GAB Decision 319
June 30, 1982 California Department of Social Services; Docket No.
81-183-CA-HD Settle, Norval; Teitz, Alexander Garrett, Donald
The California Department of Social Services appealed a decision by
the Office of Human Development Services, disallowing $18,851,690 in
federal financial participation claimed under Title IV-A of the Social
Security Act. The disallowance was based on a federal audit which used
a statistical sampling technique to determine the amount claimed by the
State from July 1, 1975 through June 30, 1979, for Aid to Families with
Dependent Children -- Foster Care (AFDC-FC) payments, for children who
were ineligible for that program.
Although the State originally challenged the sampling method used by
the auditors, the State withdrew this issue during Board proceedings,
having obtained an expert opinion that the sampling technique was
statistically valid. State's Letter of March 22, 1982. In addition,
based on arguments and documentation submitted by the State, the Agency
agreed that six of the sample cases found by the auditors to be
ineligible were in fact eligible for AFDC-FC payments. /1/ The Agency,
therefore, revised the disallowance amount to $17,784,137.50. /2/
The major issue remaining in this appeal is whether this disallowance
is inconsistent with Agency policy, developed as part of its quality
control (QC) system, to disallow for errors in determining eligibility
(2) for AFDC based on a sample only when the errors exceed a specified
tolerance level.
For reasons stated below, we conclude that Agency policy for AFDC-FC
eligibility errors, as applicable to the facts before us, was to
disallow only for individually identified errors and that, given the
history of the quality control disallowance provisions, the Agency was
unreasonable under the circumstances of this case in disallowing the
amount extrapolated from the sample of claims. We do not, however,
agree with the State that the statute requires that FFP be provided in
errors within a reasonable tolerance even where individually identified.
Accordingly, we uphold the disallowance of the amounts associated with
individual sample cases, but reverse the disallowance to the extent it
was based on extrapolation from the sample.
This decision is based on the parties' written submissions and the
transcript of a conference held before the Presiding Board Member. /3/
Program Background
The long and rather convoluted history of the quality control (QC)
system, and of provisions for fiscal disallowances based on
extrapolation from QC samples, has been discussed in previous Board
decisions. California Department of Health Services Decision No. 170,
April 30, 1981; California Department of Social Services, Decision No.
235, January 7, 1982; Maryland Department of Human Resources, Decision
No. 246, January 18, 1982. For purposes of this appeal, the following
aspects of that history are important:
- In 1971, regulations were issued requiring that state plans under
Title IV-A, among other titles of the Social Security Act, contain
provisions for a quality control system for measuring and reducing
errors made in determining recipient eligibility and payment amounts.
45 CFR 205.40, 36 Fed. Reg. 3866, February 27, 1971. Section 205.40(
b)(1)(i) of the regulations provided that states should "(apply)
sampling (3) methods and schedules in accordance with instructions
prescribed by the Social and Rehabilitation Services." /4/
- In 1972, a Quality Control Manual was published. This Manual
provided that the "QC system described in these instructions . . . does
not include . . . AFDC foster care, . . ." (SRS 72-21205, p. I-3; Agency
Supplement to Appeal File, Tab 7. Subsequent revisions of the QC Manual
contained similar provisions. See, e.g., (SRS) 74-04010, p. 3; QQA
Pub. No. 006 (4-78), p. 3; Agency Supplement to Appeal File, Tab 7.
- In 1972, the Department first proposed to take disallowances based
on extrapolation from a QC sample to a state's entire AFDC caseload.
This proposal engendered much controversy, leading to publication of
regulations which would disallow only for errors determined by applying
certain prescribed statistical sampling methods, and, then, only for
errors in excess of established "tolerance levels." The Department
stated that the tolerance levels were based on a recognition that, under
the administrative structures then existing in most states, "a
requirement . . . that states eliminate all erroneous payments, with a
resultant disallowance of Federal financial participation in any
erroneous payments is unrealistic." 40 Fed. Reg. 21737, May 19, 1975.
- QC fiscal disallowance provisions at 45 CFR 205.41, published in
1975 and establishing tolerance levels of 3% for eligibility
determination errors and 5% for overpayment errors, were struck down in
the case of Maryland v. Mathews., 415 F. Supp. 1206 (D.D.C. 1976). The
Court discussed the complexity of the eligibility determination process
and found that "the tolerance levels were arbitrarily established at 3%
and 5% without the benefit of an empirical study" and, therefore,
section 205.41 was framed in an arbitrary and capricious manner and as
an abuse of discretion. 415 F. Supp. at 1214. The court enjoined the
enforcement of the regulation to disallow claims or otherwise adjust
grant awards to the plaintiffs, including the State of California.
Appeal File, Tab 17.(4) - Following the Maryland v. Mathews decision,
the Department retroactively revoked the disallowance provisions,
stating: "(The) revocation . . . results in States being held fiscally
accountable for individually identified payments to ineligibles . . ."
42 Fed. Reg. 14717, March 16, 1977. The revocation notice cited 45 CFR
233.10(b)(1) for the proposition that FFP is available only in proper
payments, absent prescribed tolerance levels.
- Action Transmittal AT-77-30 (APA), March 16, 1977, issued upon the
revocation of the QC disallowance provisions, required states to "adjust
claims for Federal reimbursement to exclude individual incorrect
payments identified by the State or through Federal reviews." Action
Transmittal AT-77-55 (APA), May 17, 1977, stated: "For cases excluded
from the (QC) universe, i.e., AFDC foster care, . . . States were
required to adjust their claims for these individual payments to
ineligibles and overpayments." The transmittal explained that, with the
revocation of 45 CFR 205.41, the policy "always applicable to cases
excluded from the QC universe became applicable to cases included in
that universe."
- The Department subsequently issued revised provisions for
disallowances based on QC samples. 44 Fed. Reg. 12578, March 7, 1979.
These regulations provided that QC disallowances would not be taken
against any state meeting either an absolute standard of error or a
prescribed rate of error reduction. Even then, disallowances could be
waived if a state met any one of a number of waiver conditions. Later
versions included the possibility of waiver if a state was taking timely
corrective action reasonably designed to reduce errors. 45 Fed. Reg.
6326, January 25, 1980. At least as late as 1980, the Department had
never actually taken a disallowance based on extrapolation from a QC
sample.
Case Background
During the time period in question here, the AFDC-FC program was a
component of the AFDC program under Title IV-A of the Social Security
Act. /5/ To be eligible for AFDC-FC payments, a child had to be one (a)
who was removed from the home "as a result of a judicial determination
that continuation therein would be contrary to the welfare of such
child;" (b) "whose placement and care are the responsibility of"
the (5)
appropriate state or local agency; (c) who has been placed in a foster
family home or child care institution as a reslut of the judicial
determination; and (d) who received AFDC in the month court proceedings
were initiated, or would have received such aid if an application had
been made. Section 408(a) of the Act. Foster care provided to children
meeting these requirements qualified as "aid to families with dependent
children." Section 408(b).
Although the record indicates that the Agency had performed audits of
the State's AFDC-FC program for time periods prior to that in question
here, the Agency did not allege that those audits revealed any systemic
problems with the State's administration of the program. However, the
Agency decided that it would perform an audit of the State's AFDC-FC
program over a four-year period, using a random multistaged, stratified
sampling technique. Although the record shows that staff persons from
the State agency were informed concerning the sampling process, it also
shows that from an early stage the State objected to the sampling on a
number of different grounds.
The federal auditors' findings, set out in a final report issued
February 6, 1981, ACN 10251-09, were that the 133 of the 675 sample
AFDC-FC payments were ineligible for FFP. According to the auditors, of
these 133 payments, 87 were for children who were not eligible for AFDC,
38 for children who were not removed from the home as a result of a
judicial determination or who were no longer under a court's
jurisdiction, 6 for children in ineligible institutions, and 2 for
illegal aliens.Audit Report, p. 5. Extrapolation of these findings, as
modified during later proceedings, to the universe of the State's claims
for AFDC-FC payments during the audit period formed the basis for the
Agency's determination that the State had incurred unallowable costs of
$35,568,275 (federal share $17,784,137.50).
On appeal, the State argued that the statute required FFP in payments
which were unavoidable, as established by a reasonable tolerance for
errors. The State argued that, in previous decisions refusing to apply
tolerance levels to individually identified errors, the Board had
misinterpreted the Maryland v. Mathews case regarding the availability
of FFP in erroneous payments. The State further argued that,
irrespective of the provisions in the QC Manual, the regulations on the
QC system did apply to AFDC-FC cases and that, therefore, during at
least part of the audit period here, the Agency was enjoined from
disallowing based on a sample. In any event, the State argued, the
Agency can do no more than apply its policy of disallowing for
individually identified errors. (6) The Agency responded that the QC
system did not apply to AFDC-FC errors, and, therefore, no tolerance
levels should apply. The Agency distinguished the QC sampling system
from the use of a sample here, primarily on the grounds that the QC
system required the State itself to sample in a comprehensive manner on
an ongoing basis. According to the Agency, the policy, set out in the
revocation notice and action transmittals, of disallowing for
individually identified errors was intended to apply only to states'
adjustment for errors they identify and not to preclude the Agency from
disallowing based on federal samples. The Agency relied on the
generally accepted proposition that statistical sampling is a valid
technique for determining amounts of unallowable costs.
Discussion
In previous cases involving use of statistical samples as a basis for
disallowance, the Board has viewed the matter as an evidentiary question
and has held that a sample may provide a reliable basis for determining
the amount of unallowable costs a grantee has charged to federal funds.
See, e.g., University of California -- General Purpose Equipment,
Decision No. 118, September 30, 1980; Ohio Department of Public
Welfare, Decision No. 226, October 30, 1981. Thus, we have upheld
disallowances based on valid statistical sampling methods, even in the
absence of any policy statement specifically authorizing the use of such
methods.
The issue presented here is a different one, however. In our
opinion, the critical factor in this case is that the Agency had an
established policy for when it would take disallowances based on samples
of errors in determining eligibility in the AFDC program. For cases
within the QC universe, the policy was that the Agency would disallow
for all extrapolated errors in excess of tolerance levels. This policy
was promulgated through notice and comment rulemaking, became the
subject of an injunction, was retroactively revoked, and then was
reinstituted with new tolerance levels and provisions for waivers. For
cases within the scope of those errors covered by the QC system but
outside the QC universe, such as AFDC-FC payments, they stated Agency
policy was to disallow only for individually identified errors.
The Agency argued that the policy of disallowing only for
individually identified errors does not preclude disallowances based on
samples because it refers only to adjustment by the states for errors
they identify. Both the revocation notice and AT-77-30 refer to errors
identified "through federal reviews" as well as by states, however.
The phrase "individually identified" must be read in the context
of
the provisions, developed after extensive notice and comment rulemaking,
describing the circumstances under which the Agency would disallow (7)
based on samples. In this context, "individually identified" clearly
means "not based on extrapolation from a sample." The Agency's assertion
that we should read it another way is not persuasive in view of this
context. We also note that the Agency itself has cited to these
policies as a basis for why the Board should uphold disallowances of all
individually identified errors without applying a tolerance, and has
explained the rationale for applying tolerances as inextricably linked
to use of extrapolation from a sample. See, Decision No. 246, pp. 3-4;
Decision No. 235, p. 2. /6/ Although the Agency asserted that it later
began to develop sampling policies for disallowances in all programs,
based on standards promulgated by the American Institute of Certified
Public Accountants, we do not see how this can supersede the express
policy for AFDC eligibility determination errors set out in Agency
issuances, where the State was given no notice of a change.
Thus, we conclude that this disallowance, to the extent it was based
on extrapolation from a sample, was inconsistent with Agency policy to
disallow only for individually identified AFDC-FC eligibility
determination errors. While this is the primary rationale for our
decision here, we also conclude that, given the history of the QC
disallowance provisions, the Agency acted unreasonably in extrapolating
from this sample without applying a tolerance level.
We are not convinced by the Agency's argument that this sample is
distinguishable from a QC sample. According to the Agency, federal
auditing by sample has a built-in tolerance level "by the fact that only
a limited portion of the program is reviewed at any given time and that
. . . many kinds of errors are (never identified) by federal auditors."
Transcript (Tr.), as corrected, pp. 39-40. (In contrast to the quality
control program," the Agency argued, "here the auditors are looking
at
claims over a limited time period and at a limited subject matter." Tr.,
p. 39. In advancing these arguments, however, the Agency is attempting
to equate a somewhat tenuous protection resulting from circumstance and
practicality (and the fact that federal auditing resources are limited)
to the legal protection afforded to the states by a policy of
disallowing based on extrapolation from samples only when a state's
error rate exceeds a specified tolerance. We find it hard to accept the
Agency's position, given the facts of this case. The disallowance here
covered the State's entire AFDC-FC caseload over a four-year period.
Since the Agency is not permitting any tolerance for errors, nor any
possibility of waiver (8) (even though the State has taken some
corrective actions), it is potentially more burdensome than any QC
disallowance.
This disallowance is also unfair in another respect. Although the
Agency could not state the precise reason why AFDC-FC cases were
excluded rom the QC universe, the Agency recognized that AFDC-FC
eligibility determinations can be more complex than basic AFDC
determinations. Agency Brief, p. 13; Tr. pp. 34-36. Thus, a likely
reason for exclusion of the cases was a recognition that, if AFDC-FC
disallowances were based on extrapolation from a sample, an even higher
tolerance should apply. Indeed, errors in the basic AFDC eligibility
determination process itself were the reason for disallowing almost
two-thirds of the sample payments disallowed here. We do not see much
support for permitting a tolerance for such errors when they occur in
the basic AFDC program but not when they are part of an even more
complex eligibility determination process.
We are not persuaded by the Agency's argument that, by using the
lower limit of the 90 percent confidence level to determine the
disallowance amount, the Agency is allowing a tolerance for errors. The
QC disallowance provisions protected the states by using both the
tolerance levels and the lower limits of confidence. Moreover, the
purpose of using the lower limit was explained as assuring that any
disallowance would reflect actual error rates (40 Fed. Reg. 32955,
August 15, 1975), whereas use of a tolerance level is, in effect, a
policy not to disallow for certain errors which are unavoidable, given
the complexity of the eligibility process.
The Agency has here unreasonably substituted another statistical
sampling technique for QC sampling to determine AFDC eligibility errors,
and then disallowed on the basis of the sample without affording the
State the benefit of a tolerance.
Thus, we conclude that, to the extent it was based on extrapolation
from the sample rather than on individually identified errors, this
disallowance was inconsistent with Agency policy and was unreasonable in
view of the history of the QC disallowance provisions. In reaching this
conclusion, we do not fully adopt either party's position in the case.
Accordingly, we have set forth below a more complete analysis of various
issues raised by the parties. We have also explained the limited scope
of this decision and how it is consistent with previous Board decisions
on use of tolerance levels.
I. Interpretation of the statute.
In Decision No. 170, cited above, the Board addressed the question of
whether unavoidable errors must be considered a necessary cost of the
(9) Medicaid program, in which the federal government must
participate.As part of its discussion of this issue, the Board stated:
The Court in Maryland rejected the states' argument there that
erroneous payments could be considered payments made as "aid or
assistance" under Title IV-A. 415 F. Supp. at 1211-1212. Moreover, the
Court discussed the Secretary's authority to promulgate a regulation
allowing FFP in erroneous payments up to a reasonable established
tolerance level as deriving from the Secretary's authority under Section
1102 of the Act. /7/
Decision No. 170, p. 9.
Thus, the Board concluded, the Court's holding does not necessarily
mean that the statute may be directly interpreted to provide for
tolerances. In Decision No. 246, also cited above, in discussing the
Agency's authority to recover FFP in payments to ineligibles, the Board
stated more affirmatively: "Section 403(a)(1) (of the Act) allows FFP
only for payments under a State plan, and has been interpreted to mean
that improper payments 'are not to be matched by federal funds,'"
quoting from Maryland v. Mathews, 415 F. Supp. at 1212.
In its appeal brief here, the State disagreed with the Board's
interpretation of the Maryland v. Mathews opinion on this point.
According to the State,
The argument made to the Court was that all payments, erroneous or
not, made in the operation of the state plan were entitled to FFP. 415
F. Supp. at 1211. The Court did not agree with that conclusion.The
Court's conclusion, however, was not that no erroneous payments were
entitled to FFP.
Appeal Brief, p. 17 (emphasis in original)
The State took the view that, when the Secretary exercises his
authority under section 1102 to set a tolerance level,
Anything below that level is presumed to have been made pursuant to
the state plan because the (10) elimination of all errors is totally
unrealistic as well as impossible. The Secretary does not have the
power to refuse FFP in all erroneous payments and, in fact, to have
issued regulations refusing FFP in levels below three and five percent
was an abuse of discretion because those levels were without empirical
support and, based on available evidence, too low.
Appeal Brief, pp. 18-19 (emphasis in original)
In support of its position that the Department itself recognized this
view of the Maryland v. Mathews decision, the State cited a statement
from the revocation notice that "the Court ruled that the Act does
contemplate Federal matching of a reasonable level of erroneous
payments." 42 Fed. Feg. 14717. The State also argued that its
interpretation was "supported by the Secretary's own justifications and
comments when proposing and adopting the first disallowance/tolerance
level regulations." State's Reply Brief, p. 5. The State relied on
statements to the effect that those regulations would, for the first
time, "exclude" FFP in all overpayments and payments to ineligibles,
arguing that this meant that the statute itself did not exclude FFP.
We do not agree with the State that the statute requires FFP in
erroneous payments up to a reasonable tolerance. While the above cited
statement from Decision No. 170 may have somewhat mischaracterized the
states' argument in Maryland v. Mathews, the Court's response that
improper payments "are not to be matched by federal funds" supports
the
Agency position that, generally, FFP is not available in payments which
do not meet state plan provisions.
The State's reliance on Department statements is misplaced. In
context, these statements do not contradict the Agency's position here
that, absent regulations establishing tolerance levels, FFP is not
available in erroneous payments. The Court's conclusion was that a
regulation establishing a tolerance level was not inconsistent with the
Act and was therefore within the Secretary's section 1102 authority.
That the Act "contemplates" matching when the Secretary has exercised
such authority does not necessarily mean that the Act requires such
matching in other circumstances.
Preamble statements to the early QC disallowance provisions
concerning exclusion of FFP in all erroneous payments must also be taken
in context. With the adoption of the first QC disallowance regulations
in 1973, the Department also amended 45 CFR 233.10(b)(1) to "make clear
that Federal financial participation is available only in State payments
made in accordance with approved State plan conditions prescribing
standards of eligibility." 38 Fed. Reg. (11) 8743, April 6, 1973. The
previous version of this section had provided that, with certain
exceptions, "the Federal agency does not ordinarily determine, for
purposes of financial participation, whether State plan provisions which
are not required by the Federal Act, regulations, or policies have been
met." 36 Fed. Reg. 3867, February 27, 1971. Under the earlier version,
therefore, no question of FFP was raised for individuals who were
ineligible under state standards but who would have been eligible if
state standards were as broad as permitted by the federal statute. The
change to explicitly deny FFP in those payments was explained as
follows:
In view of the revised approach toward corrective action based on
information produced by the quality control system, the State plan
itself is the appropriate basis for determining Federal financial
participation, as it promotes sounder and more consistent administration
. . . .
38 Fed. Reg. 8743.
The Agency policy statements concerning disallowance of individually
identified erroneous payments cite to 45 CFR 233.10(b)(1) as a basis for
the position that FFP is not available in such payments. We conclude
that this position is correct.
Moreover, in instituting the QC disallowance provisions, the
Department was for the first time extrapolating to a state's entire
caseload of claims. References to exclusion of all erroneous payments
must be considered in this context.In our view, the QC disallowance
provisions was primarily a disallowance policy, not a policy defining
allowable costs. Therefore, they are not inconsistent with the Agency
position that, generally, FFP is not available in erroneous payments.
II. Applicability of the QC disallowance provisions to AFDC-FC.
The State argued here that the QC regulations did apply to AFCI-FC,
stating,
If the Board is going to uphold a disallowance of AFDC funds based
upon projections from a sample, then it must require application of the
QC tolerance level for those periods of time when such levels were in
effect and must disapprove of disallowances based on projections of
erroneous payments during times when the Secretary was enjoined from
making disallowances based on the three and five percent error rates.
Appeal Brief, p. 14. (12) While we are not upholding this
disallowance to the extent it is based on a projection, we wish to make
it clear that we do not agree with the State as to the scope of the
Court's order.
The regulations struck down in the Maryland v. Mathews case
incorporated by reference the QC Manual, which specifically excluded
AFDC-FC cases from the QC universe.Thus, no disallowance based on
projection from a QC sample would have reflected AFDC-FC errors. We do
not read the Court's order, enjoining the enforcement of the regulation,
as enjoining an act which would not have been taken under the regulation
in any event.
On the other hand, the State argued persuasively that the AFDC-FC
component of the AFDC program was not specifically excluded from the
terms of the regulation itself (read without the Manual instructions),
and that a mere change in the QC Manual instructions could have included
the cases within the disallowance provisions. The regulations apply to
Title IV-A eligibility determinations, and it appears that the exclusion
of AFDC-FC cases from the QC universe was merely a matter of
administrative convenience or because such cases required a higher
tolerance, if disallowances were to be based on samples. These factors
support our conclusion above that the regulation, Manual, revocation
notice and action transmittals together set forth a policy for AFDC
eligibility determination errors. Where such errors are determined by
QC sample, tolerance levels apply. Otherwise, disallowance will be made
only for individually identified errors.
III. Disallowances based on extrapolation from a sample in general.
The Agency relied on previous Board decisions and several court cases
which upheld the use of extrapolation from samples as a basis for
determining disallowance amounts. Our decision here does not contradict
our previous position that, under certain circumstances, a statistical
sampling technique may be a reliable basis for a disallowance. We hold
here merely that, under the particular circumstances of this case, use
of extrapolation was inconsistent with Agency policy and unreasonable.
Our decision is limited to errors in determining eligibility in the
AFDC program, where those errors result from the normal administrative
processes. As the State recognized here, extrapolation might be
appropriate in situations where a state was making erroneous payments
because of noncompliance with a federal program requirement or a
systemic failure in administration.These types of errors are arguably
outside the scope of the QC disallowance policy. (13) Here, there was
no showing that the State's errors resulted from anything more than the
usual difficulties of determining eligibility. Moreover, this is not a
situation where the State agreed to use of a sampling technique to avoid
the burden which might be placed on it by claim-by-claim review. If
such were the case, we might find that the State had agreed to
application of a different policy than would otherwise apply.
We also note that the underlying rationale for accepting statistical
evidence is the practical difficulty of obtaining claim-by-claim review.
See, e.g., State of Georgia v. Califano, 446 F. Supp. 404, 409 (D. Ga.
1977). Here, while we recognize that review of the State's entire
AFDC-FC caseload might not be feasible, individual identification of
errors may be considered a method which the Agency itself chose by
excluding AFDC-FC from the QC universe and then failing to adopt a
specific policy for AFDC-FC disallowances based on samples as
statistical sampling techniques and standards were developed.
IV.Consistency of this decision with previous Board decisions.
In Decision No. 170, the State of California appealed a disallowance
of individually identified claims processing errors made in the
operation of the State's Medicaid program.No regulations providing for
tolerance levels for such errors were in effect during the relevant time
period. California argued, nonetheless, that the Board should read the
Act as permitting a reasonable tolerance for errors and, based on
evidence submitted by the State, determine that the errors were
reasonable, unavoidable, and de minimis. We held that, even if in some
circumstances the Board could, by adjudication, formulate a rule
consistent with Departmental policy as expressed in a later regulation,
it would be inappropriate for the Board to exercise such authority in
the circumstances of that case.
Decision Nos. 235 and 246 involved disallowances of individually
identified errors in AFDC-FC. The Board upheld these disallowances,
without permitting any tolerance for errors, primarily because no
tolerance level for AFDC eligibility errors was in effect for the
relevant time periods and because we concluded that the statute did not
require FFP in erroneous payments.
We agree with the State that, since all of these decisions involved
disallowances of individually identified errors only, the decisions are
distinguishable from this case, which involves a sample. Indeed, the
Agency itself relied in previous cases on individual identification as
the critical factor which meant that no tolerance levels should apply.
(14) We note that, in Decision Nos. 235 and 246, there is language
supporting the Agency position that the QC disallowance provisions did
not apply to AFDC-FC cases, which refers to the QC Manual as an
interpretation of the regulations. Here we have stated instead that the
QC Manual was incorporated by reference into the regulation, which is
technically more correct. The result is the same with respect to the
issue of whether tolerance levels applied to AFDC-FC disallowances.
However, the latter view is more consistent with our position that the
QC disallowance provisions and other Agency issuances established a
complete disallowance policy for AFDC eligibility determination errors.
Conclusion
For the reasons stated above, we reverse the disallowance to the
extent it was based on extrapolation from a sample, but uphold the
disallowance of the individually identified errors in the sample. This
decision does not preclude the Agency from disallowing for any other
individually identified errors in the State's AFDC-FC program during the
audit period.
(15) DATE: December 31, 1982
RULING ON AGENCY'S MOTION FOR RECONSIDERATION
The Office of Human Development Services submitted a Motion for
Reconsideration of the Board's decision in California Department of
Social Services, Decision No. 319, June 30, 1982. Under the Board's
procedures at 45 CFR Part 16 (1981), the Board has the power to
"reconsider a Board decision where a party promptly alleges a clear
error of fact or law. . . ." Section 16.13. Below, we first discuss why
we have determined to reconsider our decision even though there is a
question whether the Motion was prompt. We then reconsider our decision
and conclude that, while the intent of several Agency policy statements
might have been limited in scope, as the Agency alleges, our
interpretation of Agency issuances as a whole was the more reasonable
interpretation. We further conclude that, even if we had erred in
determining what Agency policy was, we reached the correct result in our
decision, based on the alternative ground which we discussed in the
decision and explain more fully here.
I. The Question of Promptness
California objected to the Motion for Reconsideration on the grounds
that the Motion was not promptly filed, that the Agency had waived any
right to request reconsideration, and that the Agency had ample
opportunity to brief the issues during Board proceedings. In support of
its waiver argument, the State submitted a copy of an August 11, 1982
letter, concerning the Agency's intent to review individual cases to
determine the amount associated with individually identified errors
since that part of the disallowance was upheld by the Board.
We do not agree with the State that the Agency waived reconsideration
merely because the Agency acted to implement our decision. The State
did not provide any support for its view that this constituted a waiver,
nor did the State claim to be prejudiced by the action. However, the
State's argument that the Motion was not "promptly" filed has some
merit. The Agency admits at page 12 of the Motion that it received the
Board's decision in early July 1982. It was not until September 10,
1982, that the Agency filed notice of its intent to request
reconsideration, and the Motion was not filed until September 30, 1982.
As the State notes, the (16) underlying issues raised in the Motion are
issues which were considered during Board proceedings.
On the other hand, the Agency has pointed to some mitigating
circumstances (such as Department reorganization) which contributed to
the delay, and has raised important questions concerning the effect of
our decision. Accordingly, we reconsider our decision below, in light
of the new arguments and evidence presented with the Agency Motion.
II. Summary of Decision No. 319
In Decision No. 319, the Board reversed a disallowance of Aid to
Families with Dependent Children -- Foster Care (AFDC-FC) payments to
the extent that the amount of erroneous payments had been determined
through extrapolation from a sample of claims. Although the Board
affirmed its position from previous decisions that a statistical sample
may be reliable evidence of the amount of unallowable costs in a
universe of claims, the Board held that use of extrapolation in this
particular case was inconsistent with the Agency's own policy and was
unreasonable in the historical context. The Board referred to and
summarized other Board decisions describing the development of the
quality control (QC) sampling regulations for the AFDC program and the
controversy over the provisions for fiscal disallowances using
extrapolation from QC samples.
In finding that Agency policy during the time period in question was
to disallow AFDC-FC eligibility determination errors only when they were
identified individually, and not to disallow based on extrapolation from
a sample, the Board relied primarily on the following factors:
The policy, developed through notice and comment rulemaking, for
disallowing eligibility errors in the AFDC Program as a whole was to use
samples taken according to 45 CFR 205.40 and extrapolated under the
conditions prescribed in section 205.41. Although AFDC-FC cases were
within the terms of these regulations, the regulations also refer to
Agency instructions on QC sampling contained in the QC Manual, and
various versions of the QC Manual excluded AFDC-FC cases from the
sampling universe. (The Agency has still not explained why this was
done.) We considered this exclusion of AFDC-FC cases as part of the
Agency's disallowance policy, since these errors were still within the
scope of the QC regulations.
In explaining the effect of the revocation of section 205.41, the
revocation notice and action transmittals refer to a policy of
disallowing for "individually identified" errors. In the (17)
historical context, this clearly means error identification not based on
extrapolation from a sample.
Action Transmittal 77-55 described the revocation policy as the
policy which always applied to cases excluded from the QC universe.
Thus, it was reasonable to conclude that Agency policy concerning
AFDC-FC eligibility determination errors was to disallow only
individually identified errors.
The Agency argument that the revocation notice and action
transmittals referred only to state-identified errors did not have merit
in view of the plain language of the issuances, which includes errors
identified "through Federal reviews."
Thus, the Board's conclusion on Agency policy was based on a reading
of the Agency issuances as a whole, including the QC regulations and
Manual.
III. The Agency's Major Argument on Reconsideration
The Agency's major argument in its Motion is that the board
incorrectly interpreted the Agency's revocation notice and action
transmittals and that, therefore, the Board's decision was erroneous.
The Agency contends that these documents have to be viewed in the
context of the QC program and, so viewed, are only instructions to the
states on how to treat the individual errors identified in the QC
sample. The phrase "through federal reviews," the Agency contends,
has
to be interpreted to mean federal reviews of subsamples of the states'
QC samples and not other types of federal audits or reviews. According
to the Agency, the documents did not address the Department's general
disallowance policy nor limit the Agency's authority to go into a state
and independently review or audit a particular program, using valid
statistical sampling methods to extrapolate the amount of improper
payments.
IV. Our Evaluation of the Agency's Argument
We are not persuaded by the Agency's argument that our conclusion
regarding Agency policy was erroneous, although we concede that the
meaning and intent of the revocation notice and action transmittals may
be subject to varying interpretations. We based our conclusion on a
reading of the Agency issuances as a whole, including the regulations
and Manual. We considered it significant that the AFDC-FC errors here
were within the scope of the QC regulations, although excluded from the
samples by the Manual. Further, we considered the relevant context to
include not only the QC system and how it operated but also the history
of the fiscal disallowance provisions, developed after much negotiating
between the states (18) and the Agency over when and how disallowances
would be based on samples. In this context, the references in later
issuances to disallowing "individually identified" errors clearly
was
meant to contrast with extrapolating from a sample. The Agency has
pointed to nothing which contradicts this interpretation.
Moreover, we think that the Agency view that the notice and action
transmittals were sorely directed to states' adjustments for QC sample
errors is unconvincing for the following reasons:
The phrase "through Federal reviews" used in the documents can
reasonably be read to mean more than federal review of a subsample of a
QC sample, and indeed, in a later action transmittal, SSA-AT-78-16, May
4, 1978, the Agency clarified that the fiscal adjustment policy in
AT-77-55 applied to "all cases in which errors in payment are identified
from any source. . . ." p. 2 (emphasis in original).
In any event, the key statement in AT-77-55 concerning the provisions
which always applied to AFDC-FC cases necessarily refers to errors
outside the QC samples since these cases were excluded from the samples.
While the action transmittals may be primarily concerned with how the
states should adjust their claims for errors identified through the QC
system, they also clearly deal with a change in disallowance policy for
eligibility errors in the AFDC program. Both the revocation notice and
AT-77-30 state that, because of the revocation of the disallowance
provisions in section 205.41, the "Department is now required to
disallow" federal financial participation in all individual incorrect
payments. The possibility of disallowance is the force behind the
requirement to adjust.
The affidavit which the Agency submits in support of its contentions
concerning the underlying intent of the action transmittals is by the
current Associate Commissioner, Office of Family Assistance, Social
Security Administration. Since the affidavit contains no affirmation
that the affiant has any personal, contemporaneous knowledge of the
purpose of the issuances, it does not carry much weight.
Thus, we continue to think that it is reasonable to imply from the
Agency issuances applicable to the particular dispute in question a
policy to disallow AFDC-FC eligibility determination errors only when
individually identified.
On the other hand, as the Agency points out, the issuances do not
state such a policy explicitly. Moreover, the emphasis in the (19)
action transmittals is on the states adjusting for errors, rather than
on disallowances.The Agency has now explained that, while tolerance
levels were in effect, the states were not required to adjust on an
ongoing basis for errors to which the tolerances applied but could delay
adjusting until an error rate was determined. Since tolerance levels
were not used for AFDC-FC cases, the Agency says, the states were
required to adjust for these errors as they were identified. Thus, we
concede that the scope and intent of the Agency policy might have been
limited to states' adjustment for errors in the absence of regulations
permitting tolerances.
Although we concede that the revocation notice and action
transmittals may be interpreted more narrowly, as discussed above, we
continue to think that our original interpretation of the issuances as a
whole was the more reasonable one. In any event, even if the Agency did
not have a policy to disallow only for individually identified AFDC-FC
errors, we would conclude that we reached the correct result in Decision
No. 319. Below, we discuss why and also address additional arguments
raised in the Motion.
V. The Effect of a Lack of Policy for AFDC-FC Errors
As we stated in the decision, and explain more fully here, it was
unreasonable under the circumstances of this case for the Agency to base
its disallowance on extrapolation from a sample. Factors which lead us
to conclude that we could not accept the statistical sampling evidence
here included: (1) that there was no explicit policy for AFDC-FC
disallowances based on samples; (2) that a policy had been developed
establishing conditions for disallowances based on samples in the AFDC
program; (3) that this policy included application of various
safeguards, including allowance of tolerances for errors (which
tolerances were found to be too strict); and (4) that these errors were
generally within the scope of that policy and, indeed, a large
percentage of the errors here were precisely the kind of error covered
by the policy.
Even accepting the Agency's view that the scope of the notice and
action transmittals was more narrow than we previously thought, we are
left at most with a lack of policy concerning AFDC-FC disallowances
based on samples. The Agency still has pointed to nothing which states
an explicit sampling policy for AFDC-FC errors or sets out a "general"
disallowance policy and explicitly applies it to AFDC-FC errors.
Although the Agency does present evidence regarding administrative
practice, we find if of little weight in discerning Agency policy during
the time period in question here. The Agency points to disallowances
allegedly determined by extrapolating from a sample, taken with respect
to cases excluded from the QC universe. These disallowances were not
issued prior to 1980, however, and (20) therefore are not
contemporaneous with the Agency issuances. In addition, nothing the
Agency has submitted shows that these disallowances were for the type of
eligibility determination error we are concerned with here, nor is it
clear whether these samples were performed under circumstances which we
specifically excluded from the scope of our decision, such as where a
state has agreed to sampling. Moreover, the fact that the Agency may
have taken AFDC-FC disallowances based on samples does not alter our
conclusion that it was unreasonable to do so here.
The Agency reasons that a lack of an "explicit" sampling policy
for
these errors means that the Agency therefore may apply sampling methods
and audit policies applicable to all programs. This does not
necessarily follow, however. The Agency did develop a policy for when
disallowances for eligibility determination errors in the AFDC program
would be based on extrapolation from a sample. This policy was
developed with extensive public participation and was the subject of
litigation and negotiation with the states. Given this factor, the
Agency was not free to use a different sampling method for eligibility
determination errors within the scope of the AFDC program merely because
that was Department practice in other programs. Although the Associate
Commissioner's affidavit suggests that, after revocation of the QC
disallowance provisions, the Agency was free to use extrapolation from a
sample for disallowance of any AFDC errors, we note that the Agency has
never asserted that it did in fact take any such disallowances in the
basic AFDC program, and an Agency representative stated that, to the
best of his knowledge, it had not. Transcript, p. 72; see, also, Tr.,
p. 48. We think that this is consistent with our view that to do so
under the circumstances would have been unfair.
The Agency contends that the Board erroneously concluded that a
higher tolerance level should apply to eligibility determinations in
AFDC-FC cases than in the basic AFDC program. The Board never reached
the issue of tolerance levels, however. Since the Agency could not
explain during Board proceedings why AFDC-FC cases were excluded from
the universe, and acknowledged that AFDC-FC eligibility determinations
can be more complex than basic AFDC determinations, the Board merely
stated that "a likely reason" for the exclusion was a recognition
that a
higher tolerance should apply. Decision No. 319, p. 8. The Agency now
explains that this is not so because, in the ovewhelming majority of
AFDC-FC cases, the basic AFDC determination has already been made in the
other elements of eligibility can be readily determined. Motion, p. 10.
We do not know, however, if the Agency, in reviewing AFDC-FC cases,
would count as an error a mistake in a basic AFDC determination made
prior to the time the foster care elements are determined. If so, the
possibility of error in AFDC-FC cases would be greater and our statement
was not incorrect. Even accepting the Agency (21) position as true,
however, does not undermine our basic conclusion that extrapolating from
a sample but applying absolutely no tolerance would be unfair under the
circumstances. There were other reasons supporting that conclusion,
including the finding that errors in the basic AFDC eligibility
determination process itself were the reason for questioning almost
two-thirds of the sample payments found to be in error.
Thus, we reaffirm our conclusion that, since the Agency had developed
a policy for disallowing based on samples in the AFDC program and since
the errors here were within the scope of that policy, to base this
disallowance on extrapolation from a sample was unreasonable. We will
not accept the use of statistical sampling methods under these
circumstances, and, therefore, we uphold the decision we reached
previously.
IV. The Type of Error Included in the Disallowance
The Agency in its Motion also raised the question of whether the
errors discovered by the audit here were the type of error covered by
the QC system, occurring in the normal course of the eligibility
determination process, or derived instead from systemic problems with
the State's administration of the program. In support of its allegation
that the errors here were the latter type, the Agency merely points to
language in the audit report indicating that mistakes made in
determining AFDC eligibility occurred because information from the
social worker's or probation officer's files had not been provided to
the eligibility worker. From this, the Agency concludes that the errors
here were caused by a failure of the State to have appropriate
administrative processes in place rather than "with the complexity of
the calculations" in determining eligibility. Motion, p. 11.
In recognizing that certain "systemic" errors might be
distinguishable from QC-type errors, however, the Board did not limit
the range of normal eligibility determination errors to errors arising
from the complexity of the calculations involved. Rather, the Board
focused on the difficulties of the eligibility determination process, as
discussed in the Maryland v. Mathews decision. Information gathering
and transfer may be one aspect of this. Certainly, the QC system covers
more than calculation errors.
Moreover, the auditors' finding that some social workers or probation
officers did not provide full information to the eligibility workers is
not tantamount to a finding that there was a systemic failure in the
State's administrative procedures. Although the audit report (at page
1) does state that "the eligibility workers were not provided adequate
guidance" and that the State had requested funds from the legislature to
develop performance standards for the (22) AFDC-FC program, the audit
report does not elaborate on these points. The Agency did not develop
these points in its Motion, nor submit any supporting documentation, nor
provide any analysis as to how such a finding would put the errors
outside the scope of the QC system. We also note that, although the
State acknowledged during the hearing that its arguments concerning
unavoidable errors in eligibility (or payment amount) determinations
would not apply to certain other types of errors (Tr., pp. 21-22), the
Agency never raised at that time an issue that the errors here might be
one of those types of errors. Indeed, the State claimed, and the Agency
did not deny, that the State had instituted a program, similar to QC,
for reducing errors in the AFDC-FC program such as those covered by the
audit here.
Thus, the Agency in its Motion not only misconstrues the Board's
conclusion distinguishing certain systemic errors, but fails to
convincingly support its own allegations.
Conclusion
Having reconsidered Decision No. 319, we conclude that we reached the
correct result, for the reasons stated above. This, of course, in no
way implies that sampling is not a valid technique or that we would
reject use of that technique in other cases where its use would be
appropriate. /1/ There were three additional sample cases which the
State initially contested, on the basis that the period for
retention of records associated with those cases had expired and,
therefore, it was unfair to disallow for those cases. The Agency
responded that, since those payments were not claimed for federal
financial participation (FFP) until a period within several years of the
audit, they were within the record retention requirements of 45 CFR
74.21 and 74.22. The State did not reply to this argument, nor continue
to claim that these sample cases were allowable. Accordingly, we uphold
the Agency finding that these cases were ineligible. n2 At the hearing,
the Agency submitted a computer printout showing that it had calculated
the amount for ineligible children for the universe of claims as the
lower limit of "90 percent confidence limits." The State contested
use
of this figure, but we do not address this argument in view of our
holding below. /3/ In a post-conference submission, the State
objected to some of the Agency's proposed clarifications of the
transcript. Since the clarifications are not relevant to our decision
here, we do not find it necessary to rule on the objections. /4/
The Social and Rehabilitation Service (SRS) was then responsible for
administering the AFDC program. After a reorganization in 1977, the
Social Security Administration (SSA) took over the program. In 1981,
the Office of Human Development Services became responsible for
administering the AFDC-FC component only. For purposes of this
decision, when we refer to "Agency" we mean whichever Departmental
component was then responsible for administering the program.
/5/ The child Welfare and Adoption Assistance Act of 1980, Public Law
96-272, June 17, 1980, established a new Foster Care Program as Title
IV-E of the Act. /6/ See, also, Appeal Brief, p. 16, quoting
from remarks at the conference in the cases decided by the Board in
Decision No. 170. These remarks were mistakenly attributed to the
Presiding Board Member in those cases but were, in fact, made by an
Agency representative. /7/ Section 1102 provides that the
Secretary "shall make a publish such rules and regulations, not
inconsistent with (the Act), as may be necessary to the efficient
administration of the functions with which (he) is charged" under the
Act.
OCTOBER 22, 1983