended "to lessen the emphasis on foster care placement and to encourage greater efforts to find permanent homes for children either by making it possible for them to return to their own families or by placing them in adoptive homes." S. Rep. No. 336, 96th Cong., 1st Sess. 1 (1979). Among other changes, this legislation established a new federal program, Title IV-E, Federal Payments for Foster Care and Adoption Assistance. 42 U.S.C  670 et seq. The foster care component of the Aid to Families with Dependent Children (AFDC) program, under Title IV-A of the Act, was replaced by Title IV-E, effective no later than October 1, 1982. Title IV-E authorizes appropriations to enable a state "to provide, in appropriate cases, foster care and adoption assistance for children who otherwise would be eligible for assistance" under the state's Title IV-A program or, in the case of adoption assistance, would be eligible for benefits under the Supplemental Security Income program found at Title XVI of the Act. The appropriations made available for Title IV-E are sums "necessary to carry out the provisions of" Part E, and are to be used to make payments to any state with an approved state plan. Section 470 of the Act. Section 471(a) of the Act sets out state plan requirements. Section 471(b) provides for approval of any complying plan, but authorizes the Secretary to withhold or reduce further payments to a state if the Secretary finds that a state plan no longer complies, or that in the administration of the plan there is a substantial failure to comply with the plan. Section 472 of the Act describes the foster care maintenance program, including the children on whose behalf such payments must be made. Basically, they must be children for whom the state has the responsibility for care and placement, who would otherwise be eligible for Title IV-A, and who are removed from the home of a specified relative as a result of a judicial determination, meeting specified requirements, or a voluntary placement agreement. Coverage of children removed pursuant to a voluntary placement agreement is subject to additional conditions set out in subsections (d), (e), and (f). Section 473 describes the adoption assistance program. Section 474(a) of the Act establishes states' entitlement to federal funding for three separate categories of Title IV-E expenditures: foster care maintenance payments "under section 472" of the Act for children in foster family homes or child care institutions; adoption assistance payments under section 473 of the Act; and expenditures found necessary by the Secretary for the proper and efficient administration of the state plan. Total payments in a fiscal year may not exceed the amount allotted to a state under section 474(b) for that year (except in certain specified circumstances). Section 475 contains definitions relevant to Title IV-E, including a definition of "foster care maintenance payments." The State's arguments on the statutory language focus on the state plan requirements in section 471. Having an approved state plan, however, only makes a state "eligible" for payments. Section 471(a). The amount of funding to which a state is "entitled" is established by section 474(a). Section 474(a)(1) provides that a state with an approved plan is entitled to payment each quarter, at a specified rate, for "the total amount expended during such quarter as foster care maintenance payments under section 472 for children in foster family homes or child-care institutions." (Emphasis added.) The State is not entitled to funding for payments which are not "under 472" (nor for payments which do not meet the definition of "foster care maintenance payments" in section 475 or which are not for children in foster family homes or child-care institutions). Thus, contrary to what the State argued, the plain wording of the statute conditions FFP on the expenditures meeting the requirements set out in section 472. Reading the statute to authorize funding in payments not meeting these requirements would be contrary to the intent of the Adoption Assistance and Child Welfare Act of 1980. In order to achieve the purposes of preventing unnecessary placements and protecting children placed in foster care, Congress narrowly circumscribed the types of payments in which it would participate. Moreover, Congress tied eligibility for Title IV-E to Title IV-A, by incorporating the eligibility conditions of IV-A into section 472. To permit a state to ignore those limits would frustrate Congressional intent and expand the program beyond the purposes stated in section 470. Section 474(a) implicitly gives ACF the authority to disallow. A disallowance is essentially a determination that a grantee is claiming or has claimed federal funds to which it is not entitled under the statute, or other program requirements. ACF could not properly administer the program in accordance with the statute if ACF could not act to ensure that federal funds are spent only for authorized purposes. B. The Act provides for "retroactive" adjustments for overpayments to a state. Section 474(d)(2) of the Act provides additional support for ACF's assertion that it has authority to disallow funds claimed in excess of the amount to which the State is entitled under Title IV-E. Section 474(d) establishes a scheme of quarterly grant awards to states, based on estimates adjusted by the Secretary. Subsection (d)(2) provides -- The Secretary shall . . . pay to the State, in such installments as he may determine, the amounts so estimated, reduced or increased to the extent of any overpayment or underpayment which the Secretary determines was made under this section to such State for any prior quarter and with respect to which adjustment has not already been made under this subsection. The State argued that this section supported its contention that, under Title IV-E, ACF is limited to prospective remedies against the states. This view is contrary to the plain language of the section, which refers to overpayments for "any prior quarter." While any adjustment under section 474(d)(2) would be prospective in the sense that it would be made to an award for a future quarter, any such adjustment would also be retrospective in the sense that it would relate to a prior quarter. The plain language of section 474(d)(2) is also a basis for rejecting the State's contention that adjustments are authorized only if a state exceeds its total allotment (or allocation) of funds. Funds are allotted under section 474(b) for a fiscal year. If section 474(d)(2) was limited to only one particular type of overpayment, it would not refer to "any overpayment . . . for any prior quarter." Interpreting this authority as narrowly as the State did would also preclude the Secretary from making adjustments if the estimated award to the State for a quarter was greater than the State's actual expenditures during the relevant quarter. Clearly, Congress could not have intended such a result. Moreover, as ACF noted, judicial interpretations of similar language in other federal grant programs leads to a conclusion that ACF has the authority to disallow retroactively for payments which do not meet federal requirements. In Maryland Dept. of Human Resources v. Dept. of Health and Human Services, 763 F. 2d 1441 (D.C. Cir. 1985), the court found that virtually identical language in Title XX (at section 2002(b) of the Act) evidenced the government's "statutory right of recovery" and the state's corresponding indebtedness to the government for misused funds. 763 F. 2d at 1444, 1457. Similarly, in Georgia Dept. of Human Resources v. Califano, 446 F. Supp. 404 (N.D. Ga. 1977), the district court construed identical language contained at section 1903(d)(2) of the Act as authorizing the Secretary to question a state's claims and to reimburse a state only for proper expenses. 446 F. Supp. at 412. In Bell v. New Jersey, 461 U.S. 773 (1983), the Supreme Court interpreted a similar provision from the Elementary and Secondary Education Act of 1965 as evidence of the government's statutory right to recover misspent federal grant funds. 1/ The Court rejected the same argument which New York presented here, i.e., that the federal government was limited to prospective relief. Rather, the Supreme Court noted that the statute's plain language recognized the federal government's right to recover funds not meeting program requirements. 461 U.S. at 780-783. The judicial analyses of statutory language analogous, if not identical, to section 474(d)(2) of the Act support a conclusion that ACF has authority to disallow retroactively where it finds an improper expenditure of Title IV-E funds. The basic scheme set out in Title IV-E is typical of state plan grant programs (sometimes referred to as "entitlement" or "formula grant" programs), and is parallel to the language of similar grant programs under the Act such as Titles IV-A and XIX. While there are some differences in the structure of Title IV-E, this can be explained by the fact that Title IV-E sets up two programs -- one for foster care maintenance payments and one for adoption assistance. We find no merit in the State's argument that Title IV-E is distinguishable from Titles XIX and IV-A on the ground that the State cannot recover improper maintenance payments from foster care children, whereas it can recover erroneous Medicaid payments from providers and erroneous AFDC payments from recipients. We see no reason why this factor, even if true, should permit the State to retain FFP in excess of what is authorized under the statute. 2/ Indeed, the State said here that children in foster care not eligible for Title IV-E (or for emergency assistance payments) would be automatically covered by the State's own foster care program. Thus, the State is no worse off from a retroactive adjustment of FFP than it would have been if it had covered the maintenance payments for ineligible children with State funds initially (as it should have). We also reject the State's argument that Title IV-E is distinguishable because it establishes a block grant program. A block grant program is one in which a state has broad discretion to shift funds between related categories. The State pointed to nothing in the wording or history of Title IV-E to support its view that Title IV-E establishes a block grant program. Title IV-E is not listed as a block grant program under Department regulations at 45 C.F.R. Part 96 governing such programs. Even if Title IV-E were a block grant, however, this would not broaden the State's entitlement beyond funding authorized in the statute. Block grant programs are subject to audit, and a state must repay amounts found to have been expended improperly. 45 C.F.R.  96.32. Finally, the State pointed out that section 1116(d) of the Act does not specifically provide that the Secretary shall reconsider disallowances under Title IV-E, although it does specifically mention Title IV-A and other Titles of the Act. This omission does not support the State's position here, however. First, section 1116(d) addresses states' right to a reconsideration process when a disallowance is taken. 3/ Absence of specific mention of a right to reconsideration of a disallowance does not automatically imply a lack of authority to disallow. Second, ACF has treated as an oversight Congress' failure to amend section 1116 to add Title IV-E, and has in its regulations provided to the states the same right of reconsideration of a disallowance as states have under other state plan grant titles of the Act, that is, the right to appeal a disallowance to this Board. See 45 C.F.R.  1355.30(a); 45 C.F.R. Part 16, Appendix A, B. C. ACF regulations treat the provisions of section 472 as conditions for FFP. Title IV-E regulations at 45 C.F.R. subchapter G, clearly treat section 472 of the Act as establishing conditions for FFP. Section 1356.60(a)(1) provides that FFP -- is available to States under an approved title IV-E State plan for allowable costs in expenditures for: (i) Foster care maintenance payments . . . made in accordance with 45 CFR 1356.20 through 1356.30 of this part, section 472 of the Act and section 102(d) of Pub. L. 96-272, the Adoption Assistance and Child Welfare Act of 1980 . . . . This section was enacted as part of the "fiscal requirements" for Title IV-E, described as the "fundamental requirements for receipt and expenditure of Federal financial participation" under the program. 47 Fed. Reg. 30922 (July 15, 1982). No provision of these regulations makes FFP available in expenditures for payments for foster care services not made in accordance with section 472 of the Act. Section 1355.30, enacted at the same time, makes applicable to Title IV-E "regulations . . . which are generally applicable to all grants and are particularly relevant to programs with FFP in State expenditures." 47 Fed. Reg. 30923. These regulations include the Board procedures used for reconsideration of Social Security Act disallowances. In sum, the statute and regulations authorize FFP only in payments made in accordance with section 472 of the Act and implementing regulations, and permit retroactive adjustments to grant awards for FFP paid in excess of the authorized amount. A determination that a state has been overpaid FFP is a disallowance, subject to reconsideration by the Board. II. ACF's authority to disallow is separate from its authority to take action based on substantial failure to comply. As mentioned above, the State argued that ACF could enforce the Act only by invoking section 471(b). The State also argued that, by using a disallowance mechanism here rather than finding a substantial failure to comply, ACF was depriving the State of its right to a non-compliance hearing. In its reply brief, the State argued that its position was supported by a recent Supreme Court decision. For reasons explained in this section, we conclude that these arguments have no merit. A. ACF did not find a substantial failure to comply, nor was it required to. Section 471(b) of the Act provides that -- in any case in which the Secretary finds, after reasonable notice and opportunity for a hearing that . . . in the administration of the plan there is a substantial failure to comply with the provisions of the plan, the Secretary shall notify the State that further payments will not be made to the State under this part, or that such payments will be made to the State but reduced by an amount which the Secretary determines appropriate, until the Secretary is satisfied that there is no longer any such failure to comply, and until he is so satisfied he shall make no further payments to the State, or shall reduce such payments by the amount specified in his notification to the State. This section thus permits the Secretary (through ACF) to withhold all or part of the amount that would otherwise be paid to a state under the system of quarterly payments set out in section 474. Thus, it affects funding to the state for future quarters. The purpose of such action is to bring a state into compliance. Similar provisions appear in other Titles of the Act establishing state plan grant programs. Yet, we know of no administrative or court interpretation that such a provision must be viewed as an exclusive enforcement mechanism absent a specific reference to disallowance authority. Indeed, the Georgia court rejected such an interpretation. While the court's decision was based in part on section 1116(d) of the Act, the court also found that the Secretary's right to disallow was implied from the overpayment provision in section 1903(d) and from the common law right to recover federal funds improperly or illegally disbursed. Georgia, 446 F. Supp. at 412, citing Mount Sinai Hospital of Greater Miami v. Weinberger, 517 F.2d 329 (5th Cir. 1975). The State did not argue that ACF had made any finding that the State was substantially failing to comply with its State plan. In fact, the State argued that its failures related to "minor" or "technical" requirements. While we do not agree with the State's characterization of section 472 requirements as "minor" or "technical," we nonetheless consider it within ACF's discretion not to invoke the substantial compliance remedy. Providing for foster care maintenance payments in accordance with section 472 is only one of many state plan provisions required by section 471(a). Moreover, ACF made no finding that the claiming errors found for sample payments made in FYs 1984 and 1985 in New York City arose from any noncompliance with the State plan which was continuing to the present so that prospective cessation of part or all of the State's Title IV-E funding is necessary to force the State to comply. This draconian enforcement mechanism has only rarely been used by this Department in any program. Past Board decisions have recognized the traditional distinction made in the Act between disallowances and compliance actions. See New Jersey Dept. of Human Services, DAB No. 259, at 6-20 (1982); Massachusetts Dept. of Public Welfare, DAB No. 438, at 22-27 (1983); Colorado Dept. of Social Services v. Dept. of Health and Human Services, Civil No. 92-F-653 (D Co., filed July 17, 1992), affirming Colorado Dept. of Social Services, DAB No. 1277 (1991). We will not repeat those lengthy analyses here. However, we note that in Colorado the courts upheld the Board's view that a disallowance action was appropriate for recovering expenditures made in the past, even if the basis for disallowance was failure to meet state plan requirements, so long as the federal agency did not find that there was a continuing and substantial failure to comply. Moreover, in New Jersey v. Dept. of Health and Human Services, 670 F.2d 1284 (3rd Cir. 1981), the court found that the procedures relevant to compliance actions should have been invoked by the Secretary rather than disallowance procedures, but found that this was harmless since "nothing . . . precludes our treating the [DAB] action . . . as . . . rendered after an `opportunity for hearing' sufficient to satisfy the requirements" of section 1116 of the Act for compliance actions. At note 13. In New York State Dept. of Social Services, DAB No. 1246 (1991), the Board stated -- the Secretary's compliance remedies and disallowance remedies are not mutually exclusive and serve different purposes. . . . The compliance remedy grants the Secretary sweeping powers when a state is in substantial noncompliance with program standards. In a compliance action the Secretary is authorized to terminate all funding to the state in order to give it compelling incentive to bring its program back into compliance. In keeping with the coercive nature of the remedy, the amount of money the Secretary may withhold is not necessarily related to the actual costs of the noncompliance at issue. In contrast, a disallowance action provides a specific and focused remedy pursuant to which the federal government may disallow precisely identified amounts which were not spent in accordance with program requirements. DAB No. 1246 at 5. While the amount disallowed here is very large, that does not change the essential nature of the federal action as one to recover FFP erroneously paid in past expenditures not made in accordance with program requirements. We also note that the main reason states have argued that compliance procedures should be used is that, under section 1116(a)(3) of the Act, final determinations in compliance proceedings under listed sections of the Act may be appealed directly to the courts of appeals. Section 1116(a)(3) of the Act, like the provision in section 1116(d), was not amended to refer to Title IV-E. ACF regulations treat this as an oversight and adopt procedures at 45 C.F.R. Part 213 for compliance actions in Title IV-E, as for other titles. We think that is the correct interpretation. The State's arguments are inconsistent, however, since they would distinguish Title IV-E from other state plan grant titles for purposes of disallowances, but would treat the titles the same for purposes of compliance actions. C. The Supreme Court decision on which the State relied supports ACF rather than the State. In its reply brief, the State argued that the ACF position that the Act authorized both disallowances and substantial compliance actions was rejected by the Supreme Court in Suter v. Artist M., U.S. , 112 S. Ct. 1360 (1992). The State argued in effect that the analysis in Suter supported the State's view that having an approved plan entitled the State to any federal funding the State claimed and that the only mechanism for enforcing the Title IV-E requirements (other than the provisions of section 472(e)) is through a compliance action. The State's arguments in our view seriously misrepresent both the scope of the Suter decision and what the Supreme Court said. The issue in Suter was whether the Child Welfare and Adoption Assistance Act of 1980 conferred upon the child beneficiaries of the Act a right to enforce the requirement that the State make "reasonable efforts" to prevent a child from being removed from his home, and once removed to reunify the child with his family. The petitioner children in Suter argued that section 471(a)(15) of the Act conferred a private right of action. The Supreme Court rejected this argument based primarily on its analysis of section 471(a) as merely requiring the State to have a plan approved by the Secretary containing the listed features, including a provision that the State would make reasonable efforts in each case. The Supreme Court did not directly address the issue of whether ACF had the authority to disallow payments not made in accordance with section 472. Contrary to what the State argued, nothing in Suter implies that having a complying plan is sufficient to entitle the State to any foster care payments the State claims. Nor does Suter stand for the proposition that the compliance mechanism in section 471(b) is the only enforcement mechanism available to the Secretary. The Suter decision specifically mentions two enforcement mechanisms for the reasonable efforts requirement. After referring to the substantial compliance provision in section 471(b) as one of the other sections of the Act providing enforcement mechanisms, the decision then states: The Act also requires that in order to secure federal reimbursement for foster care payments made with respect to a child involuntarily removed from his home the removal must be "the result of a judicial determination to the effect that continuation [in the child's home] would be contrary to the welfare of such child and (effective October 1, 1983) that reasonable efforts of the type described in [section 471(a)(15)] of this title have been made."  [472(a)(1)]. 112 S. Ct. at 1368. Thus, the Supreme Court recognized that section 472 of the Act established conditions for federal funding. This section could reasonably be considered as creating an "enforcement mechanism" only if the Secretary has the authority to deny funding for foster care payments not meeting the section's requirements. The State nonetheless argued: The reasonable efforts requirements of  471(a)(15) are expressly incorporated in  472(a)(1). Because  471(a)(15) is not a precise requirement providing notice to states that the provision created enforceable rights not expressly stated, it follows that  472(a)(1) is similarly not a precise requirement creating enforcement rights not expressly stated. State Reply at 30. We disagree. First, we note that the issue here is not whether section 472(a)(1) creates rights enforceable by children; the issue is whether the State is entitled to FFP in payments not meeting section 472 requirements. In any event, it does not logically follow that a result based on wording in one section of the Act applies to a second section of the Act simply because the second section incorporates the first. As discussed above, section 471(a) of the Act contains state plan requirements. On the other hand, section 474(a) sets out a state's entitlement to payment and limits payment to "foster care maintenance payments under section 472." Thus, there is an express statement in the Act limiting FFP to payments made under section 472. 4/ Since the condition for funding is clear, the Supreme Court's decision in Pennhurst State School and Hospital v. Halderman, 451 U.S. 1 (1981), is inapposite here. 5/ The State similarly reads too much into the analysis in Suter rejecting the petitioner children's analogy between the reasonable efforts provision in section 471(a)(15) and the reasonable rate requirement in section 1902(a)(13) of the Act, considered in Wilder v. Virginia Hospital Assn., 496 U.S. 498 (1990). 6/ The Supreme Court said that "each statute must be interpreted by its own terms." 112 S. Ct. at 1367, n.8. The State argued, based on this statement, that it is "clear from the Supreme Court's interpretation that the Secretary's ability to recoup payment under Title XIX does not mean that a similar ability is afforded under Title IV-E." State Reply at 32. We agree as a general proposition that each statute must be interpreted on its own terms. On the other hand, parallel provisions in different titles of the same Act would generally be assumed to have the same meaning, absent evidence to the contrary. Haig v. Agee, 453 U.S. 280 (1981); Northcross v. Bd. of Educ. of Memphis City Schools, 412 U.S. 427 (1973); Forman v. United States, 767 F.2d 875 (Fed. Cir. 1985); see generally Sutherland Stat. Const., sections 51.01-51.03 (4th ed.). Unlike the provisions of sections 471(a)(15) and 1902(a)(13) distinguished in Suter, the provisions in sections 474(d) and 1903(d) regarding adjustments for overpayments contain parallel language. The State pointed to nothing in the context, legislative history, or administrative interpretation of these overpayment sections which would indicate any difference in intent. Finally, we reject the State's argument that the Supreme Court's finding in Suter that compliance with the reasonable efforts requirement in section 471(a)(15) of the Act will vary with the circumstances of each individual child somehow precludes the use of sampling by the OIG. The Supreme Court found that allowing private actions would interfere with the states' discretion to decide what reasonable efforts to provide in each individual case. The audit here does not interfere with that discretion. The audit findings related to reasonable efforts were not based on section 471(a)(15), but on section 472(a), which requires a judicial determination, by a state court, that reasonable efforts have been made. The auditors found for some children either that no such determination had been made, or that the State had not documented that determination. This is far different from a decision which would dictate to a state what services it must provide to a particular child in the first instance. In sum, withholding future funds based on a finding of substantial failure to comply is not the only mechanism Congress established, and which ACF may use, to enforce the Act's requirements. III. ACF is not precluded from disallowing the full amount of any overpayment to the State. The State argued that the appropriate "penalty" for certain alleged eligibility violations is not total loss of FFP. The State argued that ACF found children ineligible, regardless of the significance of the particular criteria ACF found were not met, and disallowed the total amount of FFP claimed for the children during the audit period. According to the State, the disallowance is "based in large part on the State's failure to comply with certain eligibility criteria that are most properly classified as technical requirements." State Br. at 49. The State argued that, by exacting a penalty as severe as total loss of FFP for children who fail to meet even the most minor of the eligibility criteria, ACF defeats Congress goal of aiding and improving the states' foster care programs. State Br. at 50. The State argued that exacting the maximum penalty for every violation, without regard to the significance of the transgression, is an approach disfavored in our system of law. State Br. at 51. The State analyzed contract law principles, concluding that, since "there are no provisions in Title IV-E that explicitly mandates forfeiture of federal funding for failure to meet minor eligibility provisions, the Agency ought not to be allowed to apply the statute so as to work a forfeiture where the State's error is not of the kind that would warrant full withholding of FFP." State Br. at 57. Like other arguments raised by the State, these arguments depend on State terminology -- such as "penalty" or "forfeiture" -- used inappropriately here. The State is entitled to FFP only as authorized by Congress; thus, requiring the State to repay funds it claimed improperly cannot fairly be characterized as either a "penalty" or a "forfeiture" (especially for payments which it otherwise would have covered entirely with its own funds from the beginning). See Bennett v. Kentucky Dept. of Education, 470 U.S. 656, 662-663 (1984); Bell v. New Jersey, 461 U.S. 773, 782 (1983). As discussed in this section, we also conclude that the State's arguments have no merit because they are based on contract principles which do not apply in the grant context, and because ACF reasonably determined that all requirements specified in the Act should be enforced. A. The contract principles relied on by the State do not apply here. The State relied here on the "venerable principle of contract law that failure to satisfy a minor provision in a contract does not automatically entitle the other party to terminate the contract and withhold all consideration." State Br. at 51. The State said that this doctrine of substantial performance was adopted by courts "for the very purpose of mitigating the harsh effects of requiring a party to forfeit all rights under a contract for the slightest breach." State Br. at 51. The State asserted that this doctrine provided a suitable analogy since the Supreme Court had stated in Pennhurst that funding legislation "is much in the nature of a contract: in return for federal funds, the States agree to comply with federally imposed conditions." 451 U.S. at 17. The State argued: In determining whether the State has "materially breached" the Title IV-E "contract" and thus relieved the Agency of its obligation to provide federal funds in particular cases, the Agency cannot simply look to whether the State has failed to comply with one of many eligibility provisions, however minor, or has failed to document eligibility precisely in the manner required by the Agency. This would be tantamount to one party withholding his entire consideration for a minor violation or deviation by the other, the approach clearly disfavored under contract law. Rather, the Agency must determine whether the State has substantially performed its obligations under the IV-E "contract," thereby entitling the State to its promised consideration (FFP) with or without some offset, if appropriate, to account for the minor failure. State Br. at 52-53. Finally, the State argued that ACF's "choice of remedy . . . must be rejected because it bears no relation to the injury sustained by the Agency as a result of the State's failure to comply with eligibility criteria." State Br. at 55. These arguments are all premised on reading the analogy of grants to contracts in Pennhurst as meaning that principles of contract law should be applied wholesale to grants. The State cited nothing in case law or elsewhere that supports this proposition, however, and we know of no court that has gone that far. Congress itself has specifically distinguished in the Federal Grant and Cooperative Agreement Act of 1977, Public Law 95-224, between federal assistance, including grants, and federal procurement contracts. Surely, there are some ways in which a grant is like a contract, but there has been considerable debate about the extent to which the analogy can be applied. Even those who advocate seeing a grant as a contract, however, recognize that a fundamental difference is that a grant is governed by statute, whereas a contract is not. See R. Cappalli, Federal Grants and Cooperative Agreements (1982); P. Dembling and M. Mason, Essentials of Grant Law Practice (1991). ACF does not have an "obligation" to provide funds for expenditures not meeting the statute's requirements. Indeed, ACF is not authorized to provide such funds, and, where ACF determines that the State has been overpaid (and that determination is upheld after reconsideration), ACF must adjust the grant award by the amount of the overpayment, under section 474(d). The State's arguments also mischaracterize ACF's actions here, in the following ways: o The State implied that ACF is terminating the grant or withholding all funding under the grant. This is simply not true. ACF is simply disallowing the amount of FFP overpaid to the State for certain expenditures made by the City for ineligible children during a particular period. o The State implied that the disallowance bears no relationship to "the injury" caused to ACF by the State's failures. To the contrary, the "injury" here is properly measured by the amount of FFP the State claimed in excess of the amount to which it was entitled, and ACF's disallowance is a conservative estimate of that amount. o The State ignored the fact that ACF found that the State had failed to meet fundamental program requirements, such as the requirement that there be a judicial determination. Moreover, ACF found that, in many of the sample cases, the State had failed to meet more than one requirement. Finally, as discussed next, we also reject the State's premise that the requirements at issue here are "minor or technical" requirements that should not affect the availability of FFP. B. The State's characterization of particular prerequisites for federal funding as "minor or technical requirements" which should not affect the availability of FFP has no merit. The State asserted that ACF's disallowance of more than $64 million in FFP claimed for foster care maintenance payments was based "in large part on the State's failure to comply with certain eligibility criteria that are most properly classified as technical requirements." State Br. at 49. The State asserted that "[a] loss of federal funding of such magnitude will severely impede the State's and [City's] ability to provide necessary foster care services to needy children." State Br. at 50. The State asserted that many Title IV-E eligibility requirements "do not go to the essence of the purposes of the [Act]" so that a failure to meet any of these requirements does not take a child out of the group that Congress intended to protect. State Br. at 54. The State asserted that to deny FFP for such technical violations "infringes on the underlying purpose of the [Act] -- to provide funds to states for foster care services." State Br. at 54 (emphasis in original). The State described this disallowance as a harsh penalty for "minor or technical violations" of certain Title IV-E eligibility requirements. State Br. at 69. The State presented general arguments for why ACF could not disallow all FFP claimed for foster care maintenance payments for cases where (1) the State had failed obtain or apply for a social security number; (2) the judicial determination did not include certain "magic words"; (3) the State had transferred the responsibility for care and placement of a child to a private agency; (4) a child had resided with a relative before the relative was certified as the foster care provider; and (5) the State failed to obtain a judicial determination within 180 days of a voluntary placement. We conclude that the State's arguments misstated the underlying purposes of the Act. The object of Title IV-E was not just to provide funds to the states for foster care. The State would have us conclude that Title IV-E is a form of revenue sharing and that so long as expenditures are for foster care determined necessary by the states, federal participation is available. In fact, Title IV-E reflects Congress' very specific concerns with foster care drift and adequate safeguards and controls to assure that federally funded foster care payments meet certain requirements, such as the ones at issue here. While the State can certainly fund under its own program foster care payments which do not meet Title IV-E requirements (and some funds are available under section IV-B of the Act for foster care), the State cannot justify improperly shifting to Title IV-E payments not meeting the conditions established by Congress simply by labeling those conditions as "minor or technical." While some Title IV-E eligibility requirements are fairly characterized as more technical than others, Congress chose to make FFP available only where section 472 requirements were met. Moreover, the State is obliged to maintain documentation for its foster care cases adequate to show that federal requirements were met and that the State was not overpaid under Title IV-E. See West Virginia Dept. of Health and Human Resources, DAB No. 1257, at 10 (1991). Therefore, the State had no reasonable expectation that FFP would be available in payments for foster care services which it did not document as meeting all statutory conditions for federal participation. As we explained above, ACF is not imposing a penalty on the State when it recovers FFP claimed for expenditures which are unallowable, i.e., those payments which do not meet federal requirements. As explained below, however, we find that some issues raised by the State concerning particular eligibility requirements go to whether the auditors' determinations in individual cases were correct and are, therefore, more appropriately considered during later proceedings in this case when we examine ACF's findings in individual sample cases. As explained below, without evaluating case specific information, we were able to form only a preliminary conclusion that it appears that in some instances the auditors may have misapplied certain ACF policies. Our reasons why we reject the State's general arguments on certain requirements the State considered "minor" or "technical" are as follows. 1. Social security number -- ACF determined that 22 children for whom maintenance payments were claimed had neither a social security number nor applications for a number prior to April 1, 1985. The State asserted that the requirement that a child in foster care either have or have applied for a social security number was a technical requirement not material to congressional intent underlying Title IV-E. The Board has held that sections 402(a)(25) and 472(a)(4) of the Act required a social security number as a condition of eligibility for Title IV-E until April 1, 1985 when this requirement was eliminated. See Maryland Dept. of Human Resources, DAB No. 1225, at 9-10 (1991); Pennsylvania Dept. of Public Welfare, DAB No. 1278, at 25-26 (1991). While the State asserted that there should be a lesser "penalty" for such a technical violation of eligibility requirements, the State had explicit notice by statute and an ACF policy issuance that this was a condition of eligibility. See ACF PIQ-83-2, May 19, 1983. We do not repeat here the more comprehensive discussions in DAB Nos. 1225 and 1278. We note, however, that Congress did not make retroactive the elimination of this requirement from section 402(a)(25) of the Act for Title IV-E when it enacted a more comprehensive wage tracking system for Title IV-A and certain other programs in 1985. See section 1137 of the Act. Moreover, the mere fact that Congress did not mandate a wage tracking system for Title IV-E does not mean that social security numbers served no purpose in that program. Accordingly, as we concluded in DAB Nos. 1225 and 1278, maintenance payments can properly be determined unallowable for failure to meet this requirement. 2. Judicial orders lacking requisite findings -- ACF found that judicial determinations for 65 children did not meet statutory requirements. Section 472 (a)(1) requires that -- the removal from the home . . . [be] the result of a judicial determination to the effect that continuation therein would be contrary to the welfare of such child and (effective October 1, 1983) that reasonable efforts of the type described in section 471(a)(15) have been made. Section 471(a)(15) requires states to provide in their state plans that they will make reasonable efforts prior to placing a child in foster care to prevent the need for placement or to make it possible for the child to return home. Of the cases sampled, the auditors found that for 52 children placed in foster care after October 1, 1983, the judicial determination did not find that reasonable efforts were made to prevent removal or make it possible for the child to return home. They also found that for 13 cases where the child was placed in foster care prior to October 1, 1983, the judicial determination did not contain any language to the effect that placement was in the best interests of the child or that continuation in the home was contrary to the child's welfare. The State asserted that ACF required that the court orders "explicitly state that removal from the home was in the best interests of the child and that reasonable efforts had been made" without requiring that there be any documentation in the case records to support these findings. 7/ State Br. at 60. The State argued that ACF deemed a case eligible "[a]s long as the `magic words' appear on the court order", but found a case ipso facto ineligible without these words. Consequently, reasoned the State, ACF had reduced the judicial determination requirement from "one of substance to one of form only." State Br. at 60-61. The State asserted that ACF's mechanical application was inconsistent with Congress' broad purposes in enacting Title IV-E. State Reply at 63. The State asserted that virtually all the sample cases in this category contained court orders removing the child from the home and that in virtually all of those cases the case records would support the requisite findings. State Br. at 61. The State asserted that, in light of the confusion concerning the reasonable efforts requirement during the mid-eighties and the actual language required in court orders, it would be reasonable for the Board to examine the case records to determine whether "findings of `best interests' and `reasonable efforts,' and thus the court's order of removal or placement, were supportable." State's Br. at 63. However, the State modified this proposal in its reply brief when it agreed with ACF that the requirement for a judicial determination was intended by Congress as a substantive protection for children in foster care. The State then argued that the Board should determine whether the court orders together with the material in the case records would support the requisite findings. State Reply at 64-65. The State asserted that the absence of "magic words" in the court order should not necessarily result in permanent loss of FFP. As ACF noted, we have upheld its disallowance of FFP where a state has failed to establish that the requisite judicial determinations were made. See Nebraska Dept. of Social Services, DAB No. 1250 (1991); DAB No. 1257 (1991); New Hampshire Dept. of Health and Human Services, DAB No. 1296 (1992). Moreover, as discussed in DAB No. 1257, Congress regarded the requirement for a judicial determination as "an important safeguard against inappropriate [state] agency action." DAB No. 1257, at 3. ACF explained that -- After October 1, 1983, in order to continue to meet the requirements of this section, the judicial determination must include a finding to the effect that continuation in the home would be contrary to the welfare of the child, and also to the effect that reasonable efforts were made to prevent or eliminate the need for removal and to make it possible for the child to return to his home. The court, after hearing the evidence, must be satisfied that reasonable efforts . . . have been made. Review and approval of the [state] agency's report and recommendation alone are not sufficient . . . the court must make a determination . . . . ACYF-PA-84-1 (January 13, 1984), ACF Ex. 7. This policy announcement also stated that the states could extend the documentation used to meet the judicial determination concerning the child's welfare to meet the "new" reasonable efforts requirement. While the State is correct that there was at some point confusion concerning the reasonable efforts requirement, the confusion concerned the documentation necessary to evidence the required judicial determination. The language of the statute coupled with the clarity of the policy announcement leaves no question that there must be a judicial determination. ACF denied that the auditors had applied a "magic words" approach when auditing the City's sample cases. While there is a clear and unequivocal statutory requirement that there be certain judicial determinations in Title IV-E cases, the question presented is whether the documentation in any specific case is adequate evidence that the required judicial determinations were made. Here, we note that ACF more explicitly informed the states in ACYF-PIQ-86-02 (May 8, 1986) about the use of the state agency's petition coupled with a court order to meet the requirements. State Ex. 17. While ACF is right that there was no basis to permit the states to delay implementing the reasonable efforts requirement, its own explicit policies permitted more reliance on the state petition as evidence of the requisite determination prior to October 1986. See State Exs. 18, and 19; ACF Ex. 9. Therefore, we conclude that ACF cannot during the years audited here require that the State have in every instance the documentation stated to be adequate in the 1986 policy. Nevertheless, the proposal by the State in its reply brief that the Board examine the case records in light of the judicial determinations would still substitute the Board's determination for a judicial one. The issue is not whether the requisite determinations were supportable but whether the documentation relied on by the State is adequate (in light of the documentation policies then in effect) to establish that the judicial determinations were made. The State's arguments do raise the possibility that the auditors were overly strict with regard to the documentation which they found acceptable. To the extent the State contests particular sample cases, we will consider during later proceedings whether the requisite determinations were made. We conclude here, however, that no FFP is available if the requisite judicial determinations were not made. 3. Children whose care and placement were not the City's responsibility -- The auditors found that 15 of the sample cases were ineligible because the City did not have responsibility for the child's care and placement. The State asserted that three of these cases were children whose custody had been transferred by the City to a private placement agency in order to facilitate adoption. The State asserted that the City had "retained responsibility for case management and case supervision." 8/ State Br. at 66. The State conceded that these children were ineligible under Title IV-E because of the transfer, but asserted that "the appropriate remedy should not be loss of FFP." State Br. at 66, n.32. The State asserted that these children received the substantive protections mandated by section 472(a)(2) and that while ACF could deem the transfer of custody to be a "technical" violation, this is not a basis to find that the State failed to comply substantially with the IV-E criteria. The transfer form used by the City specified that the City transferred "exclusive control and custody . . . conferring . . . all [the City's] rights of custody and control." ACF Ex. 10. The State conceded that it did not have responsibility for the care and placement of these children within the meaning of section 472(a)(2) when it conceded for our purposes here that the transfer meant that this eligibility condition was no longer met. This is a basic, not technical, requirement which ensures that placement decisions, including adoptive placements, are made by the State in accordance with applicable program requirements. Certain protections under the Act apply only to children for whom the State has legal responsibility. Moreover, it is the State's legal responsibility for the child's care which is the explicit statutory condition. While the City may have under its own procedures continued to exercise case management and case supervision, there would be no assurance of this, or of provision of the protections to the child, absent the legal responsibility. See State Affidavit of Veronica Lynch. As we stated in Maryland Dept. of Human Resources, DAB No. 1225 (1991) on page 7 -- even if the State agency was responsible for providing services to support the child's placement, this did not constitute responsibility for the child's care. . . . if Congress had intended to make title IV-E funds available for any case in which the State agency took an interest, it would not have required specifically that the State agency have responsibility for the child's placement and care. The State will have an opportunity during later proceedings in this case to address why ACF's interpretation of the effect of the City's transfer of custody is not sound. However, we conclude here that no FFP was available for expenditures for foster care services for children for whom the State did not have the requisite responsibility. 4. Children not physically removed from the home -- The auditors found 11 children ineligible because they had resided with a relative for more than six months and were not physically removed from the home of that relative and placed in foster care, but continued to live in the same home once it was designated a kinship foster care home. 9/ For its arguments here, the State did not contest the auditors' finding that these children were ineligible because they were not removed from home prior to entering foster care. The State asserted that "[a]lthough the children may not have been physically removed from a relative's home at the time they entered foster care, they were constructively removed from their parents' homes and placed in the kinship foster care home." State Br. at 69. The State asserted that the appropriate penalty should not be permanent loss of FFP. For purposes of this stage, we state only a general conclusion that FFP is not available in payments for any child who is ineligible, despite the desirability of relative placements generally. We reserve for the later proceedings in this matter any further analysis of whether particular sample cases were in fact eligible. Although the State proceeded to address ACF's arguments concerning the necessity for physical removal from the home, we will address any issues concerning physical or constructive removal as satisfying the Act's requirement that a child be removed from the home of a specified relative later as well. We decline to address these issues here because we have insufficient information regarding the facts of specific cases. Moreover, we have determined that the basis for the finding of ineligibility needs clarifying. The discussion in the audit report raised the possibility that the auditors misapplied the applicable requirements. The audit report stated that -- Because each child had been living with the relative more that six months prior to that relative's certification as a foster parent, the children could not fulfill the eligibility requirements of Section 472 that: (1) they were removed from the home of a specified relative and placed in foster care, or (2) removal was the result of a judicial determination that to remain in the home of the specified relative would be contrary to their welfare. Audit Report at 13, State Ex. 2. The six-month time period at issue in determining whether the Title IV-E eligibility requirements are met is six months prior to the time the voluntary agreement was entered into or court proceedings were initiated, not six months prior to the date on which the home in which the child is placed is certified as a foster family home. 10/ We agree with the State that it is not clear what the auditors meant by the finding that these children had resided with the relative foster parents for more that six months "prior to becoming foster children." Audit Report at 13. (The State said that in some cases the children had entered foster care before going to live in the kinship foster home.) Also, we do not find the facts set forth in the audit report adequate to show that these children were not removed from the home of a specified relative or did not receive or were not eligible to receive payments under Title IV-A within six months of the voluntary agreement or initiation of court proceedings. Thus, while we do not consider either the requirement for removal from the home of a specified relative or the six month requirement for AFDC eligibility to be merely technical, we will need more information to determine whether these requirements were properly applied by the auditors. 5. No judicial determination within 180 days of voluntary placement -- The auditors found that FFP in payments made for 46 children voluntarily placed in foster care were unallowable (for periods after 180 days passed) because the State did not obtain a judicial determination within 180 days of placement. Section 472(e) provides -- No Federal payment may be made . . . in the case of any child who was removed from . . . home pursuant to a voluntary agreement . . . and has remained in voluntary placement for a period in excess of 180 days, unless there has been a judicial determination (within the first 180 days of such placement) to the effect that such placement is in the best interests of the child. The State asserted that the statute imposes two requirements, a substantive one for a judicial determination and a technical one requiring such a determination within 180 days of placement. State Br. at 64; State Reply at 70-71. The State asserted that for voluntary placements where the State does not obtain the requisite judicial determination within 180 days but ultimately does obtain such a determination, FFP should be withheld only for the period (after day 180) prior to the judicial determination. State Br. at 65. We conclude that there is no merit to the State's assertion. The statutory provision is clear that "[n]o payment" is available unless the requisite judicial determination has been obtained within the first 180 days of placement. Moreover, 45 C.F.R. 1356.30 explicitly provides that -- (a) As a condition of receipt of . . . (FFP) in foster care maintenance payments for a dependent child removed from his home under a voluntary agreement the State must meet the requirements of: (1) Section 472 of the Act . . . . (b) [FFP] is available only for voluntary foster care maintenance expenditures made within the first 180 days after the date of the original foster care placement unless there has been a judicial determination . . . within the first 180 days of the date of that original placement to the effect that the continued voluntary placement is in the best interests of the child. To accept the State's position here would be to substantially undercut a protection specifically set by Congress for children voluntarily placed in foster care. If anything, the amount of time afforded the State to obtain a judicial determination is generous. To conclude that this is a technical requirement causing only temporary cessation of FFP would remove the force of the 180-day time limit as an incentive to states to act in a timely manner. In sum, most of the requirements at issue here, properly applied, are essential to fulfilling the purposes of Title IV-E. Even the more "technical" requirements are nonetheless eligibility criteria of which the State had notice. The State had no reasonable expectation of FFP in payments for children not meeting the criteria. IV. ACF has authority to disallow based on valid statistical sampling methodologies. ACF determined the disallowed amounts here by projection (extrapolation) from a statistical sample. In this section, we address issues raised by the State concerning ACF's general use of statistical sampling as a basis for Title IV-E disallowances. In section V. below, we address questions raised by the State concerning the particular statistical methods used. The State acknowledged that "the Board and the courts have traditionally upheld the use of statistical sampling as an audit tool to establish rebuttable facts where there is a clear expression of statutory authority, such as in the AFDC quality control program, or as a reasonable interpretation of the Secretary's authority to determine `overpayments.'" State Br. at 38 (emphasis in original), citing Georgia, 446 F. Supp. 404; Pennsylvania, DAB No. 1278 (1991). The State argued, however, that prior Board decisions need not be controlling and that, in this instance, use of sampling was unlawful because it represented a change in ACF policy which was not promulgated using notice and comment procedures or, alternatively, because the State did not have adequate and timely notice. The State also argued that, if sampling was permissible at all, the State should be given the benefit of a "tolerance" for errors, such as that which applies in the AFDC quality control program. For reasons explained below, we conclude that none of these arguments has merit. A. Valid statistical sampling provides reliable evidence concerning the amount of unallowable payments for which a grantee claimed FFP. This Board has consistently held that valid statistical sampling provides reliable evidence of the amount of unallowable costs a grantee has claimed. See, e.g., New York State Dept. of Social Services, DAB No. 1079, at 5 (1989); California Dept. of Social Services, DAB No. 816, at 4-5 (1988), and cases cited therein. These decisions were based in part on the rationale in Georgia, which upheld use of sampling methods to determine the amount of unallowable state Medicaid payments for physician claims. The Georgia court reasoned that sampling had been recognized as a valid audit technique and that statistical results had been recognized by courts as reliable and acceptable evidence in establishing adjudicative facts. 446 F. Supp. at 409; see also Rosado v. Wyman, 322 F. Supp. 1173 (E.D. N.Y. 1970), aff'd, 402 U.S. 991 (1971). The court noted: "Audit on an individual claim-by-claim basis of the many thousands of claims submitted each month . . . would be a practical impossibility as well as unnecessary." Georgia, 446 F. Supp. at 410. As this Board has noted, sampling (and extrapolation from a sample) done in accordance with scientifically accepted rules and conventions has a high degree of probability of being close to the finding which would have resulted from individual consideration of numerous cost items and, indeed, may be even more accurate, since clerical and other errors can reduce the accuracy of a 100% review. See DAB No. 1079, at 5. The State cited several court cases for the proposition that prior Board decisions need not be controlling in determining issues currently before the Board. While we agree with the State that prior Board decisions need not be controlling, the State provided no reasonable basis for us either to change our prior rationale or to find it inapplicable here. The auditors here needed to audit a foster care caseload containing numerous cases. It would not be practical to perform a 100% review, nor would such a review have been likely to result in a more accurate determination of unallowable costs than a sample review. Assuming that the statistical sampling method chosen was valid and was properly applied (which we find below it was), the results are reliable evidence of the amount of unallowable foster care maintenance payments made by the State and charged to federal funds. 11/ The fact that Georgia addressed use of sampling as a basis for a disallowance under Title XIX, which gives the Secretary authority to determine "overpayments" of FFP in Medicaid, does not distinguish that case. As discussed above, section 474(d) of the Act specifically authorizes the Secretary to adjust a state's claims for IV-E funds to the extent any overpayment was made in a prior quarter. The wording of this section parallels the comparable Medicaid provision in section 1903(d) of the Act. As discussed above, if ACF has paid a claim for FFP under Title IV-E for costs which do not meet applicable requirements, that constitutes an overpayment to the State to be adjusted. Numerous courts have held that specific authority to use statistical sampling as an audit tool to produce evidence on the factual issue of the amount of unallowable claims for federal funds is not necessary. See, e.g., Chaves County Home Health Service, Inc. v. Sullivan, 931 F. 2d 914 (D.C. Cir. 1991), cert. denied, U.S. , 112 S. Ct. 1160 (1992); Michigan Dept. of Education v. U.S. Dept. of Education, 875 F. 2d 1196 (6th Cir. 1989); Mile High Therapy Centers, Inc. v. Bowen, 735 F. Supp. 984 (D. Colo. 1988). The State pointed to nothing in the language or history of Title IV-E indicating any intent to preclude use of sampling for determining overpayments of FFP. Congress can be presumed to have intended that reasonable audit tools be used, so long as they provide reliable evidence. Moreover, the State has been given ample opportunity, in response to the audit and in Board proceedings, to rebut the evidence developed through the sample (and will be given a further opportunity to contest ACF's findings in individual sample cases). B. Use of sampling here is not unlawful on the ground that it represents a change in agency policy subject to notice and comment rulemaking. The State argued that ACF "policy prior to the inception of Title IV-E was not to use statistical sampling as a basis for disallowing foster care funds." State Br. at 37, citing Louisiana Dept. of Health and Human Services, DAB No. 580 (1984). The State argued that ACF's "change in policy to use statistical sampling as a basis for disallowing IV-E funds and to no longer limit recoupment to individual errors seriously affects the State's interest in the amount of FFP it receives" and, therefore, constitutes "a substantive interpretation that must be promulgated according to the [notice and comment] procedures required by the Administrative Procedure Act, 5 U.S.C. 553." State Br. at 37. The State relied on the case of Batterton v. Marshall, 648 F. 2d 694 (D.C. Cir. 1980) in support of its position that notice and comment procedures were required. Use of sampling here does not represent a policy change, nor does it affect any substantive right of the State. Thus, notice and comment rulemaking was not required. In Louisiana, this Board discussed the history of agency policy on use of statistical sampling in the AFDC program to identify errors within the scope of the AFDC quality control program (involving required ongoing sampling of each state's AFDC eligibility determinations). In light of that history, the Board found that the agency could not fairly apply retroactively to AFDC foster care payment errors a change in policy from disallowing only individually identified errors to disallowing errors determined through statistical samples. The change in policy had been promulgated in an action transmittal issued to states on December 13, 1982, Action Transmittal (AT) 82-33. The Board has previously held that the rationale in Louisiana was limited to eligibility determination errors within the scope of the AFDC quality control program and did not apply to "payments made under the completely separate IV-E program." Pennsylvania Dept. of Public Welfare, DAB No. 1278, at 7 (1991). The State presented no reason here why we should reconsider this holding and find that a policy developed for one program, applicable to an element of that program with a particular history of how and when sampling would be used, should apply to another program. While the IV-E program replaced the AFDC foster care program, neither Congress nor ACF expressed any intent to adopt the AFDC quality control system or related disallowance policies for the new IV-E program. Thus, the State had no reasonable expectation for Title IV-E that ACF would disallow only individually identified errors. Since we do not agree that use of sampling for Title IV-E represented a policy change, we conclude that the State's reliance on cases discussing Administrative Procedure Act (APA) requirements for rules or policies which constitute a change in existing law, policy, or practice is misplaced. See State Br. at 37. In any event, even if use of sampling represented a change, it is not the type of change requiring notice and comment rulemaking. Notice and comment rulemaking is required under the APA only for legislative-type rules; it is not required for interpretative rules, general statements of policy, or rules of agency procedure or practice. 5 U.S.C.  553(b)(3)(A). In Ohio Dept. of Human Services, DAB No. 1202 (1990), we rejected an argument that an agency's policy on statistical sampling methods was a legislative-type rule. We found that it was a general statement of policy or rule of agency procedure, since it represented a means for gathering evidence on a state's compliance with statutory requirements, rather than an inflexible standard that must be applied. This decision was upheld in district court. State of Ohio, Dept. of Human Services v. Sullivan, 789 F. Supp. 1395 (S.D. Ohio 1992), appeal docketed, No. 92-3560, 6th Cir., June 11, 1992. The district court agreed that the sampling methods were "simply a means of gathering and analyzing evidence." 789 F. Supp. at 1405. Like the methodology considered in Ohio, the methodology used by the auditors here is simply a means of gathering evidence; it does not substantively change what the State is required to do in order to be entitled to FFP. Moreover, the particular audit methodology used here does not conclusively establish the amount disallowed. The State has had ample opportunity to show that the method did not generate reliable evidence. Thus, the State's reliance on Batterton is misplaced. That case involved a sampling methodology developed by a federal agency to calculate unemployment statistics, which were then used as a basis for determining allocation of grant funds among states, with no provision for adjustment of the allocation through adjudication or any other method. The methodology in Batterton thus affected substantive rights since it established the amount of funds available to grantees for authorized purposes. See also Maryland Dept. of Human Resources v. Sullivan, 738 F. Supp. 555, 561 (D.D.C. 1990) (program instruction describing both the sampling methodology and pass rates to be used in determining eligibility for section 427 funds is a interpretative rule, exempt from notice and comment procedures). C. The State had adequate and timely notice of the ACF policy to use sampling. Under the APA, policy statements and agency procedures generally have to be published in the Federal Register, but timely and actual notice can substitute for publication. 5 U.S.C.  552(a)(1). The State had such notice here. Action Transmittal 82-33, issued December 13, 1982, explained that Department policy had always been that payments excluded from the AFDC quality control universe were subject to any recognized audit technique to determine the amount disallowed, including statistical sampling methods. Louisiana, DAB No. 580, at 7. Policy Announcement (PA) 83-02, issued May 4, 1983, reiterated that "for title IV-A - Foster Care . . . it is the policy of the Department to use sampling as the basis for determining expenditures ineligible for Federal matching." State Ex. 4, at 2 (emphasis in original). Finally, in order to ensure that there was no misunderstanding about the use of statistical sampling in Title IV-E, ACF issued Policy Announcement 84-2 on March 7, 1984, stating: It is now and has always been the policy of the Department to use valid statistical sampling methods, including extrapolation from a sample to a universe, to determine the amount of expenditures eligible for Federal financial participation . . . . State Ex. 5. The State did not deny receiving these issuances, but argued that the Board had "implicitly rejected" the policy set forth in PA-83-02 and PA-84-02 in the Louisiana decision. The State argued that the "stated policy and contradictory Board ruling further served to deprive the State of clear notice of the change in policy in advance of incurring expenditures under Title IV-E," and that the "policy issuances were also untimely since they were not issued prior to the State's filing of its state plan under Title IV-E, which was effective April 1, 1982." State Br. at 40. Contrary to what the State argued, the Board's decision in Louisiana did not, explicitly or implicitly, reject the policy stated in these issuances. What the Board rejected in Louisiana was the attempt in AT-82-33 to describe prior issuances as consistent with the policy stated in AT-82-33. The Board said that AT-82-33 was a change in disallowance policy for Title IV-A which could not fairly be applied retroactively to erroneous payments within the scope of the AFDC quality control program since states may have made decisions concerning where to devote resources for correcting errors, based on the prior policy of disallowing AFDC-foster care erroneous payments only if individually identified. Here, payments under Title IV-E have never been within the scope of the AFDC quality control program, nor could a state reasonably have thought at any time that only individually identified Title IV-E errors would be disallowed. Thus, the potential adverse effect from retroactive application of AT-82-33 present in Louisiana simply is not present here. In any event, the general policy stated in AT-82-33 -- to use extrapolation from statistical samples as a basis for disallowance -- is being applied prospectively in this case. AT-82-33 was issued December 31, 1982, and the expenditures at issue here were incurred during the period October 1, 1983 to September 30, 1985. The State's attempt to characterize the Board's decision in Louisiana as somehow introducing confusion is unavailing. The Board specifically limited its decision in Louisiana to the period and type of error in question there, emphasizing that it found "nothing wrong with the Department's general policy of using extrapolations from statistical samples to produce disallowances." DAB No. 580, at 2. 12/ The State cited no authority for its position that the relevant event was the filing of its Title IV-E plan, and we know of none. Under grant programs such as Title IV-E, a state's entitlement to FFP arises when it incurs allowable expenditures. Notice given prior to the time those expenditures are incurred is generally adequate and, in our view, the State definitely had adequate notice here. Indeed, since the State could never have determined reasonably that there would be Title IV-E disallowances only for individually identified errors, use of statistical sampling in Title IV-E could reasonably be viewed solely as an audit technique. The State itself does not have to take any steps to implement the technique, and its only interest is in ensuring that the sample findings are accurate and that the sampling method used is reliable. Thus, notice prior to the use of the technique (which the State clearly had) is adequate here. We further conclude that Bowen v. Georgetown University Hospital, 488 U.S. 204 (1988) is inapposite here. That case involved unauthorized retroactive application of a legislative-type regulation establishing reimbursement rates for Medicare hospitals. As explained above, the policy here is being applied prospectively and is not a legislative-type rule. D. ACF was not required to apply a tolerance level to the sample results. The State argued that a tolerance level, such as that applied in the AFDC program to AFDC eligibility determination errors identified through the quality control system, should also be applied here. The Board has previously addressed the question of whether tolerance levels should be applied to errors not within the scope of a quality control program. The Board held that, while establishing such a tolerance for errors was within the Secretary's authority under section 1102 of the Act, such a tolerance was neither required nor appropriate for Board adjudication on a case-by-case basis. California Dept. of Health Services, DAB No. 170 (1981), aff'd, California v. Settle, 708 F.2d 1380 (9th Cir. 1983). The mere fact that the Secretary (and later Congress) determined that a tolerance was appropriate for AFDC eligibility determination errors identified through the quality control system does not necessarily mean that one is appropriate for the Title IV-E errors at issue here. The quality control system involves ongoing sampling by the states (with further subsample review by ACF), rather than sampling in the context of a federal audit. The rationale for allowing a tolerance in AFDC was the complexity of the eligibility determination process (with its presumption in favor of eligibility), which makes it impossible for a state to totally avoid errors. Under Title IV-E, a state must make IV-A eligibility determinations for some children, but for, others, can rely in part on eligibility determinations already made. Many of the errors here arose from applying requirements which are unique to IV-E, such as the requirement for a judicial determination, and which are much easier to apply than the income and resource provisions of Title IV-A. The State submitted parts of audit reports of other states' Title IV-E programs and alleged that the "fact that other states have also incurred large disallowances under Title IV-E demonstrates that a zero tolerance level, i.e., requiring 100 percent compliance, is not reasonably achievable." State Br. at 41. According to the State, this "also demonstrates that the fault lies not with the states, but rather with an overly complex law that is entirely too difficult to administer, given the exactitude expected by the OIG and the Agency and the limited resources granted the states." State Br. at 41. The amounts proposed for disallowance in audit reports are not necessarily equivalent to amounts actually disallowed, however. Audits simply recommend disallowances, and the ultimate disallowance will be a lesser amount if ACF or the Board disagrees with the auditors or if a state submits new documentation. In any event, the submitted audit reports do not support the State's assertion that the audit findings demonstrate that the errors were attributable to an overly complex law. Indeed, most of the errors identified in the audit reports appear to be attributable to states' failures to obtain or document the judicial determinations required by section 472 of the Act. This is a basic, straightforward program requirement, which ACF has implemented by accepting statements in court orders "to the effect that" the required determinations were made, a standard which can hardly be considered "exactitude." Moreover, some of the other audit findings for other states also contradict the State's assertion, for example, findings that a state had a regulation in conflict with the federal requirements, or had failed to implement required procedures, or had identified the children as ineligible but had failed to exclude related payments from FFP claims because of a computer problem. State Exs. 9-12. We also note that our experience with Title IV-E disallowances suggests that states have in some instances knowingly shifted other foster care cases to Title IV-E since otherwise the states would have to cover the payments entirely with their own funds, or have failed to take simple administrative steps such as providing courts with model orders meeting Title IV-E requirements. The State referred to the use of a zero tolerance as a change in agency policy, arguing that since "the State did not voluntarily and knowingly agree to the federal rules, it is not bound by the rules." State Br. at 43, citing Perales v. United States, 598 F. Supp. 19 (D.C. N.Y. 1984), aff'd, 751 F. 2d 95 (2nd Cir. 1984); Pennhurst, 451 U.S. 1 (1981). As explained above, however, the statute and regulations are clear in conditioning FFP under Title IV-E on a state's meeting certain requirements. The State had no reasonable basis for believing it would be granted any tolerance for errors in meeting these requirements. Moreover, ACF's decision not to apply a tolerance here is not unreasonable. The requirements at issue here implemented statutory changes which were intended to narrowly circumscribe federal funding for foster care in order to reduce foster care drift and promote permanency planning for children. Holding states to a higher standard for preventing errors under Title IV-E than under Title IV-A could reasonably be viewed as consistent with this intent. In sum, we conclude that ACF's use of sampling here was both authorized and consistent with ACF policy. We also conclude that ACF was neither required to apply a tolerance level here, nor unreasonable in refusing to do so. V. The particular statistical methods used here produced reliable evidence and were not inconsistent with ACF policy. In this section, we first describe the statistical sampling methods used as a basis for the disallowance of maintenance payments. We then address the State's challenges to these methods. The State did not challenge the methods here by producing any expert testimony or other evidence disputing the statistical validity of the particular methods used. Rather, the State argued generally that the statistical sample used here was defective because it did not comply with ACF policy and because it projected errors occurring under the AFDC program to expenditures incurred under Title IV-E. In its reply brief, the State also argued that the OIG sampling plan contravened OIG policy. For the reasons stated below, we conclude that these arguments have no merit. A. The sampling methods used here were valid. ACF provided evidence concerning the methods used. ACF Affidavit of Margaret Wallace. The State did not dispute that evidence, so we have used it as a basis for our description here. The sample universe consisted of children in New York City for whom foster care payments had been made by New York City during the audit period, which payments New York State then claimed for FFP. Sample universe listings were provided to the auditors by the State and were generated by the State's computerized Welfare Management System (WMS). These listings included 20,446 children for FY 1984 and 23,638 children for FY 1985. Before reviewing the cases, the auditors conducted a validation study to determine whether the listings generated by the State accurately represented the sample universe. The validation study disclosed that the lists were inaccurate, so the State produced a second set of listings. On the basis of a second validation study, the auditors determined that the new universe listings were reasonably accurate. The auditors then drew a scientific sample from the listings provided by the State, that is, a sample in which each item in the universe had an equal chance of being selected. The sample was selected by a computer using random number generator software. The selected sample consisted of 300 cases (150 cases from each fiscal year). The auditors used a stratified sampling technique, in which each stratum represented children on whose behalf payments had been made and FFP claimed for at least one day in the relevant fiscal year. 13/ The auditors then held an entrance conference with City and State representatives to discuss the objectives of the audit, the intended audit approach, and any procedural matters that might arise in conducting the audit. The auditors also requested that the City produce the case files for the children selected in the sample and provide access to payment records which would enable the auditors to determine the amount of maintenance payments made on behalf of these children. The auditors reviewed the payment records to determine the amount of payments made on behalf of the sample children for the applicable fiscal year from which they were selected. The auditors also determined whether such payments had been claimed for FFP and whether any adjustments or retroactive claims had been made by the City. The auditors also reviewed the case files for the sample children, using an audit questionnaire which had been developed after consultation with ACF and the State. Each completed audit questionnaire was reviewed by the auditor in charge of this phase of the audit, and a copy of each questionnaire was given to the City, which had the opportunity to submit more documentation regarding matters questioned by the auditors. An exit conference was held with City and State officials, who then furnished additional information. To project the results of the sample, the auditors first compiled the amount audited for each sample child and determined the amount ineligible for FFP. The auditors used this compilation to create a computer data file and then used a variables sample appraisal program for stratified samples on the computer. This appraisal program (which has been validated using National Bureau of Standards methodology) projected the sample results to the universe from which the sample was drawn. The sample projection disclosed that, at the 90% confidence level, between $141,385,044 and $182,665,197 in unallowable maintenance payments had been made and claimed for FFP. The single most likely estimate of the amount of unallowable payments was $162,025,120. Consistent with OIG audit policy, however, the auditors recommended disallowance at the lower limit of the confidence interval, that is, at $141,385,044. Using the lower limit allowed the auditors to estimate with a 95% probability that the true amount of unallowable payments equalled or exceeded the recommended disallowance. As a result of a supplemental review of the audit findings for sample cases by ACF, the auditors adjusted the findings and determined the lower limit of the confidence interval for the adjusted findings, $64,123,732 in FFP. B. The sampling methodology did not violate ACF or OIG policy. As noted above, the State did not challenge the statistical validity of the sampling methods used. Instead, the State contended that the statistical sample was "defective" because it did not comply with ACF policy, as set forth in ACYF-IM-85-4. The State said that that policy required a minimum random selection of 200 sampling units and "clearly envisioned that the selection would be drawn from a universe of cases within a two to four quarter time frame." State Br. at 45. ACYF-IM-85-4 discusses regular reviews of foster care claims which ACF intended to conduct as part of its implementation and oversight of Title IV-E. The issuance specifically notes: These reviews are in addition to reviews that will be conducted from time to time based on such criteria as claims whose allowability is questioned and audits. State Ex. 6, at 1. The issuance also notes that other reviews "may be conducted using valid and reliable sampling." State Ex. 6, at 3. Thus, contrary to the State's assertion, the sampling here does not violate ACYF-IM-85-4, since that policy is limited to a particular type of review to be conducted by ACF. The policy by its own terms does not apply to the OIG audit at issue here. 14/ Even if ACYF-IM-85-4 did establish a policy for OIG audits (which it does not), we would not find the sampling methodology used here to be "defective" on that basis. The State provided no evidence to show that the sample size used here (300 children) was insufficient in light of the size of the sample universe. The State's objection was based on the sample size for each fiscal year (150 children) and on the reference in ACYF-IM-85-4 to a "sample of at least 200 payment units." The reference to payment units, however, indicates that the sample would be a sample of payment units, rather than a sample of children; the universe of payment units is larger than the universe of children since usually more than one payment is made on behalf of any child. Moreover, we find no support in ACYF-IM-85-4 for the State's assertion that ACF contemplated that a sample of 200 payment units would be drawn within two or four quarters. While the policy refers to review of regular, current quarter claims (as opposed to retroactive claims for expenditures incurred in prior periods), the policy also indicates that the contemplated reviews would be made "at least every three years . . . ." Thus, we do not read the policy as binding ACF to use a larger sample size for a fiscal year than that used here. We also conclude that the OIG sampling plan did not violate OIG policy. The State focused on wording in the auditors' sampling plan which said that the sample size was "arbitrarily selected." ACF Ex. 1, at 4. The State said that this violated OIG policy which indicates that determining sample size is a "key element" of any sampling plan and which indicates that a sample performed for "error analysis" should contain 20 items for each characteristic being reviewed. State Reply at 45-47, citing State Ex. 22. 15/ First, we note that the OIG document cited by the State contains both policies and procedures. The statements relied on by the State are not in the section labeled "POLICY" but appear in a section labeled "SAMPLING PLAN." This section describes the sampling plan as summarizing "the audit team's ideas concerning a sample" which are then reviewed by a statistical sampling expert who may modify the plan. Read carefully, the section cannot be read as binding OIG to use of any particular sample size. It merely describes how the auditors "should" determine sample size and seek approval for it. Moreover, the reference in the OIG document to "error analysis" requiring 20 units for each characteristic audited appears to apply only to "attributes" sampling "used to estimate the rate or proportion of a characteristic or group of characteristics in an universe." State Ex. 22, at 4. The auditors here chose "variables" sampling, which is "used to estimate quantitative characteristics, usually dollar amounts, in an universe." Id. The OIG statistical expert approved this approach, as well as the sample size chosen, and the State presented no evidence from a statistical expert challenging either the approach or the sample size. Finally, we note that the State did not allege that it was somehow prejudiced by use of the particular sample size chosen. This is not surprising since sample size affects the precision of sample results in estimating the most likely true value, and a smaller sample size generates a wider confidence interval. Since ACF disallowed only the amount established by the lower limit of the confidence interval, the State potentially benefited from the use of a sample size totaling 300 units, rather than a larger sample size. C. The sampling method does not improperly project Title IV-A errors to Title IV-E. The State argued that another "defect in the sample was the projection of errors that occurred under the Title IV-A program to a universe of expenditures incurred under the Title IV-E programs." State Br. at 45. The State gave as an example of this type of error a sample unit where ACF found that the child's record lacked sufficient information establishing that the child was from a family that met AFDC eligibility requirements when the child was placed in foster care in 1972. The State argued that the sample failed to meet the Board's requirement that the sampling methodology be sound because the sample "uses a 1972 error to disallow 1984-85 expenditures." State Br. at 46. The State asserted: "The time period in which the error occurred must be the same period, or at least proximate to the period, of expenditures to which the error is projected." State Br. at 46. The State argued that OIG used an error that occurred at a time when ACF did not have a policy that permitted extrapolation and when "there was no quality control program for AFDC using a valid statistical sampling methodology to establish on an extrapolated basis the allocation of 1984-85 expenditures." State Br. at 46. The State also argued that the OIG, in finding the State responsible to "maintain sufficient documentation 16 years later . . . holds the State to a record retention requirement far in excess of the three-year record requirement set by regulations." State Br. at 46-47. According to the State, the reason the records were missing was that ACF had advised the State that it did not have to keep financial information concerning the AFDC eligibility of the family placing the child with future records related to the child. The State said further that it was unreasonable of the OIG to expect the State to produce records from 1972. The State asserted that "the OIG must judge the allocability of expenditures for Title IV-E by reviewing only the records of children entering the program under the title IV-E criterion, effective April 1982." State Br. at 48. These arguments confuse the question of the validity of the sampling methodology -- or its soundness -- with issues regarding the correctness of an audit finding that a payment was unallowable. Even if an auditor was incorrect in finding that a child was ineligible for Title IV-E payments or in concluding that the State had failed to document properly the child's eligibility, that would not render invalid or unsound the statistical sampling method used. The State also confused questions related to allocability of costs with questions concerning the substantive allowability of the costs. The issue here is not whether maintenance payments -- or errors -- were properly allocated to Title IV-E rather than to Title IV-A. The payments disallowed here were all claimed under Title IV-E. Title IV-A becomes relevant only because section 472 of the Act makes eligibility for Title IV-E maintenance payments contingent on the child's Title IV-A eligibility. Section 472 requires this eligibility generally within six months of the month the child is voluntarily placed or court proceedings are initiated. Thus, under the statute, AFDC eligibility in 1972 for a child placed in 1972 may have also been a condition for the child's eligibility under IV-E, both when the child was first claimed under Title IV-E in 1982 and when payments were claimed in 1984 and 1985. The State is required to document the Title IV-E eligibility of any child for the period for which the expenditure is claimed. The record retention requirement for expenditures claimed for each federal fiscal year is three years from the time the expenditure report is submitted for the last quarter of the federal fiscal year, or longer if an audit is initiated within that time period (as it was here). 45 C.F.R. Part 74, Subpart D. Thus, the State must retain any documentation necessary to establish the allowability of Title IV-E payments made in 1984 and 1985 until this audit is fully resolved. The question of what documentation the State was required to retain in order to establish the allowability of payments made in 1984 or 1985 for children who entered the foster care system in an earlier year is best addressed in the context of our review of individual sample cases. If in the course of that review, we conclude that the auditors applied a standard inconsistent with ACF policy, we would overturn any findings based on the incorrect standard, and require a corresponding reduction in the projected disallowance. Finally, the question of whether an AFDC disallowance could have been based on statistical sampling for an error in determining AFDC eligibility in 1972 is irrelevant here. The disallowance here relates to Title IV-E payments made in 1984 or 1985, and, as discussed above, the applicable policy was to permit disallowances for this period based on statistical sampling techniques. 16/ In sum, we conclude that the sampling methodology used here was a valid one, which (assuming the findings in individual sample cases are correct) produced reliable evidence of the amount of unallowable foster care maintenance payments the State claimed under Title IV-E. We further conclude that the sampling methodology was not inconsistent with ACF or OIG policy. VI. The pro rata method ACF used to determine unallowable administrative costs is inconsistent with the State's approved cost allocation plan and is otherwise unreasonable. In this section, we discuss the State's arguments concerning ACF's method of calculating unallowable administrative costs which ACF determined were associated with the disallowed maintenance payments. ACF disallowed $27,991,567 in Title IV-E administrative costs based on a "pro rata" method of calculation. The State's major arguments were that the pro rata method conflicted with the State's approved cost allocation plan (CAP) and that ACF had failed to establish that the method was valid and reasonable. For reasons explained in this section, we conclude: o ACF was reasonable in inferring from the audit findings (which showed that the State had made numerous errors in claiming maintenance payments under Title IV-E) that the State had likely also overstated its Title IV-E administrative claims. o ACF was also generally reasonable in using an estimation technique to determine an amount by which administrative cost claims were overstated as a result of the errors identified in the audit. o ACF did not show that the particular estimation technique used here -- the pro rata method -- was reasonable. To reliably estimate the amount of administrative costs improperly claimed as Title IV-E costs, the estimation technique must be based on a rational relationship between the types of administrative costs actually claimed by the State and the types of errors found in the audit sample. The pro rata method used here is flawed because: (1) it incorrectly assumes that the activities covered by the State's claims are Title IV-E activities only if performed for children who are accurately determined and documented as eligible for Title IV-E maintenance payments; (2) it incorrectly assumes that Title IV-E administrative costs will increase proportionately to the amount of maintenance payments made; and (3) it in effect amends the approved CAP without meeting the standards for retroactively changing an allocation method. Below, we first discuss generally why we consider it reasonable for ACF to determine that some disallowance is warranted and that the disallowance could be based on an estimation technique. We next explain what a CAP is and generally how the State's CAP worked. We then describe the pro rata method used to calculate the administrative cost disallowance and explain why we find that the pro rata method conflicts with the approved CAP and is otherwise unreasonable. We also discuss why ACF's reliance on prior Board decisions is misplaced. Finally, we explain why our decision does not preclude ACF from further examining the State's allocation of administrative costs to determine whether that allocation can reasonably be adjusted to account for the sample case findings, consistent with the CAP. A. ACF was reasonable in determining that the State had overstated administrative costs and in estimating the amount of the overstatement. At the outset, we note that our decision here is limited to overturning the particular method ACF used in calculating a disallowance of administrative costs. If the State was operating its program in such a way that it was significantly overstating which maintenance payments were for Title IV-E eligible children, the State most likely was also overstating which costs were necessary for the proper and efficient administration of its Title IV-E program. We also conclude that ACF was reasonable in using a method which involved estimating the amount by which the State's claims for administrative costs had been overstated. The State's claims for administrative costs are necessarily based (at least in part) on allocation methods which are essentially estimation techniques. It is impracticable for a state to identify with particularity each of the costs incurred for a specific objective; moreover, some costs benefit more than one program. Thus, estimation techniques are acceptable in allocating costs, so long as they have a rational basis which results in an equitable distribution of costs. See Oklahoma Dept. of Human Services, DAB No. 963, at 4-6 (1988). The State did not argue that ACF could not use an estimation technique, but did argue that, under past Board decisions, ACF had the burden of showing that its method of calculating the disallowance was reasonable. The Board decisions relied on by the State hold that an agency using a statistical sampling method generally has the burden of coming forward to show that the sampling methods were valid. The State relied on University of California -- General Purpose Equipment, DAB No. 118, at 5 (1980), and Ohio Dept. of Public Welfare, DAB No. 226, at 3 (1981). However, what an agency must show to meet its burden may be affected by the overarching burden on the grantee to document in the first instance that its claims meet applicable federal requirements. See DAB No. 226, at 2-3. Here, ACF did not actually sample the State's administrative costs. Instead, ACF used the results of a sample of maintenance payments to estimate the amount by which the State's administrative costs were overstated. Use of such an estimation technique is, however, comparable to sampling in that ACF is relying on it as evidence to establish a particular disallowance amount. Thus, as the proponent of the method, ACF has the burden of showing that it is reasonable under the particular circumstances here and produces reliable evidence to support the amount of the disallowance. Moreover, as discussed below, since superimposing the pro rata method on top of the approved CAP method results in a retroactive change to the CAP, ACF has the further burden of showing that circumstances exist which are sufficiently compelling to warrant changing the CAP retroactively. B. The State was required to claim administrative costs in accordance with an approved CAP. Section 474(a)(3)(C) authorizes FFP at a 50% rate for costs found necessary by the Secretary of HHS for the proper and efficient administration of the state plan under Title IV-E. 17/ ACF regulations describe costs of certain administrative activities which must, may, or may not be claimed under Title IV-E. These regulations specifically require that a state's "cost allocation plan shall identify which costs are allocated and claimed under this program." 45 C.F.R.  1356.60(c). Procedures for submittal and approval of CAPs for public assistance programs are set out in Department regulations at 45 C.F.R. Part 95, Subpart E, made applicable to Title IV-E by 45 C.F.R.  1355.30. See also 45 C.F.R.  205.150. The State was required to submit to the Department's Division of Cost Allocation (DCA) a CAP describing the procedures used to identify, measure, and allocate costs to each of the programs administered by the State and containing certain specified information. 45 C.F.R.  95.507. The State must amend its CAP if certain specified events occur. 45 C.F.R.  95.509. The Director, DCA, is responsible for approving or disapproving CAPs and amendments, after consulting with the affected operating divisions of HHS such as ACF. 45 C.F.R.  95.511. An approved CAP is not a rigid unalterable "contract" binding the parties, and approval of a CAP cannot make a cost allowable or allocable contrary to statute or regulation. Nor does approval of a CAP constitute approval of specific costs claimed. See Oklahoma Dept. of Human Services, DAB No. 963 (1988); California Dept. of Social Services, DAB No. 855 (1987); State Affidavit of Roger Nelligan, Attachment (Att.) (CAP approval letter). Based on its analysis of CAP approval requirements, the Board has concluded that it would not uphold an allocation method which was clearly inequitable, which had been approved based on incorrect, inconsistent, or incomplete data submitted to DCA, or which was prohibited by statute or regulation. The Board has also concluded, however, that approval of a CAP by DCA gives rise to a presumption that the approved allocation methods are valid. The Board thus defers to DCA's expertise, absent a compelling basis for concluding that the approved plan was improper. Oklahoma, DAB No. 963, at 6-7. 18/ One method of allocating costs among Title IV-E and other benefitting programs is to use a caseload count of eligible individuals to develop percentages which are then applied to determine each program's share of administrative costs. Other methods are based on the percentages of employees' time spent on various activities. The State asserted that the allocation method the State used during the audit period was a random moment study (RMS) method. 19/ According to the State, this methodology had been revised in 1983 and approved by DCA with an effective date of October 1, 1983. The State described the RMS system as follows: The RMS estimates the percentage of time the staff of the New York City Department of Social Services ("NYCDSS") spend on various social services programs. To arrive at the estimates, the State generates a random sample of NYCDSS employees and corresponding "moments." At the date and time of the random moment, the selected employee is interrupted and questioned by a NYCDSS sample taker about the activity on which the employee is working. The employee's response is recorded on an RMS form. The responses for all employees sampled during a given fiscal quarter are collected and allocated by the State to the appropriate State or federal program. The State then generates a list of the percentage of responses allocated to each program. These percentages are applied to a cost pool comprised primarily of the salaries, benefits, and overhead costs associated with the NYCDSS employees, and the resulting products are charged to various State and federal programs as administrative costs. State Br. at 13-14 (footnote omitted); see also State Affidavit of Roger Nelligan.. The State asserted and ACF did not deny that its administrative claims were made in accordance with the approved CAP. C. The method ACF used to calculate the administrative cost disallowance was not correlated with the RMS. After reviewing information on sample cases submitted by the State in response to the audit report, ACF determined that it could say with a 95% confidence that at least 52.96% of the maintenance payments claimed by the State in FYs 1984 and 1985 were unallowable. ACF determined, "[c]onsistent with the approach taken in the original audit report," that this meant that "52.96% of administrative costs are also unallowable." State Ex. 1, at 4. The parties here referred to the auditors' approach as a "pro rata" method since it assumes that the percentage of total State administrative costs which are unallowable is the same as the percentage of total maintenance payments found unallowable. The State argued that application of this pro rata method, on top of the allocation of administrative costs already made by the State using RMS results, was inconsistent with the State's approved CAP. The State said that ACF failed to show any correlation between the percentage of unallowable maintenance payments and the RMS results. ACF did not deny that it did not correlate the method of calculating the administrative cost disallowance with the RMS method. The auditors explained this as follows: The audit staff did not question the random moment study method; however, based on the high error rate in the sample, the auditors did question whether the information in the RMS was being collected properly. However, it would not have been feasible or accurate to attempt to correlate the cases in the sample with the cases represented in the RMS studies. If the administrative costs were calculated based solely on the number of cases in the sample, 76.6% of the administrative costs would be unallowable. If we also included the 22 cases for which no case folder could be located, 78.6% of the administrative costs would have been unallowable. Since no two cases can be expected to consume the identical amount of administrative costs, we decided to use the RMS results as a base to which we applied the percentage of errors disclosed by the audit to determine the amount of unallowable administrative costs. The amount of maintenance payments in each case was an estimate of the amount of administrative costs associated with that case over the period covered by the audit. ACF Affidavit of Margaret Wallace at 7 (paragraph numbers omitted). We share the auditors' concern that a high error rate in determining eligibility may have affected the accuracy of the State's administrative claims. As noted above, the record does not clearly establish that all of the administrative costs were allocated using RMS percentages, rather than caseload count. See n.19. Moreover, the State admitted that the RMS involved questioning City employees on the activity they were performing and the program to which the activity related. State Affidavit of Roger Nelligan at 2. If a City social worker identified the program as Title IV-E based on an erroneous eligibility determination, this could have resulted in what would essentially be an erroneous response to the RMS. 20/ As discussed more in detail below, the problem with the auditors' approach is that the erroneous payments found in the audit sample do not necessarily indicate a proportionate rate of errors in allocating administrative costs to Title IV-E. Even if the auditors could not feasibly correlate individual sample cases with the cases represented in the RMS studies, this does not excuse a failure to use an estimation method which reasonably relates the errors found in the audit with the administrative costs in fact claimed by the State. We find unpersuasive ACF's defense of the pro rata method, in spite of the lack of correlation with the RMS, on the basis that this is a question of allowability of costs, rather than a question of allocability (which is the only question addressed by a CAP). Allocability is one aspect of allowability. To be allowable under Title IV-E as administrative costs, the costs must be for the type of activity listed in 45 C.F.R.  1356.60(c), they must be the type of costs which are generally allowable under applicable cost principles (for example, only certain types of compensation are allowable); and they must be of benefit to the Title IV-E program. Here, ACF did not question whether the type of activity claimed was within those listed in  1356.60(c) as necessary for the proper and efficient administration of Title IV-E. Nor did ACF question whether the type of cost (such as salaries of the City employees) was allowable under the applicable cost principles. Essentially, ACF's disallowance is premised on the idea that administrative costs did not benefit the Title IV-E program if they were incurred in relationship to an ineligible child. This is a question of allocability. The method of determining allocability to Title IV-E is established in the State's CAP. An RMS system focuses on activities, and not directly on numbers of eligible children or eligible payments. Neither party here provided complete information on what activities are covered by the RMS or how particular activities are allocated. From past cases, however, we can say that if activities identified in the RMS are performed only to meet requirements of Title IV-E, and do not benefit any other program, the State's allocation of the entire costs of these activities to Title IV-E would not be called into question based solely on ACF's findings here. 21/ Moreover, the State presented evidence that, with respect to certain preplacement activities (such as preparing for judicial determinations), the RMS allocated these activities to Title IV-E if they were performed for children who had a plan of foster care that potentially would qualify as Title IV-E eligible within a reasonable time after placement. State Affidavit of John Reilly at 2. Some of these children may have never been placed in foster care. What this means is that there is not a proportionate relationship between the costs claimed by the State under the RMS and the allowability of maintenance payments claimed by the State. We do not find here that there is absolutely no relationship between the RMS and identification of children as Title IV-E eligible. If activities identified in the RMS may be allocated to Title IV-E under the RMS only if performed for a child who meets Title IV-E requirements, ACF could reasonably infer that the social workers responding to the RMS incorrectly identified children as Title IV-E eligible at the same rate as children in the sample were erroneously determined eligible by the State. Making adjustments to the RMS results to account for inaccurate responses would be consistent with the approved CAP (assuming the CAP can reasonably be read as permitting adjustment where responses are based on inaccurate information). See n.20. We conclude, however, that ACF's method here was flawed because ACF did not establish any rational relationship between the pro rata method and the CAP method in fact used to generate the State's claims for administrative costs. D. Use of the pro rata method alters the approved CAP methodology in an unauthorized way. As explained above, retroactive changes to CAPs are permitted only in very limited circumstances. Here, ACF in effect superimposed an allocation method on top of the approved RMS method, thus altering the allocation (at least with respect to Title IV-E). We cannot uphold a disallowance based on this retroactive change since ACF did not show that the approved CAP methodology resulted in a substantial inequity to the federal government, nor that the information the State submitted with its proposed CAP was incomplete or inaccurate, nor that it violated federal law. Implicit in ACF's position here is ACF's view that the State overclaimed administrative costs under Title IV-E by a substantial amount. That position, however, is based on the pro rata method of calculating the overstatement and therefore contingent on the reasonableness of the pro rata method. ACF attempted to justify its further allocation by stating that the information provided by the State was inaccurate. The information ACF referred to, however, was information provided by social workers responding to the RMS. ACF did not allege any inaccuracy in the information the State submitted with its CAP proposal regarding State organization and other matters relevant to approving the CAP methodology. We also note that a CAP methodology should not be evaluated only in light of one aspect of one program, but must be considered from the standpoint of the affected programs as a whole. See DAB No. 963, at 20-21. Also, as we have noted before, the CAP has particular importance in Title IV-E in allocating costs which may be of the type which ACF regulations recognize as activities that could be allocable in their entirety to more than one program. In sum, ACF's imposition of the pro rata methodology on top of the approved CAP methodology amounted to a retroactive change to the CAP, which ACF did not show was warranted. E. The pro rata method is otherwise unreasonable. The pro rata method is an unreasonable method of calculating unallowable administrative costs under the particular circumstances here. The pro rata method is based on the assumption that the State overclaimed administrative costs proportionately to the amount of its unallowable maintenance claims. This assumption is not reasonable. As the State pointed out, the State could, under applicable requirements and its RMS, allocate to Title IV-E costs of some administrative activities which are not directly linked to the Title IV-E eligibility of particular children. For example, the RMS would identify as costs of Title IV-E certain preplacement activities performed on behalf of children who were candidates for Title IV-E foster care, but for whom placement in foster care was ultimately avoided, so that no maintenance payments were made. State Reply at 12-17; State Affidavit of John Reilly. 22/ ACF disagreed with parts of the State's analysis of particular Title IV-E administrative activities. 23/ But ACF did not assert that all of the administrative activities for which the State claimed costs were allocable to Title IV-E only if performed on behalf of Title IV-E eligible children. Also, the auditors' choice of a pro rata method based on amount of unallowable maintenance payments (rather than numbers of children erroneously determined eligible) incorrectly assumes that if higher payments were made, this meant that the child was in foster care for a longer period, and that the administrative expenditures associated with that child would be proportionately greater than the costs of a child for whom the maintenance payments were less. The State pointed out that this is not a reasonable assumption, for two reasons. First, most of the administrative costs are incurred in actually placing a child into foster care. (The State said that ACF's own estimate was that 25 to 40 percent of administrative costs were associated with foster care candidates. State Br. at 24.) Second, some children's payments are higher solely because they have special needs or because of the type of placement. (The State said that in 1984 the basic rate for a child in a group home was $1,645.50 to $1,980 per month depending on special needs, while the rate for a child under five in a foster home was $242 per month. State Reply at 20.) ACF did not directly dispute this (and, indeed, acknowledged that most of the administrative costs would be associated with placement); ACF asserted, however, that more administrative costs would likely be incurred for children who were in care longer. This may be true generally, but does not establish that administrative costs are higher in direct proportion to the amount of maintenance payments. 24/ Contrary to what ACF argued, past Board decisions do not support retroactive use of a pro rata method under the circumstances here. ACF relied on Office of Management and Budget (OMB) Circular A-87, Attachment A, Section C.1(a) and "basic grant law" for the proposition that "[a]dministrative activities that are not directly linked to any individual child cannot be allocated solely to Title IV-E." ACF Br. at 57. ACF read the Board's decision in Florida Dept. of Health and Rehabilitative Services, DAB No. 821 (1987), as ruling that, under OMB Circular A-87, "a grantee's entitlement to FFP for administrative costs is linked to the allowability of the underlying direct program costs, and that a proportional disallowance of the former is reasonable when the latter are unallowable" ACF Br. at 58. This reading improperly derives a general principle from a Board analysis dependent on the specific facts of the case. In Florida, the administrative costs at issue were overhead-type costs allocated using a method called an indirect cost rate. By definition, an indirect cost rate is a ratio of indirect costs (those that cannot be readily identified with a specific cost objective) to a direct cost base. The point in Florida was that the indirect cost rate had to be applied to allowable direct costs of the project to determine the allowable indirect costs of that project. Thus, a disallowance of direct costs would automatically justify a proportionate disallowance of indirect costs. Similarly, in Pennsylvania Dept. of Public Welfare, DAB No. 563 (1984), Pennsylvania admitted that its claim for administrative expenditures had been based on a calculation of a pro rata percentage of its expenditures for foster care maintenance payments (under Title IV-A). Pennsylvania did not dispute the agency's pro rata calculation of an administrative cost disallowance. ACF also misconstrued Board decisions specifically addressing allocation of costs to Title IV-E. ACF failed to distinguish between cases which addressed prospective amendments to CAPs and those which addressed retroactive imposition of new allocation methods. The importance of this distinction is illustrated in the Board's decision in Oklahoma, DAB No. 963. In that decision, the Board reversed a disallowance of costs incurred prior to October 1, 1984 for recruiting and approving foster family homes and allocated to Title IV-E pursuant to an approved CAP. The Board found that it was within DCA's discretion to approve a CAP methodology allocating to Title IV-E all costs of recruiting and approving homes. Oklahoma had provided evidence that the entire pool of recruited and approved homes served to fulfill Title IV-E requirements and that virtually every home was used by a IV-E child at some time. Thus, the total costs could be considered as providing a benefit to Title IV-E, and the nature of the costs did not make it inequitable to charge Title IV-E. On the other hand, the Board upheld a disallowance of costs incurred after October 1, 1984, since ACF provided notice to Oklahoma as of that date that it was required to allocate recruitment and approval costs among Title IV-E and other benefiting programs using a pro rata method, and Oklahoma had agreed to amend its plan accordingly. In sum, a pro rata method of allocation is not always required, nor is it always reasonable. ACF here improperly superimposed a pro rata allocation on top of the approved allocation method, based on incorrect assumptions about the relationship between the allowability of the State's maintenance payments and its administrative claims. F. Our decision does not preclude ACF from further examining whether the State overstated its administrative costs. Our decision does not preclude ACF from further examining the State's administrative cost claims, the method of allocation, and its relationship to the accuracy of eligibility determinations. Our decision here is a narrow one, limited to the particular pro rata method ACF applied on top of the RMS allocation method for social worker costs, because -- o It is likely the State's numerous errors in determining eligibility (which we assume for purposes of this decision) also affected its identification of Title IV-E administrative costs. o The information in the record on the State's approved CAP is very general. The DCA approval letter indicates that some costs may have been allocated using a method different from the RMS. Also, since we do not have the instruction forms or other detailed information concerning the RMS, we do not know the extent to which RMS responses regarding certain activities may have included information about children's eligibility. o The State provided no support for its assertion that the parties agreed not to adjust RMS data to account for subsequent determinations about eligibility. Absent any specific agreement not to make such adjustments, it is reasonable to assume that such adjustments should be made to ensure the accuracy of the RMS percentages and equitable distribution of costs. Thus, while we reverse the disallowance calculated using the pro rata method superimposed on the RMS results, our decision does not preclude ACF from taking a further disallowance based on a reasonable method, consistent with the approved CAP. Since we still need to address ACF's findings in individual sample cases, and this may affect the numbers and kind of errors upon which any further determination is based, we suggest that anything other than a preliminary analysis by ACF may be premature at this point. If, after our decision on individual sample cases, ACF takes a further disallowance of administrative costs, the State would have an opportunity to appeal that determination to the Board. Of course, the parties may at any time engage in negotiations to resolve issues concerning either or both types of cost. Conclusion We uphold the disallowance of $64,123,732 FFP in foster care maintenance payments, subject to downward adjustment based on our further review of ACF's findings in individual sample cases. We reverse the disallowance of $27,991,567 FFP in administrative costs calculating using the pro rata method, but do not preclude ACF from recalculating an administrative cost disallowance using a reasonable method, consistent with the approved CAP. _______________________________ Judith A. Ballard _______________________________ Donald F. Garrett _______________________________ Cecilia Sparks Ford Presiding Board Member 1. The State asserted that ACF's reference in its brief to "misspent" funds was intended "to create the erroneous impression that federal funds were spent by the State on concerns far removed from child welfare." State Reply at 42-43, n.16. ACF used this term when discussing Bell. The Supreme Court has used the term "misused" funds to refer to State expenditures which did not meet statutory or regulatory conditions for federal funding. Addressing Bell in a later case, the Supreme Court noted that "funds were misused if the State did not fulfill its assurances that it would abide by the conditions [of the program at issue]." Bennett v. Kentucky Dept. of Education, 470 U.S. 657, 664-665 (1983). 2. The Board has in the past interpreted the overpayment provision in section 1903(d) of the Act as permitting adjustments of FFP claimed in excess of a state's entitlement, even where the state could not recover from Medicaid providers who had been overpaid by the state. Board decisions on this issue have been upheld by three courts of appeals. Massachusetts v. Secretary, 749 F.2d 89 (1st Cir. 1984), cert. denied, 472 U.S. 1017 (1985); Perales v. Heckler, 762 F.2d 226 (2d Cir. 1985); and Missouri v. Bowen, 804 F.2d 1035 (8th Cir. 1986). Congress ultimately amended Title XIX to authorize states to retain FFP where a state had overpaid a provider but could not recover from the provider because the provider was bankrupt or the payments were uncollectible. See section 1903(d)(2)(D) of the Act. No parallel provision permits a state to retain FFP in foster care maintenance payments made on behalf of ineligible children. 3. Section 1116(d) provides -- Whenever the Secretary determines that any item or class of items on account of which Federal financial participation is claimed under title I, X, XIV, XVI, or XIX, or part A of title IV, shall be disallowed for such participation, the State shall be entitled to and upon request shall receive a reconsideration of the disallowance. In 1968, section 1116 of the Act was amended to delete "IV" which followed "I" in the sequential list of titles and to add "or part A of subchapter [title] IV" following "XIX of this title." See Pub. L. No. 90-248 (1968). As noted earlier Title IV-E was enacted in 1980 to be effective no later than October 1, 1982. 4. In footnote 12 in Suter, the Supreme Court contrasted the wording of the state plan requirement in section 471(a)(15) with the wording of section 472(e) (providing that "[n]o federal payment may be made under this part" for a child voluntarily placed in foster care for more than 180 days unless within that period there is a judicial determination that placement is in the best interest of the child). The Court said that this showed that Congress knew how to impose precise requirements aside from the submission of an approvable state plan. The State asserted that this meant that the requirement in section 472(e) was the only provision which could be enforced independently from compliance procedures. Suter does not support this proposition, however, since footnote 12 referred to 472(e) merely as an example of one of the sections imposing precise requirements. The footnote does not say that 472(e) is the only such requirement. A careful analysis of the statutory scheme and the legislative history of the Child Welfare and Adoption Assistance Act shows that section 472(e) establishes an additional condition which must be met if a state opts to cover children under voluntary placement agreements, beyond those conditions which must be met for children under court orders. 5. In Pennhurst, the Supreme Court rejected the argument that acceptance of federal grants under the Developmentally Disabled Assistance and Bill of Rights Act required states to provide mentally handicapped persons with appropriate treatment in the least restricted environment. The Court observed that such a requirement would have imposed a massive and largely indeterminate financial obligation on the states, and stated: "Congress must express clearly its intent to impose conditions on the grant of federal funds . . . ." Here, the intent was clear that section 472 requirements were conditions for funding. Moreover, the Supreme Court has clarified that "Pennhurst does not suggest that the Federal Government may recover misused federal funds only if every improper expenditure has been specifically identified and proscribed in advance." Bennett v. Kentucky Dept. of Education, 470 U.S. 656, 666 (1984). 6. In Wilder, the Supreme Court concluded that section 1902(a)(13) created rights enforceable by hospitals in a private action. 7. As ACF noted, the State's brief is misleading. ACF did not require that the case record fully support the court order since it recognized that testimony or other material available to the court but not in the case record might have been the basis for the court's determination. 8. The State maintained that in these cases the transfer was of "physical custody" only, not "legal responsibility" so that Title IV-E status was unaffected. The State indicated that it would make this argument during the later proceedings, but that for purposes of its arguments here would concede that this transfer rendered the children ineligible. 9. The State conceded that these children were ineligible for purposes of its arguments here, reserving until later proceedings its argument that the auditors improperly found that these facts rendered the children ineligible. State Br. at 68. 10. Under section 472(a)(4) a child removed from the home of a specified relative and placed in foster care (foster family home or child care institution) as the result of a voluntary placement or judicial determination is eligible for foster care maintenance payments if -- such child-- (A) received aid under . . . [Title IV-A (AFDC)]in or for the month in which such agreement was entered into or court proceedings leading to removal . . . were initiated, or (B)(i) would have received such aid in or for such month if application had been made . . . or (ii) had been living with a relative specified in section 406(a) within six months prior to the month in which such agreement was entered into or such proceedings were initiated, and would have received such aid in or for such month if in such month he had been living with such a relative and application therefor had been made. 11. As ACF pointed out, the State's position here is diametrically opposed to the position which the State took in Mercy Hosp. of Watertown v. New York State Dept. of Social Services, 79 N.Y. 2d 197 (1992). See ACF Ex. 3. 12. We also note that part of the audit period here had passed before Louisiana was issued, in October 1984. The State provided no evidence of when it received the Louisiana decision, nor of how it interpreted Louisiana at the time. 13. In its reply brief, the State argued that the sample did not meet the standard for having each item in the universe have an equal chance of selection since "children in care during both 1984 and 1985 were not removed from the universe for 1985." State Reply at 44-45. The State provided no scientific opinion to support its view that this created a defect in the sample. Moreover, the State's argument is logically sound only if you assume that the relevant universe is children for whom a payment was made in either 1984 or 1985. This is in our view not a correct assumption. The universe was constructed as each child for whom a payment was made in 1984 plus each child for whom a payment was made in 1985. The fact that a child for whom a payment was made in 1984 may have also been a child for whom a payment was made in 1985 is irrelevant. Treating the child as one item in 1984 and as another item in 1985 is consistent with the stratified sampling method used. ACF submitted an affidavit by an OIG statistical sampling expert which indicates that the projected disallowance was calculated based on "the amount of payments made on behalf of the children in our sample for the applicable fiscal year from which they were selected" and for which the State claimed FFP. Thus, even if the same child were selected for both fiscal years, the associated payments would be different. We also note that, after the original listings provided by the State could not be validated, the State assured the auditors that its new listings would be complete and that the new listings "assure that all cases have an equal chance of being selected for review." ACF Ex. 1 (June 2, 1989 Memo for Record). 14. This Department has traditionally distinguished between reviews (generally conducted by the program agency) and audits (now conducted by OIG). See 45 C.F.R. Part 201, Subpart B (1970 - present). The State argued in its reply brief that the standards for a formal audit were not met because there was no reconciliation of expenditures to claims for FFP since OIG did not audit the financial records supporting the claims. State Reply Br. at 49, n.17. However, contrary to the State's assertion, the record shows that the auditors did review financial records, both to determine the amount of payments made on behalf of children in the sample during the applicable fiscal year, and to determine whether such payments had been claimed for FFP. We know of no reason why these steps should not qualify as the kind of reconciliation referred to in audit standards. 15. The State objected to ACF's reliance on, and inclusion in the record of, ACF Ex. 2, OIG Chapter 6-20, Sampling and Estimation Techniques in Auditing, because this policy is dated December 31, 1990. We see no reason to delete the 1990 policy from ACF's appeal file, but we do not rely on it. We address the State's arguments with reference to the 1986 policy cited by the State. 16. The State argued that because Congress has barred HHS from taking AFDC disallowances based on quality control results for FYs 1981-1990, "any eligibility errors attributable to an AFDC child in foster care have been forgiven." State Reply at 52. The State pointed to nothing in the referenced legislation which would indicate that Congress intended to "forgive errors" or to preclude disallowances under Title IV-E simply because they may have been based on the same errors as were encompassed in Title IV-A disallowances. In any event, we have insufficient information on the specific errors found here to determine whether the alleged errors were in fact made in determining eligibility as part of the IV-A program. The errors may have been made by the State in determining or documenting whether the foster child had been receiving assistance under Title IV-A, or in applying Title IV-A eligibility requirements as part of the process of determining eligibility for Title IV-E. Neither of the latter types of errors are encompassed in AFDC quality control samples. 17. Section 474(a)(3) also authorizes FFP at a 75% rate for certain training costs found necessary for the proper and efficient administration of the state plan. ACF apparently applied the pro rata method to the State's claims for training expenditures. Audit Report at 32 (State Ex. 2). The record does not clearly indicate how the State allocated training costs to Title IV-E. If the State used a caseload count of eligible Title IV-E children, nothing in our decision here would preclude a disallowance of training expenditures calculated by substituting a more accurate caseload count, adjusted based on a statistically sound projection from our ultimate findings on individual sample children. 18. This standard of deference is supported by the regulations establishing the process for amending a CAP. Under the regulations, an approved CAP amendment may have retroactive effect only if "needed to avoid a significant inequity to either the State or the Federal Government," or if the information provided by the State is later found to be materially incomplete or inaccurate, or if the previously approved CAP violates federal law. 45 C.F.R.  95.515. 19. While ACF did not directly challenge this assertion, the State's own evidence raises a question concerning the accuracy of the assertion. DCA's approval letter conditions approval of the CAP on the State's submitting "revised RMS and SSRR forms and instructions" to DCA and furnishing the "results of the RMS and SSRR" on a quarterly basis. State Affidavit of Roger Nelligan, Att. This indicates that the State had another system -- SSRR -- for claiming at least part of the IV-E administrative costs. Moreover, the State's own description of the RMS appears to limit it to allocation of one particular cost pool, related to social services. It is not clear whether this cost pool would include all of the costs allocated to Title IV-E. As discussed below, this is one reason why our decision does not preclude ACF from further examining specifically what the CAP provided, how it was implemented, what the results were, and whether those results may appropriately be correlated with the results of the case sample. 20. The State asserted in its reply brief that ACF and the State "agreed that administrative costs would be allocated based on a child's eligibility determination at the time the costs were to be divided among the various programs." State Reply at 25. The State also said that under the agreed methodology the State was not required to make adjustments based on a subsequent finding of ineligibility, nor could the State retroactively increase its claims if it subsequently determined that a child was eligible. State Reply at 25. The State provided absolutely no support for these assertions, however, and ACF was not given an opportunity to respond. If the State provided satisfactory evidence that both parties had agreed that responses to the RMS be based on eligibility determinations at the time the response was given and that no subsequent adjustments for errors would be made, this might preclude a disallowance which was in effect an adjustment for eligibility determination errors. Absent such evidence, however, it is more reasonable to assume that the RMS methodology would not preclude such adjustments where they are otherwise reasonable. 21. For example, the State may have a rate-setting system established solely to set rates for Title IV-E maintenance payments to child care institutions, and the RMS may allocate the entire costs of rate-setting to Title IV-E. The audit findings alone do not provide a basis for disallowing any of these costs. 22. We express no opinion here on whether this State practice was consistent with the approved CAP. Indeed, our analysis on the whole was hampered by the fact that neither party provided the relevant CAP documents. 23. For example, the State asserted that costs of determining eligibility for Title IV-E were properly considered Title IV-E costs, even where the child is found to be ineligible (that is, where there is a negative eligibility determination), but ACF disputed this. The parties' arguments relating to the costs of "negative eligibility determinations" misstate the issue somewhat. Here, ACF found that the State either erroneously claimed the children as eligible or, in some cases, failed to maintain documentation of eligibility. The State is correct that it does not automatically follow from these findings that the State improperly claimed during the audit period costs of determining eligibility for the children at issue. Indeed, some of the eligibility determinations may have been made in prior years. Moreover, if the State's CAP provided for allocating all of the costs of determining eligibility to Title IV-E, regardless of the results of the determination, ACF cannot reasonably superimpose a further allocation based solely on the audit findings. 24. We recognize that the auditors were in a sense being conservative in using the percentage of payments found in error, rather than the percentage of children for whom those payments had been made. The auditors said that 76.6% of the children (compared with 52.6% of the payments) were in error. Thus, choice of a ratio of unallowable payments to total payments claimed (rather than children erroneously determined eligible to the total number of children for whom payments were claimed) appears to have benefited the State. The choice was nonetheless based on an assumption which ACF has not supported as a reasonable