Department of Health and Human Services
DEPARTMENTAL APPEALS BOARD
Appellate Division
SUBJECT:
New Mexico Children, Youth and
Families Department
Docket No. A-07-12
Decision No. 2100
DATE:
July 19, 2007
DECISION
The New Mexico Children, Youth and Families Department (CYFD)
appealed a determination by the Administration for Children and
Families (ACF) disallowing $552,374 in federal financial
participation (FFP) claimed by New Mexico under section 1130 of
the Social Security Act (Act) for the “assisted guardianship”
component of a Child Welfare Waiver Demonstration Project.1 This
component allowed CYFD to use funds available under title IV-E of
the Act to subsidize the cost of guardianships to facilitate
permanency for children who might otherwise remain in foster
care. Section 1130(g) requires that the total amount of federal
funds expended by a state for a demonstration project not exceed
the amount that would have been expended in the absence of the
project. ACF disallowed the amount by which it found CYFD
exceeded this “cost neutrality limit” over the five-and-one-half
year project period. CYFD maintains that the formula used by ACF
to calculate the cost neutrality limit does not make sense as
applied to its project data. According to CYFD, the project was
cost neutral since the payments made for children in assisted
guardianships did not exceed the amount that would have been paid
if these children had remained in foster care and continued to
receive maintenance payments funded by title IV-E. As an
additional ground for reversing the disallowance, CYFD asserts
1
The current version of the Social Security Act can be
found at www.ssa.gov/OP_Home/ssact/comp-ssa.htm. Each section of
the Act on that website contains a reference to the corresponding
United States Code chapter and section. Also, a cross reference
table for the Act and the United States Code can be found at 42
U.S.C.A. Ch. 7, Disp Table.
2
that the project ultimately saved the federal government money
because, had the children not been placed in assisted
guardianships, they would have continued to receive IV-E payments
instead of CYFD assuming all responsibility for the assisted
guardianship payments when the project ended.
As discussed in detail below, we conclude that ACF properly
required CYFD to calculate the cost neutrality limit using the
formula in the waiver terms and conditions. We further conclude
that any cost savings to the federal government after the
demonstration project ended were irrelevant. Accordingly, we
uphold the disallowance in full.
General background on section 1130 demonstration projects
Title IV-E was originally enacted as part of the Adoption
Assistance and Child Welfare Act of 1980, Public Law No. 96-262.
The primary purpose of title IV-E is to assist states with foster
care maintenance payments and adoption assistance payments for
eligible children. See section 474(a)(1) and (a)(2) of the
Social Security Act (Act), 42 U.S.C. § 674(a)(1) and (a)(2). In
1994, Public Law 103-432 added a provision to the Act authorizing
the Secretary “to conduct demonstration projects . . . which the
Secretary finds are likely to promote the objectives of part B or
E of title IV.”2 Section 1130 of the Act, 42 U.S.C. § 11320a-9.
Section 1130(d) provides that a demonstration project may be
conducted for not more than five years unless the Secretary
allows the project to continue beyond that period. Section
1130(e) provides that an application for a demonstration project
must include, among other things:
a description of the proposed project, the geographic
area in which the proposed project would be conducted,
the children or families who would be served by the
proposed project, and the services which would be
provided by the proposed project (which shall provide,
where appropriate, for random assignment of children and
families to groups served under the project and to
control groups)[.]
2
Section 1130 originally provided that the Secretary
“may authorize not more than 10 States to conduct demonstration
projects[.]” The Adoption and Safe Families Act of 1997, Public
Law No. 105-89, amended section 1130 to authorize the Secretary
to approve up to ten new demonstration projects each year from
1998 through 2002 (extended by subsequent legislation to 2003).
3
Section 1130(f)(1) provides that each state authorized to conduct
a demonstration project shall–
obtain an evaluation by an independent contractor of the
effectiveness of the project, using an evaluation design
approved by the Secretary which provides for–
(A) comparison of methods of service delivery under
the project, and such methods under a State plan or
plans, with respect to efficiency, economy, and any
other appropriate measures of program management;
(B) comparison of outcomes for children and
families (and groups of children and families) under
the project, and such outcomes under a State plan or
plans, for purposes of assessing the effectiveness of
the project in achieving program goals; and
(C) any other information that the Secretary may
require[.]
Finally, section 1130(g) provides:
COST NEUTRALITY.-The Secretary may not authorize a
State to conduct a demonstrate project under this
section unless the Secretary determines that the total
amount of Federal funds that will be expended under (or
by reason of) the project over its approved term (or
such portion thereof or other period as the Secretary
may find appropriate) will not exceed the amount of such
funds that would be expended by the State under the
State plans approved under parts B and E of title IV if
the project were not conducted.3
In 1995, ACF published a notice announcing that it was seeking
proposals for demonstration projects under section 1130, stating
generally: “The Department desires to facilitate the testing of
new approaches to the delivery of a broad range of child welfare
services. Such demonstrations can provide valuable knowledge
that will help lead to improvements in the delivery,
effectiveness and efficiency of services.” 60 Fed. Reg. 31,478
3
Title IV-B provides funding for a broad range of
social services to families and may also be used for the same
types of costs funded under title IV-E. See sections 423 and 425
of the Act. However, title IV-B, unlike title IV-E, has a
funding cap. Section 421 of the Act. ACF does not allege that
CYFD failed to take any IV-B services for IV-E-eligible children
into account in determining cost neutrality.
4
(June 15, 1995) (ACF Ex. 2). In a section captioned
“Evaluation,” the notice reiterates the statutory requirement
that the evaluation plan must provide for a comparison of methods
of service delivery and outcomes under the project and under an
existing state plan, and continues:
The Department encourages, where appropriate, that the
proposal provide for random assignment of children and
families to groups served under the project and control
groups, but is open to various other evaluation
techniques. For example, in a proposed demonstration
effort that would necessarily affect 100% of the
population to be served, a random assignment methodology
would not be appropriate.
* * * * *
If the State proposes an alternative to random
assignment, the proposal must include a justification
explaining why random assignment is not appropriate and
how the alternative methodology will meet evaluation
needs.
60 Fed. Reg. 31,478, 31,480.
The notice also addresses the topic of “Cost Neutrality” in part
as follows:
Section 1130(g) requires that the waiver demonstration
project be cost-neutral, that is, the total amount of
federal funds used to support the demonstration project,
over the approved project period, will not exceed the
amount of federal funds that would have been expended by
the State under the State plans approved under Parts B
and E of title IV if the waiver demonstration project
were not conducted. . . .
* * * * *
The Department encourages, where appropriate, the use of
random assignment of individuals for evaluation and as a
method for determining the fiscal effects of the
demonstration project but recognizes that this method
may not be appropriate for certain demonstration
projects. In randomly assigning individuals to
experimental and control groups, the costs associated
with the control group (foster care rates and
administrative costs) become the baseline for cost
neutrality (i.e., the average cost for a control group
case is assumed to be the amount that would have been
spent on each experimental case). If an alternative
5
method is proposed, then other methods of measuring cost
neutrality should be used.
* * * * *
The Department recognizes the difficulty of projecting
and measuring title IV-E and title IV-B expenditures,
and is open to methodology(ies) the State(s) may
propose. However, the Department favors random
assignment, where appropriate, as the methodology for
the evaluation component, and as a method for
determining the fiscal effects of a demonstration as
well. . . .
60 Fed. Reg. 31,478, 31,480-81.4
In 1998, ACF issued an information memorandum to the states that
made essentially the same points as the 1995 Federal Register
notice regarding the design and evaluation of the projects.
ACYF-CB-IM-98-01, dated 2/13/98 (CYFD Ex. 6), at 11-13. The
information memorandum also stated that “[t]he Department will
devise a cost neutrality formula, for quarterly (typically)
payments to the States, that will calculate an amount the State
would otherwise have received for that period for the children in
the demonstration, in the absence of a demonstration.” Id. at
13.
In 1999, ACF informed states that they were to immediately begin
using Form ACF-IV-E-1 for reporting expenditures for IV-E foster
care and adoption assistance. ACYF-OC-PI-99-01, dated 9/22/99
(ACF Ex. 6), at 1. Part 4 of the form is specifically designed
for reporting expenditures for demonstration projects. Id. at 8.
CYFD’s demonstration project
In February 1999, CYFD submitted its proposal for a demonstration
project under section 1130 of the Act. ACF Ex. 5. The waiver
4
The notice also states that the cost of the
evaluation of a state’s demonstration project will be excluded
from the cost-neutrality calculation. 60 Fed. Reg. 31,478,
31,480. It further states that the costs of the development of
the proposal, the evaluation design and the evaluation itself may
be allocated entirely to title IV-E as administrative costs. Id.
Thus, even if the project did not exceed the cost neutrality
limit, the federal government would bear extra costs.
6
terms and conditions for New Mexico’s demonstration project were
approved by ACF in June 1999 and accepted by CYFD in July 1999.
CYFD Ex. 3. The terms and conditions provided for “a threecomponent demonstration,” of which only the third component is at
issue.
The third component is described as follows:
Under the third component of the demonstration, the
State will be able to use title IV-E funds to establish
assisted guardianships, in amounts not greater than
amounts provided under subsidized adoption for children
in foster care under State custody, and for such
children for whom reunification with the biological
family or adoption are not options.
CYFD Ex. 3, at 3 (section 2.0).5 The terms and conditions
further describe the third component as follows:
Under the State-custody guardianship component, the
State may establish an assisted guardianship program for
children who meet the minimum criteria stated below.
This will be a statewide guardianship program, the
evaluation of which will be based on a random assignment
methodology.
Prior to implementation of [this component], the State
must submit to the Department, for review and approval,
a plan specifying the criteria for determining
eligibility . . . . At a minimum, the selection
criteria will provide that no child may participate in
the guardianship component unless it has been
determined: that the child cannot be reunited with her
or his parent(s); and that adoption is not possible
. . . or is not in the best interests of the child. . ..
CYFD Ex. 3, at 5 (section 2.1.c.).
5
The second component of the demonstration project was
an assisted guardianship component for Native American children
in tribal custody. ACF ultimately agreed that any eligible child
in tribal custody could be placed in an assisted guardianship and
did not require that component to use the cost neutrality formula
at issue in the case of the third component. CYFD Ex. 8, at 1;
CYFD Br. at 10. According to CYFD, tribes objected to the
initial proposal for a control group that would not be eligible
for assisted guardianship. Id.
7
A section captioned “Evaluation” states in pertinent part that
“[f]or the State guardianship component, eligible children will
be randomly assigned to a control and experimental group and
compared on process, outcomes, and costs.” Id. at 8 (section
3.0). The section also states as follows:
The sampling plan will also describe the procedures by
which children in the State-custody subsidized
guardianship component (component No. 3) will be
screened for eligibility for this demonstration, and
randomly assigned to a control group or experimental
group at the point of eligibility determination.
Id. at 10 (section 3.3). In addition, the terms and conditions
contain a list of “outcomes” that will be measured “for both the
experimental and comparison groups” in the State guardianship
component. Id. at 12 (section 3.5). The terms and conditions
further state that, as part of the evaluation, “[c]osts for the
experimental group will be compared to costs of the control group
for foster care payments, adoption subsidies, services, case
worker time, and other factors.” Id. at 14 (section 3.7).
With respect to cost neutrality, the terms and conditions require
generally that–
[p]rojects conducted under this waiver authority shall
. . . [b]e cost neutral to the Federal government over
the life of the project period. That is, the cost of
the demonstration project may not exceed the total
amount of Federal funds that would have been expended by
the State under the State plans approved under Parts B
and E of title IV if the waiver demonstration had not
been conducted . . . .
Id. at 2 (section 1.7). The terms and conditions also state in a
separate section captioned “Cost Neutrality”:
Except for costs of evaluating and developing this
project . . . , beginning with the deemed beginning date
the operation of this demonstration is to be costneutral to the Federal government with respect to
benefit and administrative costs for titles IV-B and IVE of the Social Security Act.
Id. at 16 (section 4.2). In addition, the terms and conditions
contain the following instructions under the subheading
“Claiming”:
8
The amount of title IV-E funds a State may claim under
both the guardianship components of this demonstration
will be based on an analysis of the costs of cases
within the respective control groups and will be
determined by application of the cost neutrality formula
in this Section. The calculation below will be
performed for maintenance or adoption assistance
payments and administrative payments on a quarterly
basis.
Step 1. Calculate the average title IV-E cost per
control group case by dividing the cumulative title IV-E
costs for control group by the number of control cases.
Step 2. Multiply the average derived in step (1) above
by the number of experimental cases. The result is the
cumulative cost for the experimental cases.
Step 3. The cumulative demonstration cost will be equal
to the cumulative allowable title IV-E maintenance
payments and administrative costs for the control cases
plus the cumulative cost for the experimental cases as
calculated above.
Id. (italics in original).
CYFD contracted with TriWest Group, LLC (TriWest) to conduct the
evaluation of its demonstration project. CYFD Ex. 5, at 5.
TriWest’s May 31, 2000 Final Evaluation Plan indicates that the
purpose of the project is to assess “the impact of making
assisted guardianships available to children for whom
reunification or adoption are not options[.]” CYFC Ex. 5, at 2;
see also id. at 3 (stating that the “major objectives of the
project include . . . [d]emonstration of improved outcomes in the
areas of safety, permanency, and well-being for . . . non Native
American children in State custody through implementation of
assisted guardianships.”). The plan indicates that children
eligible to participate in the State guardianship component will
be randomly assigned to either an experimental or control group.
See CYFD Ex. 5, at 2, 7 (emphasis added). The plan also states
that for this component, “the population will be Title IV-E
eligible non-Native American children in State custody who are
referred for care during the first three years of the project
(SFY 2001 - 2003) and meet the criteria described in General
Conditions for Components Involving Assisted Guardianship,
above.” Id. at 6. Those criteria were “that no child may
participate in a guardianship demonstration unless it has been
determined that the child cannot be reunited with [his/her]
9
parent(s) and that adoption is not possible . . . or is not in
the best interests of the child, and the child has been
adjudicated as required by New Mexico Code . . . .” Id. at 5.
The Interim Status Report for December 31, 2000 does not track
the project design set out in the Final Evaluation Plan, however.
Instead of providing that all children meeting the criteria
described above will be randomly assigned to either an
experimental or control group, the report indicates that all
children entering the child welfare system will be randomly
assigned to an experimental or control group. CYFD Ex. 12, at
22-25 (emphasis added).
By the end of the project, there were 3,392 children assigned to
the experimental group and 3,434 children assigned to the control
group.6 The assignments were made based on whether the
identification number the child was given upon entering the child
welfare system was odd or even, except in the case of siblings,
who were assigned to the same group. See, e.g., CYFD Ex. 15, at
9.7 The number of children in each group increased over the
duration of the project as new children entered the child welfare
system.
Basis for the disallowance
While the project was ongoing, CYFD failed to properly complete
the part of the quarterly claiming forms used to determine
whether the project was meeting the cost neutrality limit. ACF
Exs. 21, 23, and 24. Following the end of the project on
December 31, 2005, ACF gave CYFD additional time to complete the
cost neutrality calculations. See CYFD Ex. 1, at 1. Using the
formula in the waiver terms and conditions, TriWest determined
that the cost neutrality limit was exceeded by $3,373,163.11.
6
The final evaluation report shows that there were
6,291 children assigned to the experimental group, including 234
children in assisted guardianships, and 6,199 children assigned
to the control group. CYFD Ex. 21, at 31, 35. TriWest later
adjusted these numbers after determining that it had failed to
limit the pool of children to those who were eligible for
services funded with IV-E and less than 18 years of age. CYFD
Ex. 25, at 4.
7
Several children who were initially randomly assigned
to the control group were reassigned to the experimental group
and placed in assisted guardianships under an “exemption” for a
“hardship situation.” CYFD Ex. 21, at 31.
10
CYFD Ex. 25, at 3. CYFD then hired a statistical consultant to
determine why this figure exceeded the entire cost of the
assisted guardianship program (identified in the statistician’s
affidavit as $2.5 million, including both the federal and nonfederal share). CYFD Ex. 25, at 1; CYFD Ex. 10 (Affidavit of
Mary Ann Shaening, Ph.D.). According to Ms. Shaening, since only
198 children in the experimental group were placed in assisted
guardianships, the formula in the waiver terms and conditions
“inappropriately attributes all cost differences between two
large groups to the impact of a much smaller group (of only 198
children) who actually were subjected to the experimental
condition of receiving assisted guardianships,” and therefore
“does not yield accurate or meaningful results.” CYFD Ex. 10, at
1. Ms. Shaening also identified as a secondary reason for these
results the fact that the average length of time that a child was
in custody, i.e., “average days in episode,” varied significantly
between the experimental group and the control group after
excluding the assisted guardianship children as a factor. Id. at
2. Ms. Shaening determined that had the days in episode been
comparable for the two groups, the costs of the control group
would have increased, reducing the amount by which the cost
neutrality limit was exceeded to $552,374. Id. Ms. Shaening
stated that this illustrated that, even adjusting for the
difference in days in episode, the experimental group and the
control group “were not comparable.” Id. Ms. Shaening then
suggested that “[t]he most direct way of determining cost
neutrality of the waiver project . . . would be to compare the
actual Federal costs of the 198 assisted guardianships with what
these 198 children would have received in Title IV-E payments had
they remained in foster care.” Id. By letter dated May 31,
2006, CYFD presented the results of Ms. Shaening’s analysis to
ACF, noting that when the “the cost neutrality formula is applied
to the demonstration [experimental] group and the control group
(with the adjustment) . . . the cost of the demonstration group
exceeded the cost neutrality limit by approximately 1.6 percent.”
CYFD Ex. 25, at 1. The letter continued:
This method for calculating cost neutrality is not
ideal, as you know. The waiver terms and conditions’
selection criteria and the original evaluation plan
provide that the comparison groups should have been
based upon eligible children (children who cannot return
home and who are not likely to be adopted).
Unfortunately, this criteria was not readily available
in the data base of the evaluator. With additional
time, we could possibl[y] construct a comparison group
with similar demographics to those of the subsidized
guardianship group that would allow us to more
11
accurately compute cost neutrality. Additionally, the
administrative costs associated with the two groups are
not included in the current calculations. . . .
Id. at 2. Although ACF granted CYFD an additional extension
until August 31, 2006 to finalize the cost neutrality
calculations, there is no evidence of additional calculations in
the record. By letter dated September 25, 2006, ACF stated that
it “concurs with CYFD’s calculation that the total submitted
costs for the experimental group exceeded the calculated cost
neutrality limit by $552,374” FFP and that it was disallowing
this amount, i.e., the amount calculated by Ms. Shaening using
ACF’s formula after adjusting for the difference in days in
episode. CYFD Ex. 1, at 1.
Discussion
CYFD acknowledges that its project was subject to a cost
neutrality limit but disputes ACF’s determination that it
exceeded that limit. According to CYFD, its project was cost
neutral because the payments made for children in assisted
guardianships did not exceed the amount that would have been paid
if these children had remained in foster care and continued to
receive maintenance payments funded by title IV-E. As explained
below, we conclude that CYFD was required by section 1130 and the
waiver terms and conditions to use a different comparison to
determine the cost neutrality limit, that in any event CYFD’s
alternative methodology does not demonstrate cost neutrality, and
that any cost savings to the federal government after the project
ended may not be used to offset the excess costs incurred during
the term of the project for purposes of the cost neutrality
calculation.
Under section 1130(g), a project is cost neutral where “the total
amount of Federal funds that will be expended under
. . . the project . . . will not exceed the amount of such funds
that would be expended by the State under the State plans
approved under parts B and E of title IV if the project were not
conducted.” Although this language does not specify how the
latter amount is to be determined, that becomes apparent when
section 1130(g) is read in light of the requirement in section
1130(e) that a project “provide, where appropriate, for random
assignment of children and families to groups served under the
project and to control groups[.]” ACF’s 1995 Federal Register
notice, which constitutes ACF’s official interpretation of
section 1130, explains that such groups are to be used to
determine cost neutrality by making “the costs associated with
the control group (foster care rates and administrative costs)
12
. . . the baseline for cost neutrality (i.e., the average cost
for a control group case is assumed to be the amount that would
have been spent on each experimental case).” 60 Fed. Reg.
31,478, 31,480. In addition, the waiver terms and conditions
state that “eligible children will be randomly assigned to a
control and experimental group and compared on process, outcomes
and costs.”
The waiver terms and conditions also specify that
“[c]osts for the experimental group will be compared to costs of
the control group . . . .” CYFD Ex. 3, at 10. Finally, the
waiver terms and conditions set out a formula (quoted earlier)
for determining cost neutrality using this comparison. Thus,
CYFD’s comparison of the costs of the subsidies for children in
assisted guardianships with the IV-E foster care maintenance
payments that hypothetically would have been made on behalf of
the same children had they remained in foster care was not the
comparison required by section 1130 and the waiver terms and
conditions.
CYFD takes the position that it was not required to use the
formula in the waiver terms and conditions to calculate the cost
neutrality limit. Instead, CYFD maintains, the provision in the
terms and conditions that payments for an assisted guardianship
will not exceed the amount that would have been paid for
subsidized adoption for children in foster care in state custody
is the applicable cost neutrality provision. See CYFD Br. at 89, CYFD Ex. 3, at 3. However, this provision appears in the
“Implementation” section of the waiver terms and conditions with
no indication that it is intended as a test for cost neutrality.
Although CYFD might have believed that limiting the payment
amounts for assisted guardianships in this manner would ensure
cost neutrality, the cost neutrality formula was clearly intended
as the means for determining whether the project was in fact cost
neutral.
CYFD also takes the position that “cost neutrality to the Federal
government was . . . guaranteed” by the provision on the payment
amount for assisted guardianships. See CYFD Ex. 9 (Affidavit of
Maryellen Bearzi, Deputy Director, Protective Services, CYFD,
dated 13/13/07). We conclude, however, that comparing the cost
of the guardianship subsidies with the amount that would have
been spent for foster care payments for the same children is not
a reasonable basis for determining whether the demonstration
project was cost neutral.8 CYFD assumes that all of the children
8
According to CYFD, “[g]uardianship payments were
determined in exactly the same manner as adoption payments which,
(continued...)
13
who were placed in assisted guardianships would otherwise have
remained in foster care. According to CYFD, such an assumption
is reasonable because the criteria for placing children in
assisted guardianships included that the children be “in foster
care under State custody” and “children for whom reunification
with the biological family or adoption are not options.” CYFD
Br. at 2, quoting CYFD Ex. 3, at 3 (waiver terms and conditions).
However, as CYFD itself points out, whether reunification or
adoption is possible is a judgment a court makes at a particular
point in time. See CYFD Br. at 11. Thus, a child who was placed
in an assisted guardianship might have later been reunified or
adopted had he or she remained in foster care.
Even if CYFD could establish that the children in assisted
guardianships would otherwise have remained in foster care, we
are not persuaded that the amount of the guardianship payment for
each child was the same as the amount of foster care maintenance
payments that would have been paid if the child had remained in
foster care. According to CYFD, there was nothing “about
guardianship placements per se that would cause a different level
of subsidy for a particular child.” CYFD Reply Br. at 11.
However, CYFD provided no assurance that the various factors that
affect the amount of monthly foster care maintenance payments for
a child would not result in different payment amounts, on
average, for children placed in assisted guardianships than for
children in foster care. The rates actually paid per child
include non-standard allowances to meet the individual needs of
children. See CYFD Reply Br., Appendix (App.) 1, at 3-7. Some
of these items depend on whether the parents provided certain
items to the child and sought reimbursement for them, such as
vacations, extracurricular activities, mentoring, and holiday
gifts. Id. CYFD has not shown any basis for assuming that
guardians in assisted guardianships received these allowances
with the same frequency as foster parents of children eligible
for but not placed in assisted guardianships. Moreover, the
total amount of foster care maintenance payments reflects the
length of time that a child remains in foster care. As indicated
above, it is simply speculative to assume that a child would have
8
(...continued)
under [section 473(a)(3) of the Act]
‘. . . in no case may . . . exceed the foster care maintenance
payment which would have been paid during the period if the child
with respect to whom the adoption assistance payment [or in this
case, guardianship payment] is made had been in a foster family
home.’” CYFD Reply Br. at 9.
14
remained in foster care if he or she had not been placed in an
assisted guardianship.
Furthermore, CYFD’s comparison does not account for additional
administrative costs eligible for IV-E funding that might have
been incurred as a result of the project. The waiver terms and
conditions expressly provide that the project “is to be costneutral to the Federal government with respect to benefit and
administrative costs[.]” CYFD Ex. 3, at 16 (emphasis added).
CYFD nevertheless asserts that “there were no separate
administrative costs to the waiver. . . . The same social workers
did the same work . . . as they have always done . . . .” CYFD
Reply Br. at 13, n.1. This argument has no merit. Among the
questions to be evaluated in the demonstration project was “[t]he
differences between the experimental and control groups with
regard to comparable resources, services, activities, staffing,
etc.” CYFD Ex. 3, at 14; see also id. at 9. CYFD cannot simply
assume the answer based on the fact that the same social workers
and administrators provided services to children in both of these
settings. Indeed, a synthesis of findings from the seven stateassisted guardianship demonstration projects notes that
caseworkers raised “objections to the additional time, effort and
paperwork required to process guardianships.” ACF Ex. 30, at 2.
Moreover, additional social work services might have been
provided to comply with the requirement in the waiver terms and
conditions for delivery of “post-placement services of the type
generally provided by the State for adoptive placements” (CYFD
Ex. 3, at 6) and the requirement that guardianship homes be
licensed or approved or have a homestudy “as would be done for an
adoptive placement” (CYFD Ex. 3, at 5).9
As CYFD points out, the methodology on which the disallowance
amount was based is also problematic. CYFD asserts, and ACF does
not dispute, that strict application of the formula in the waiver
terms and conditions yields an amount of $3,373,163.11 in excess
of the cost neutrality limit, more than the $2.5 million total
project costs (of which $1.8 was the federal share). This
anomolous result, however, is clearly attributable to CYFD’s
failure to use the project design required by the waiver terms
and conditions. The waiver terms and conditions provide for
screening all children in foster care in state custody to
identify those children who meet certain criteria for eligibility
9
While foster care homes must also be licensed or
approved, more than one child could be placed in the same foster
home (at the same time or in succession) without the home going
through the licensing or approval process each time.
15
for assisted guardianships. The waiver terms and conditions
further provide that children meeting these criteria are to be
randomly assigned to either the experimental group or the control
group. This contemplates that both the experimental and control
groups be drawn from children who are eligible for assisted
guardianships. Although the Final Evaluation Plan prepared by
TriWest describes such a procedure, TriWest’s subsequent status
reports indicate, and CYFD acknowledges, that children were not
screened prior to assignment to the experimental or control
group. See, e.g., CYFD Ex. 12, at 22-25; CYFD Reply Br. at 5.
Moreover, only 198 of the 3,392 children in the experimental
group were placed in assisted guardianships. Following the
formula in the waiver terms and conditions, TriWest calculated
the average cost (federal share) for a child in the control group
of 3,434 children, and multiplied the resulting figure by 3,392
to determine the cost neutrality limit. The $3,373,163.11 by
which the federal share of the cumulative costs of the
experimental group exceeded this limit was plainly not
attributable solely to the very small number of children who were
in assisted guardianships. CYFD can hardly complain that it is
disadvantaged by the use of this formula, however, since it did
not set up the experimental and control groups in accordance with
the waiver terms and conditions and did not offer any valid
alternative method of determining cost neutrality.
Moreover, ACF did not disallow the full amount in excess of the
cost neutrality limit calculated using this formula. Instead,
ACF recognized the downward adjustment in the control group costs
calculated by CYFD’s statistician based on the difference in
“days in episode” between the experimental group and the control
group, which reduced the excess costs under the formula to
$552,374. CYFD nevertheless challenges ACF’s reliance on its
adjusted figure as the basis for the disallowance on the ground
that the adjustment was intended only to demonstrate that
application of the formula to CYFD’s data was not scientifically
sound. CYFD clearly benefitted from the use of this figure,
however, since ACF might have been justified in disallowing all
of the project costs in light of CYFD’s failure to carry out the
project in accordance with the requirement in the waiver terms
and conditions that “eligible children . . . be randomly assigned
to a control and experimental group and compared on process,
outcomes, and costs.” CYFD Ex. 3, at 8. This requirement is
related to the underlying purpose of the section 1130
demonstration projects, stated in ACF’s 1995 Federal Register
notice, to test new approaches to the delivery of child welfare
services that will lead to improvements in the delivery of such
services. This purpose was thwarted by CYFD’s failure to have a
control group comparable to the group of children who were placed
16
in assisted guardianships. In the absence of such a control
group, CYFD could not test whether the use of assisted
guardianships was a more effective way of achieving permanence
for children than leaving them in the foster care system, and at
what cost.10 Indeed, ACF warned CYFD of the possibility that all
project costs might be disallowed if CYFD failed to correct the
financial reports to calculate cost neutrality using data from
experimental and control groups. CYFD Ex. 22, at 4.
CYFD asserts that there were valid reasons for its failure to
follow the original project design. According to CYFD,
[t]hese included the burden to already overworked field
staff who would have been asked to establish and
document eligibility for assisted guardianship . . . and
the concern that CYFD would be vulnerable to criticism
and even legal action because a permanency option was
denied to eligible youth based on the random assignment
to a group . . . .
Notice of appeal dated 10/25/06, at 2. CYFD further states that
disqualifying otherwise eligible children from permanent
guardianships “was inconsistent with State law on permanency
planning which requires that the permanency plan for a child be
determined solely by his/her best interests” and that “prescreening” children for eligibility for guardianship “was an
undertaking for which CYFD found it did not have the resources.”
CYFD Reply Br. at 5. Although the waiver terms and conditions
required CYFD to “apprise the Department of any difficulty in
achieving the estimated sample sizes for the project” (CYFD Ex.
3, at 10), CYFD does not assert that it requested, much less
received, ACF’s approval to change the project design for this
component based on any of these considerations (which may even
have been apparent when CYFD was developing its project
proposal).11 Instead, CYFD appears to have simply ignored its
10
TriWest’s final report did compare children in
assisted guardianships to the subset of adopted children in the
purported control group. CYFD Ex. 21 (Evaluation Final Report)
at 35-37. However, it is not clear how these comparisons relate
to the project purpose of assessing the impact of assisted
guardianships on children for whom reunification or adoption are
not options.
11
As indicated previously, CYFD obtained ACF’s
approval to change the project design for the tribal guardianship
(continued...)
17
commitment under the waiver terms and conditions to include only
children eligible for assisted guardianships in both the
experimental and control groups.
CYFD also points out that neither it nor TriWest hid from ACF the
fact that they were not using the formula in the waiver terms and
conditions to ensure that the project remained cost-neutral as it
progressed. This is not an appropriate basis for reversing the
disallowance, however. The Board has repeatedly acknowledged the
prevailing view in the federal courts that equitable estoppel
does not lie against the federal government, if indeed it is
available at all, absent at least a showing of affirmative
misconduct. See, e.g., South Carolina Department of Social
Services, DAB No. 1998 (2005); Northstar Youth Services, Inc.,
DAB No. 1884 (2003), and cases cited therein (including Office of
Personnel Management v. Richmond, 496 U.S. 414 (1990) and Heckler
v. Community Health Services of Crawford County, Inc., 467 U.S.
51 (1984)). ACF’s acceptance of claim forms and reports from
which it might have discerned that the cost neutrality formula
was not being used does not constitute affirmative misconduct.
Moreover, a government agent cannot obligate the government to
pay funds in violation of statutory authority, such as project
costs in excess of the section 1130 cost neutrality limit. See,
e.g., Colorado Dept. of Social Services, DAB No. 1369, at 5
(1992), citing Office of Personnel Management v. Richmond.
Finally, CYFD maintains that its demonstration project saved the
federal government money by shifting the costs of the care of the
198 children who were placed in assisted guardianships from title
IV-E to the State after the project ended. CYFD estimates that,
had those children remained in foster care, an additional
$5,837,793 in title IV-E funds would have been expended on their
behalf from the time the project ended until they reached age 18.
CYFD Ex. 10, at 3. These alleged cost savings, according to
CYFD, would have more than offset even the unadjusted amount in
excess of the cost neutrality limit calculated by TriWest using
the formula in the waiver terms and conditions. CYFD argues that
such cost savings are “relevant to application of the definition
of cost neutrality” in section 1130(g), which it quotes in part
as follows: “the total amount of Federal funds that will be
expended under (or by reason of) the project over its approved
term (or such portion thereof or other period as the Secretary
may find appropriate) . . . .” CYFD Reply Br. at 8 (emphasis
added by CYFD).
11
(...continued)
component of its project based on similar considerations.
18
This argument has no merit. The alleged cost savings result from
CYFD’s commitment in the waiver terms and conditions to ensure
that children placed in assisted guardianships as part of the
demonstration project continue to receive support in their
placements after IV-E funding ends. The shifting of support for
these children to State-only sources is irrelevant to the purpose
of assessing the cost neutrality of the project–which is to
evaluate whether alternative approaches to child welfare can
improve service delivery and outcomes at the same or lower cost.
The language quoted from section 1130(g) at best could be read to
give the Secretary discretion to consider the impact on federal
funds of costs incurred by reason of or under the demonstration
project during some period other than the full project period.
Here, the Secretary did not elect to exercise any such
discretion. Instead, an explicit agreement between ACF and CYFD
in the waiver terms and conditions set the term of the project as
the period over which cost neutrality was to be measured. Since
the terms and conditions also included the plan for State funding
of assisted guardianships post-waiver, it is clear that the terms
and conditions rejected any option of including post-waiver costs
in the cost neutrality evaluation.
Conclusion
Based on the foregoing analysis, we uphold the disallowance in
full.
/s/
Judith A. Ballard
/s/
Constance B. Tobias
/s/
Leslie A. Sussan
Presiding Board Member
