GAB Decision 522
Docket No. 83-16
March 12, 1984
New York State Department of Social Services;
Ford, Cecilia; Teitz, Alexander Settle, Norval
Introduction.
This dispute concerns how to calculate the correct "adjusted payment
level" (APL) for benefits paid by New York to recipients of aid to the
aged, blind, and disabled (AABD) in January 1972. The APL calculation
is necessary to implement a "hold harmless" provision enacted by
Congress in 1972 as part of the supplemental security income program
(SSI).
SSI is a program of federal benefits to the needy aged, blind, and
disabled. It replaced the state-administered AABD programs beginning in
January 1974. States could supplement the basic federal SSI benefit and
could elect to have their state supplements administered by the Social
Security Administration (SSA), which also administers the basic federal
SSI benefits. Congress wished to encourage states to adopt supplement
programs and to elect federal administration of the supplements, but
also recognized that liberalized federal eligibility criteria under SSI
meant expanded caseloads. Thus, Congress devised a "hold harmless"
provision, authorizing federal funding for part of the cost of
federally-administered state supplements, to be determined by using a
statutory formula. Pub. L. 92-603, section 401(b). The formula was
designed to protect states from any increase in their costs due to
expanded caseloads, but not from any increase due to higher benefits.
The APL is part of the "hold harmless" formula; it is an arithmetic
construct representing the average money payment made by a state in
January 1972 to AABD recipients who had "no other income," as adjusted
by a $10 "food stamp bonus." In other words, the APL is a measure
of
what a state was paying to its neediest recipients under AABD and limits
what part of any state supplement is protected under "hold harmless."
Preliminary attempts at calculating an APL for New York resulted in a
provisional APL. At SSA's direction, the State undertook a review of
100 percent of its January 1972 AABD caseload to determine a final APL.
(2) The State reviewed over 248,000 AABD cases; some could be
screened by a computer, but 133,470 cases required manual review of the
recipient's case file. The process was complicated by the need to
establish what AABD payments should be included in the APL calculation
(i.e., were money payments to individuals with no income other than AABD
benefits), and what payments should be excluded from the APL
calculation. State and SSA officials negotiated from 1974 to 1978 over
criteria to be used for the review and over other matters related to
data collection, such as how to treat cases where files were missing or
data was insufficient to apply the criteria.
After performing its 100 percent review, the State hired an
accounting firm (Arthur Young & Company) to do a quality assurance
study. Arthur Young reviewed a sample of 4,426 cases, concluding that
the State's procedures were adequate. The State then submitted to SSA a
report of its review, stating that the final APL figure should be
$201.25.
Subsequently, the HHS Audit Agency performed an audit to verify the
State's submitted APL. The auditors used a sample drawn from the Arthur
Young sample to come up with their own APL figure. SSA adjusted the
auditors' findings in five cases but accepted them in all other
respects, ultimately establishing an APL of $191.56. Although this
figure is only $9.69 less than the State's figure, use of the lower rate
would mean a difference of many millions of dollars to New York. /1/
The State appealed the SSA determination to the Board under the
disputes clause of its agreement with SSA and 45 CFR Part 16. The State
challenged SSA's determinations on 28 of the individual cases in the
auditors' sample of excluded cases. The State also challenged the
statistical methodology and how the results were used. Our decision,
which we first summarize below, is based on the parties' written
submissions and a hearing.
(3) Summary and index.
Based on the analysis below, we conclude:
* New York was correct in 18 of the individual case determinations,
and SSA was correct in 10. A chart summarizing our findings is at
Appendix A. The standards for determining exclusion/inclusion of a case
differed markedly depending on how the case was categorized (e.g.,
"income" or "shared living"), so the reader will not that
facts
dispositive on one issue for a case may not be so treated for another
issue in the same case.
* Generally, SSA's statistical methodology was valid. There are two
issues, however, which we remand for further development: the use of a
"difference estimator" versus a "direct projection" approach
for
interpreting sample results; and the use of a "point estimate" to
select the exact amount proposed as an APL. We anticipate that our
individual case determinations will lead to discussions which may
resolve or moot these remaining issues, but if not, the Board will
provide further review on an accelerated basis.
For aid in using this lengthy decision, here is an abbreviated index:
I. Individual case determinations A. "other income" cases -
Background - Analysis of the 11 cases B. "Non-federally
participating" cases C. Other cases D. "Shared living" cases
-
Overview of the issues - The "household of another"
regulations as a basis for a presumption of excludability -
SSA's approach to "shared living" cases - The parties'
agreement on "shared living" - How we considered the
evidence - Summary of our conclusions on how to analyze
"shared living" cases - Analysis of the 18 cases II. The
challenge to SSA's APL calculation
(4) I. Individual case determinations.
In this section, we analyze the individual AABD cases in dispute.
Here is a brief background of the review these cases have already
undergone.
The State's manual review process to obtain its APL figure was
basically this: a reviewer would examine a case file to determine the
answers to specific questions listed on an APL questionnaire (related to
matters such as living arrangement, income, and AABD budget for January
1972); the results were reviewed by a supervisor; cases were then
coded according to the answers on the APL questionnaire, some of which
provided a basis for excluding a case; and, finally, data from the
questionnaire was entered into a computer, which screened the cases
using the exclusion codes, and other edits and ultimately produced an
average value for all of the cases which had been categorized as
includable.
In performing their review of sample cases, the federal auditors also
used the APL questionnaire, citing to documents in the case file to
support their findings on various questions. The auditors used as a
basis for their conclusions a compilation of documents including
correspondence between the parties regarding the APL calculation,
minutes of meetings, and handwritten notes based on conversations with
an SSA official who had been involved in APL discussions. The auditors
referred to this as the "Criteria File." NY appeal file, Vol. II.
In
addition to finding errors in including cases (not in dispute here), the
auditors determined that New York wrongly excluded 48 cases in
calculating the APL.
An SSA reviewer analyzed the State's arguments made to SSA
challenging the auditors' findings, and recommended reversal of these
findings in five cases. SSA adopted this recommendation and also
adopted the reviewer's conclusions that the remaining 43 cases were
improperly excluded by New York. New York appealed SSA's determinations
in 28 of the cases.
The parties discussed the cases in three general groups, and we use
the same headings below. The groups are: cases where New York alleged
there was "other income" that affected the classification; cases where
there allegedly was "shared living" producing economies of scale that
affected the classification; and a miscellaneous group of cases
involving other issues. Some of the cases fell in more than one group,
and we analyzed such cases under both headings (except for Case 2, for
reasons explained below).
Based on our analyses of the 28 disputed cases, we conclude that New
York should have been allowed to exclude 18 of the 28 cases. SSA's
determinations are upheld concerning the other 10 cases (including one
which New York conceded during the appeal).
(5) To avoid invading personal privacy, we refer to the cases only by
the numbers assigned in the parties' briefs (the numbers are the same).
See "Contents" pages of Appendix I of the State's brief, and "Table
of
Contents" of Volume II of SSA's brief. Where we must refer to an
individual, we substitute the word "recipient" for the name.
A. "Other income" cases.
1. Background.
Under section 401(b) of Pub. L. 92-603 and 20 CFR 416.2085(a)(1)
(1975), the APL calculation did not include payments to persons if they
had "other income." There was an exception:
. . . Cases in which income was totally disregarded by a State in
determining the amount of the money payment will be considered cases in
which there was "no other income." For these purposes in kind or imputed
income will be treated the same as cash income.
20 CFR 416.2085(a)(2) (1975).
Thus, payments to persons with income were to be excluded from the
APL calculation, unless the State had "totally disregarded" the income
(whether cash, in kind or imputed) when the State originally determined
what the grant would be.
SSA discussed the meaning of the regulation in a letter entitled a
"written confirmation of agreements," dated October 25, 1974:
On the subject of imputed or in-kind income cases . . . we agree that
any case identified as having imputed or in-kind income that was
considered in computing the grant will be excluded.
NY brief, App. II, Ex. K (emphasis added).
This still did not settle the issue of how to treat imputed income
(we discuss the State's response to the letter below), and in a letter
dated November 18, 1974, SSA responded to a State request for further
clarification as follows:
With respect to your third question concerning the rationale for
including "imputed income" in the computation of State debits and
credits, we provided a response in (the letter quoted above). We should
like to augment our response at this time, however, by taking exception
to your statement that "the (6) imputed income concept did not exist in
the AABD Program." While the term "imputed income" was not generally
used in state welfare programs, the concept was universally in use and
is usually referred to as "income-in-kind." All states, New York
included, have taken into account in one way or another, the value of
goods or services provided free or at reduced cost to clients, in their
budget determinations. Some states assigned a dollar value and entered
this on the income side of the client's budget, while others simply
omitted the provided item from the needs side of the budget. When, for
instance, a daughter provides shelter without charge for her father, it
makes no difference whether New York simply omits the item of shelter on
the needs side of the father's budget, or includes its monetary value on
the needs side and makes a corresponding entry on the income side. In
either case, the only difference between the State's concept of
"income-in-kind" and ours of "imputed income," is semantic
rather than
real.
NY brief, App. II, Ex. L.
2. The dispute.
In 1981, HHS auditors determined that the State did not properly
apply section 416.2085(a)(2) and the October 25, 1974 letter quoted
above, and that the State should have included, rather than excluded,
certain cases. The auditors appeared to interpret SSA's standard quite
stringently:
. . . only cases in which the imputed or in-kind income actually
resulted in a reduction of the assistance grant should be excluded from
the APL calculation. . . . Our review of the records of the five
excluded cases did not indicate adherence to the criteria. The budgets
or grants did not reflect reductions as a result of in-kind or imputed
income and the case folders contained no information to support the
contention that the client had received imputed income.
NY appeal file, Vol. I, Ex. 7, pp. 25-6. /2/
(7) On appeal, the State's position basically was that in the 11
disputed "income" cases, such things as the absence of payments for
housekeeper services or a low shelter allowance reflected the presence
of imputed income, and that SSA therefore erred in finding that the
State was wrong to exclude these cases from the APL calculation. NY
brief, pp. 58-9.
SSA's response basically was that there was insufficient proof of the
State's position, and that the State's case could only be established if
one improperly relied on "speculation and inference and . . . a
hypothetical scenario for each case. . . ." SSA brief, p. 88.
3. How the Board applied the rules applicable to "income" cases.
On its face, the basic rule in section 416.2085(a)(2) is broad and
simple: cases of persons who had income will not be used in calculating
the APL. The general rule is one of exclusion, and there is only one
exception. The exception is narrowed by on emphatic term: if income was
"totally disregarded," the case will be included in calculations,
although it normally would be excluded. The narrow exception gives rise
to a burden on the proponent of the exception's applicability--in this
case, SSA--to show that a given case is not governed by the general
rule. This burden does not derive from a general burden of proof on
either party, but arises from the specific regulation involved, which on
its face always excludes "income" cases absent affirmative evidence
that
the case should be included. /3/ But SSA is not alone in having a
burden in "income" cases. Since a determination here to exclude a
case
translates ultimately into more federal money for the State, the State
must show, when challenged on a reasonable basis, evidence to support
its determination. Thus, there is a sharing of burdens; but the
responsibilities arise in a context where we reasonably must give some
weight to the State's view of what occurred: the process involved was
administered most closely by the State and its political subdivisions,
and the critical decision-making occurred years before attention was
focused on the convoluted standard in section 416.2085(a)(2).
(8) The HHS auditors arguably reversed SSA's burden by, in effect,
maintaining that a case must be included absent specific information
showing that the grant was reduced. While the auditors' interpretation
hypothetically might be a reasonable policy, it went beyond the rule
that SSA actually promulgated. In briefing this appeal, SSA disavowed
this interpretation:
Appellant incorrectly states that in order to establish the existence
of imputed income, SSA required the case file to reflect the award of an
initial grant amount thereafter reduced by the value of the imputed
income, resulting in a reduced grant. . . . This is inaccurate. SSA
simply required some indication that the presence of imputed income had
been actually considered by (the State) in determining the grant amount.
In the absence of any such indication, a case would be properly
includable. SSA brief, p. 87.
Although we agree with the standard of review SSA says in its brief
it used, it is not entirely clear that this less stringent standard was,
in fact, used. Note, for example, this testimony of the SSA analyst on
cross-examination:
Q: . . . Let's get this clarified once and for all. Was it the SSA
position that only cases in which the imputed or in-kind income actually
resulted in a reduction of the assistance grant were to be excluded?
A: I think that is the end conclusion. . . .
TR. II, p. 400.
Further support for the proposition that the auditors' ostensible
standard was too stringent can be found in the SSA letter of November
18, 1974 (quoted above on pp. 5-6). SSA noted that imputed income
sometimes had been reflected by states omitting the value of goods or
services from the determination of needs, rather than affirmatively
showing plus and minus quantities. Also, the record contains evidence
that, at least as relevant here, there was no specific need in 1972 to
note the type of information in case files that the auditors' standard
presumably would require. /4/ Whatever the benefit of hindsight, to
impose a rigid documentation requirement meeting today's needs on
yesterday's record-keeping (9) is unfair. Also, the very word "imputed"
indicates that we are dealing with income whose existence is derived
from circumstances rather than simple static measurement. We note the
following from a State affiant (who was referring to two cases in which
the State argued artificially low rents produced imputed income):
In neither case is there any conceivable reason why the caseworker
would have recorded a higher shelter cost and then reduced it. In
truth, there never was a higher cost because the recipients' families
made it possible to avert that need. This is the nature of "imputed
income." It never really existed in the sense that actual income does.
The recipients never had it in hand. Rather, it allowed the recipient,
and, therefore, the State to avert an expense which would otherwise have
arisen.
NY brief, App. II, Burdick Affidavit, P23.
Finally, perhaps the most important factor concerns the overall
context of the narrow dispute over excluding "imputed income" cases:
it
must be remembered that the basic objective was to produce a fair
averaged payment figure reflecting a class of the State's neediest AABD
recipients--those with "no other income." Pub. L. 92-603, section
401(
b)(1). This objective was part of a strategy of luring states into the
program with a guarantee of a "hold harmless" payment level generally
reflecting the more expensive cases--i.e., those of neediest persons.
/5/ In that context, the development of the "no other income" concept
suggests what common sense would suggest: that cases with artificially
low payment levels due to the presence of actual or imputed income
should not be included in calculations because they would unfairly lower
the average. The record here suggests that this simple proposition
(which, as with many basic governmental policies, is much more difficult
to implement than announce) may have become lost to conscious view as
SSA and the State, over a period of years, much paper, and other
attenuating circumstances, became more concerned with the mechanics of
process than with the principles on which the process was supposed to be
based.
(10) The practical effect of all the foregoing is that the Board has
reviewed the 11 cases under a less severe standard than that articulated
by the auditors, and we have considered the inferences one may draw from
circumstances in and surrounding the cases to determine whether SSA has
shown that the State's evidence supporting its "exclude" judgment
does
not reasonably substantiate that judgment. Of course, this approach is
fully within the reach of what SSA said in its brief that it had to do:
to find "some" evidence" that presence of imputed income was
"actually
considered" in determining grant amount. We have also required of the
State more than mere speculation; while we have drawn inferences from
the facts in some cases which support the State, in others we have found
the State's case too insubstantial by any reasonable standard.
4. How the standard above relates to New York's process in 1972.
New York argued essentially that in 1972 the basic needs grant and
personal needs allowances were floor amounts, to the increased for rent,
housekeeper, and homemaker services and other "special needs," and
that
New York did not award increased aid if there was "income" (e. g.,
reduced rent or services provided by relatives) which reduced need;
thus, SSA erred in seeking evidence of a reduction as a way to show that
New York had "considered" the income. NY brief, pp. 57-8. SSA argued
on
appeal essentially that it was not looking for an actual reduction, but
only for better evidence than the State presented that what the State
said happened did happen. SSA brief, pp. 87-8. As we noted above, and
show further below, SSA actually may have sought more than it says it
did on appeal; but in any event, we examined the record to determine,
in accordance with SSA's standard on appeal, whether there is "some
indication" that New York "actually considered" imputed income.
We do
so in this context: SSA presented no substantial evidence rebutting the
way New York said its process worked. Furthermore, our analysis above
leads us generally to agree in principle with the State's proposition
that it should be viewed as having "actually considered" imputed income
when it is clear that more assistance would have been awarded to a
recipient "but for" the presence of the imputed income. As we note
below, the most reasonable interpretation of the "actually considered"
standard is that the case file must show that there was imputed income
directly offsetting funds which otherwise would have been awarded to
meet an identified need. If evidence supports it, the State's "but for"
proposition can fit within that standard.
(11) 5. Analysis of the 11 "income" cases.
New York argued that in 11 cases, it would have given increased aid
"but for" the presence of imputed income, so that the State had not
"totally disregarded" the income. NY brief, p. 59. SSA responded that
there was not "even a scintilla of proof, that the State ever considered
such alleged income in determining the grant amount." SSA brief, p. 88.
We agree with the State in five cases, and with SSA in six. Our
analysis follows:
(a) Cases involving low rent.
The State argued that in four cases the recipients paid "artificially
low rents" which should be reflected as imputed income. NY brief, p.
59. We conclude that the State was correct on two of the cases (9 and
13), and that SSA was correct on the other two (14 and 18).
Case 9:
These facts are undisputed about Case 9: the recipient paid less
rent than she otherwise would have had to pay because her son, who once
had lived with her, continued to maintain the lease to her
rent-controlled apartment in his name, and continued to pay the rent
with his checks. This was done specifically to "allay rent increase due
to a change in name." NY brief, App. 1, Tab 9, p. 11. /6/ The increase
would have been "large" according to a State witness. TR. I, p. 99.
New York argued that this recipient clearly would have been paid more
rent "but for" the son's assistance, and that the difference in the
actual and potential rent amounts was imputed income which was not
"totally disregarded" in determining the recipient's money payment
"since in its absence the State would have paid a higher rent figure."
NY brief, p. 60; see also, NY post-hearing brief, p. 22. The record
indicates that the son maintained the apartment in his name before,
during and after January, 1972. NY brief, App. I, Tab 9, pp. 9-14.
SSA argued as follows:
Nowhere in the welfare file is there revealed any indication that
NYS/DSS considered the rent as (12) "artificially low." Nowhere is
there
any indication that NYS gave the potential for imputed income described
by Appellant any thought whatsoever. . . . NYS/DSS awarded the
recipient the actual value of rent paid. Thus, even if the potential for
a higher rent could be considered "imputed income" to the recipient,
there is not a scintilla of proof that this potential was ever
considered.
SSA brief, pp. 89-90.
SSA further argued:
. . . SSA required some indication in the case folder that DSS had
actually been aware of and had taken into account the presence of
imputed income in establishing the grant . . . (there is no) indication
whatsoever that DSS was aware of the existence of imputed income and
adjusted the grant accordingly. SSA post-hearing brief, p. 13; see
also SSA brief, Vol. II, Tab 9.
It appears that SSA sought nothing less than an actual notation in
the recipient's case file, made in 1972, specifically noting that the
reduced amount of the grant reflected imputed income in the form of the
benefit received from the son. We briefly reiterate and apply here the
results of our general discussion above, as follows:
* We find nothing in the factual record to rebut the evidence that a
caseworker in 1972 cannot reasonably be expected to have recorded in the
Case 9 file the specific statement that SSA seeks. Furthermore, the
omission appears consistent with the descriptions of how the process
worked.
* The concept stated in SSA's October 25, 1974 letter--that only
imputed income "that was considered in computing the grant" would
be
excluded--is not as dispositive here as SSA argues. First of all, it is
not clear that there was agreement with the State on how SSA interpreted
the standard: we note that the State responded to SSA's letter by
stating only that "we agree that . . . imputed income is considered
according to our excluded. . . ." SSA brief, Vol. II, Tab 9, p. 2 of
letter of November 6, 1974. In the record before us, the State's reply
is enigmatic, and does not necessarily confirm the "considered"
standard, (13) nor, certainly, the mechanistic way it was applied.
Second, even assuming the "considered" standard was the agreed-upon
interpretation--and it is a reasonable (if ambiguous) one-- it does not
require the application SSA promotes here. SSA's reading virtually
calls for a recordation of a state of mind. An alternative
interpretation, equally permissible under the bare words of the standard
and much more reasonable in the context of this case, is that it must be
demonstrable from the case file that there was imputed income that
directly offset funds which otherwise would have been awarded to meet an
identified need. Nothing specifically requires the demonstration to
take the form of a specific notation, particularly one which likely
would not have been made at the time, and nothing precludes the use of
reasonable inferences from the case file. The case file in Case 9
clearly shows that caseworkers found that the son had endowed the
recipient with a specific monetary benefit. The case file also gives
rise to the virtually ineluctable inference that the recipient would
have received a larger grant if it had not been for the son's action.
* It can be argued that "rent control" cases as a class involve
persons whose needs were not reduced by imputed income (in the form of
reduced rent), because they simply did not need more rental allowance in
the first place. Thus, the argument would be that such cases should be
included because, although they might from one perspective reflect
reduced payments, actually there would be nothing artificially low about
such cases, and nothing unfair about including them in an averaged APL
statistic. That is, such cases arguably merely reflect the status quo
in New York, and should therefore be included in APL calculations. We
do not face this issue in this decision, because Case 9 does not involve
a normal rent control situation (and neither do not other rent cases
discussed below). While it is a close case, we believe the most
reasonable reading of the record is that Case 9 involves rent control
only coincidently. The important point is that the son in Case 9
created a substantial benefit for the recipient which she otherwise
would not have received, and based on which she did not receive further
assistance, thus fairly placing this case in the category of cases with
artificially low payments which should be excluded to avoid skewing the
average.
(14) * One also might argue that this was a gift rather than
"income," but we see no such finding or argument by SSA. The record
does not support the proposition that "income" should be defined as
narrowly as, for example, it would be in a tax context (and even if it
was, one might argue that the value of reduced rent here constituted
constructive income). Furthermore, as we have previously discussed, the
benefit here, however characterized, impacts a needy case in a way that
makes the case excludable under a fair reading of the basic purpose of
the rule excluding "imputed income" cases.
* As we have already indicated, the record suggests that the unduly
narrow constraints which SSA placed on excludability of imputed income
cases arose less from a conscious policy choice based on program
fundamentals, than from overreliance on implementation mechanisms
evolved at a distance from fundamental program meaning. Interpretations
and tools designed to facilitate objective decisions about including or
excluding a case appeared to have become inflexible ends in themselves.
See, e.g., TR. II, pp. 305-7, 361-9, 400-1. We do not suggest that
auditors and SSA review personnel were not doing their jobs properly; in
fact, the record suggests they were careful and diligent. The
difficulty is that the interpretations they were implementing were
unreasonable or inconsistent with the fundamental purposes of the APL.
For the foregoing reasons, we have determined that New York
reasonably excluded Case 9 from the APL calculations.
Case 13:
The recipient here paid $35 per month rent in a 3-room bungalow. She
lived alone in the house. Her mother, the landlord, lived in another
house on the property. New York argued that the rent was obviously
below normal, and that "but for" the benefit the recipient's mother
was
conferring on her, the recipient would have had a substantially higher
rent for which New York would have awarded more assistance. Therefore,
argued the State, the increment above actual rent should be deemed
imputed income which, since it was not totally disregarded in making the
grant, means that the case is excludable from the APL calculation.
This, of course, is essentially the same argument as made for Case 9,
above, and SSA's response essentially was the same as for Case 9. NY
brief, pp. 59-60, and App. I, Tab 13; NY post-hearing brief, (15) pp.
21-2; SSA brief, pp. 88-90, and App. III, Tab 13. The only additional
issue /7/ concerned whether the rent of $35 actually was low. On
cross-examination of the State's witness, SSA challenged his assertion
that the rent was low. His evidence was not ideal--for example, he had
not seen the house in question--but he was sufficiently well-informed
and affirmative in his testimony so that, on balance, given the lack of
any specific contradictory evidence from SSA, we believe we are
compelled to believe his testimony that the rent was, indeed, low "by
any standard." Burdick Affidavit, P22; TR. I, pp. 97-8, 123-6, 145-7.
We note that the case history indicates the house was "neat, clean and
well-furnished." NY brief, App. I, Tab 13(B), p. 4. The same page
indicates that there was a "strong smell of animals, perhaps rats around
the neighborhood." However, this does not say the place was
rat-infested, as SSA implied on cross-examination. TR. I, p. 125. The
State's contemporary "first rental schedule," which presumably gives
us
a rough idea of anticipated rental values for AABD recipients, indicates
a value of $90. NY brief, App. I, Ex. M, p. VII-16; see also, TR. I,
p. 147. Thus, overall, we find that the recipient in Case 13 paid
substantially less rent specifically because of the benefit conferred on
her by her mother, and that she accordingly received less assistance
than she otherwise would have. Having so found, we then note that the
case becomes otherwise essentially the same as Case 9 above, and we
therefore conclude, on the bases discussed previously, that New York
reasonably excluded the case.
Case 14:
Imputed income was an additional basis for exclusion argued by New
York, the primary basis being a shared living situation. NY brief, App.
I, Tab 14, p. 1. The "shared living" elements of this case are
discussed in another section.
The recipient here rented a furnished room in a family's apartment.
The rent was shown to be $48 per month in 1966, and $12 per week in
1971. New York indicated that the lack of any (16) substantial increase
in rent during that time gave rise to a presumption of imputed income
(in the form of artificially low rent), when viewed together with these
factors: the recipient's access to the entire apartment; indications
that the recipient may have been doing housework in the apartment; and
the appearance of a friendly relationship with the primary tenant.
Our review of the record reveals that these points do not justify
excluding this case on the basis of artificially low rent. The State's
primary point on this issue was that there was low rent in exchange for
housekeeping services. The State's witness stated in his affidavit that
he found on review of the file that the recipient could be "providing
housekeeping services" (P24; see also P20), but on cross-examination he
said he did not find evidence that this was specifically so in this
case. TR. I, pp. 120-21. Further, when on cross-examination his
attention was called to the statement in his affidavit (P24) that
housekeeping services were being provided throughout the apartment, he
stated that he suspected it "was an error." TR. I, p. 127.
Moreover, the State did not rely on any evidence here to show that
the rent was, in fact, low. Rather, the State pointed to evidence which
merely showed that the rent had not been raised over a period of time.
There is nothing to indicate what the rent for the primary tenant was at
any point in the period covered by the records, so we have no basis to
judge what portion the recipient was paying, or whether the primary did
or did not have any raise in rent during the period.
Finally, with respect to the recipient' relationship to the primary
tenant, while there is some information from which one can infer that it
was not an arm's-length sublet arrangement (see our discussion under the
"shared living" section below), this information is not sufficient
for
us to find definitively that the relationship benefited the recipient in
the form of lower rent than she otherwise would pay.
In short, the State's "exchange" theory was based on a misreading
of
the file, and the State's position is reduced to mere speculation about
whether the rent was low and why. Thus, we cannot agree with the State
that this case should be excluded based on an affirmative finding that
there was imputed income in the form of artifically low rent.
Case 18:
The proposition that imputed income was present in Case 18 was a
subsidiary argument for this recipient; as with Case 14, the (17)
primary issue here concerned whether there was a "shared living"
arrangement, and that element of the case is discussed elsewhere.
The element of the case relevant here is this: the recipient in this
case, argued New York, had an amicable relationship with the primary
tenant and helped with cleaning, "giving rise to the possibility of
imputed income in the form of lower rent. . . ." NY brief, App. I, Tab
14, p. 2. The State witness's testimony largely covered this case and
Case 14, above, together. Burdick Affidavit, P20, 24.
1969 and 1970 visit records showed the recipient paying $15 per week
rent for a room in a five room apartment. NY brief, App. I, Tab 14, pp.
6-7. The key phrase that New York focused on was the statement on the
7th page that the recipient "does a little cleaning around the house."
See, e.g., TR. I, p. 145. The quoted phrase on cleaning appears
separately in a paragraph dealing with some other issue (the heading is
illegible); the entire paragraph states:
We asked her what she does with her time. She told us that she
usually goes out and visits her friends and takes walks and does a
little cleaning around the house.
Thus, we need not necessarily read the key phrase as indicating any
imputed income, because it was an incidental statement included in an
answer to a question about how the recipient occupied her time. Stated
another way, it is just as easy to ascribe no relevant meaning to the
phrase as to give it meaning. In context, that is not enough: we do
not know if $15 per week was low or reasonable, and we have nothing that
tells us about the nature of the recipient's relationship to the primary
tenant. She may have been referring merely to areas of the house which
she occupied or used. Thus, the phrase is at best suggestive of a need
to look at more records, but we of course have no further records before
us. By itself, and with no further information about the rental rate
circumstances, the phrase is insufficient as a basis for excluding the
case as having artificially low rent. Even when buttressed with the
State witness's general testimony there is insufficient information to
do more than suggest one explanation among many. Therefore, we conclude
that SSA properly determined that Case 18 should not have been excluded
from the APL calculation on the basis of artificially low rent.
(b) Cases involving housekeeping services.
The State argued in three cases that there was an absence of a
supplementary grant for housekeeping or homemaking services only (18)
because of the presence of services provided by a relative (in Cases 7
and 16) or because of special circumstances involving a vendor payment
for housekeeping services (Case 4). NY brief, pp. 57-8, 60. We
conclude that the State was correct on one case (Case 4) and incorrect
on the other two (7 and 16).
Case 7:
Case 7 concerned an elderly recipient who apparently had a heart
condition and other ailments. New York argued essentially that the
recipient was eligible for a special grant for housekeeping services,
but that her daughter provided such services, relieving the State of the
necessity of awarding the grant. NY brief, App. I, Tab 7. The State's
witness stated generally that caseworkers were supposed to make use of
all available resources, including dependence on family or relatives, to
avoid expense to the State; often, he said, special grant needs never
went any further than to be communicated to the caseworker. Burdick
Affidavit, P27 and 28. He further stated that while the record did "not
contain sufficient information to permit a definite conclusion," the
general practice he described persuaded him that the State was relieved
of providing a special grant in Case 7. Id., P29.
Our review of the record indicates this problem with the State's
position: the caseworker had specifically required further information
as a condition of determining eligibility for the special needs grant,
and nothing indicates that he got the information.
The State pointed to section 352.9(c) of its regulations as providing
for eligibility of the recipient. That section merely states, in
pertinent part, that "Housekeeping services for a recipient who is
unable to perform housekeeping tasks shall be provided." NY brief, App.
II, Tab N, p. E-2. On its face, the provision requires some sort of
determination that the recipient is unable to perform housekeeping
tasks.
The brief record in this case contains several reports that the
recipient or her daughter told the caseworker that the recipient was
unable to move around much or do housework, and that the daughter
provided such housekeeping services. NY brief, App. I, Tab 7, pp. 3, 5.
Without more, the State would have a reasonable argument for excluding
the case. But there is more: on both of the visit records (which are,
with the exception of a brief recertification document which is not
material to the point under discussion, the only records), the
caseworker noted that the recipient requested housekeeping assistance
and that the (19) caseworker would pursue this, but needed verification
of the recipient's medical condition. For example, the second visit
record notes that the recipient was told that housekeeping services
"would be look(ed) into and that verification of her Health Condition
would be needed before this could be done." Id., p. 5 (emphasis added).
The State witness's statements are not a sufficient basis to
disregard this evidence. As we noted, he stated that requests for
special grants sometimes "never went any further than to be communicated
to the caseworker" if other resources were available. Burdick
Affidavit, P28. Here, however, the caseworker made a point of recording
the requests twice, suggesting how the matter could and should be
pursued, and twice requesting further information needed to determine
eligibility. The witness admitted that there was no evidence of medical
verification. TR. I, p. 127-8. Furthermore, the extent of the
daughter's services is not clear (as the witness himself noted), and
both visit records note that the daughter wanted her mother closer to
her because it was difficult to go back and forth, from which one might
infer that the housekeeping services were not provided in amount
sufficient to support an inference that the caseworker "considered"
them
in determining the grant amount. (We do not promote that inference, but
merely note it is no more strained than others which one may draw.)
Further, this also makes it less reasonable to infer that the daughter
actually provided services in January.
Thus, on the record before us, there is affirmative evidence that the
question whether the recipient qualified for a special purpose grant was
an unsettled issue. Therefore, one cannot reasonably draw an inference
that "but for" the daughter's services, such as they are, a special
needs grant would have been awarded. The circumstances are simply too
ambiguous, and the caseworker's insistence on evidence of disability
might also be read to imply that he actually questioned eligibility for
the grant. Since the inference cannot be drawn, we are left with
insufficient evidence that there was imputed income derived from the
daughter's services to meet a need which otherwise would have given rise
to a grant. Therefore, we conclude that SSA correctly determined that
Case 7 should not have been excluded from the APL calculation.
Case 16:
New York's arguments about this case were essentially the same as
those offered for Case 7, above, and the witness' statements, mentioned
above, were applicable to this case as well as Case 7. New York's
position suffers the same problems as Case 7, with the additional
weakness that it is even less clear that there actually was imputed
income.
(20) As with Case 7, this case involves an ailing elderly woman
living alone who had a relative (in this case, a cousin) who helped her.
NY brief, App. I, Tab 16. The record contains documents reporting
several requests from the recipient for housekeeping services. Id., pp.
3-6. Similarly, caseworkers twice asked for a physician's verification
of the recipient's condition. Id., pp. 4, 5. The record indicates she
was short and stooped, and therefore could not wash windows well, but
does not otherwise specify any debilitated condition: in fact, one
report said "Client's health appears to be all right" (Id., p. 3),
and
another said she was "feeling fine." Id., p. 4. The last report in
the
record indicates that in late 1970, the recipient had sent in a doctor's
note. There is nothing further and nothing to show whether she was
found eligible for a special needs grant. Contrary to what New York
argued, we do not know whether the note "justified" the request. Id.,
p. 1.
As with Case 7, we conclude that the basis upon which the State's
proposed inference of grant avoidance would be based--i.e., that she
could have been given such a grant in the first place--is insufficiently
shown, if not actually rebutted, by this record.
Furthermore, there is little to indicate that there were any services
being provided by her cousin. The only direct reference to this
possibility is a statement on one report that "It seems that
(recipient's) cousin occasionally assists her with the housekeeping
chores." Id., p. 3. The only other reference to the cousin occurs in
the context of a statement that since recipient did not speak English,
the caseworker accompanied her to her cousin's to continue the
interview. Id., p. 4. The single reference to an occasional assist
with housekeeping hardly seems sufficient as a basis for a finding of
imputed income in January 1972.
Since the presence of imputed income is questionable, and since there
is no evidence that the recipient here would have been awarded a special
needs grant for housekeeping services, we conclude that SSA correctly
determined that Case 16 should not have been excluded from the APL
calculation.
Case 4:
This case concerns a "vendor payment," which, in this context, is
a
payment for housekeeper services made to a third party for the benefit
of a recipient. It is undisputed that cases involving vendor payments
were to be excluded from the APL calculation. NY brief, App. II, Tab K.
The issue is whether Case 4 was a vendor payment case in January 1972.
SSA reviewed an archive called the "Duane Street Composite Roll,"
which
SSA said "represents all (21) payments made by NYC DSS. . . ." SSA
brief, Vol. II, Tab 4, p. 1. SSA found no record that any vendor
payment was made for this recipient in January, 1972. Id. New York
acknowledged that lack of specific evidence of a vendor payment in
January 1972, but argued that there was evidence "that leads to a strong
inference that she (the recipient) did participate in the program for
that month." NY brief, p. 61. The State also argued that SSA was
erroneous in assuming that vendor payments would be recorded on the
Duane Street Composite Roll, and submitted an affidavit to that effect.
Id., and see Reilly Affidavit, NY brief, App. II. Mr. Reilly's
affidavit contains a single paragraph (number 6) in which he
categorically, and conclusorily, states that vendor payments would not
be reported on the Composite Roll for January 1972. /8/
SSA did not dispute the Reilly affidavit in its post-hearing brief,
and did not deal with the issue at the hearing. See, e.g., TR. II, p.
311. We have no basis to do anything other than accept the Reilly
affidavit at face value, and we therefore find that the mere fact that
the recipient in Case 4 did not have vendor payments recorded for
January 1972 in the Duane Street Composite Roll is, by itself, an
insufficient reason to say that Case 4 is not a vendor payment case.
The question thus becomes whether evidence in the case gives rise to
a reasonably well-based inference that the recipient in Case 4 did
benefit from vendor payments in January 1972. Here, we bear in mind
that SSA's primary argument on this case did not go to this issue, but
to the question of what appeared in the Composite Roll (although, of
course, SSA took the position of rejecting inferences generally, a
matter we have discussed above). Although the materials in the record
are not all one would like them to be, we conclude that, in the context
of this case, they do give rise to a reasonable inference that a vendor
payment was made in January 1972.
The first item in the record is a letter dated July 15, 1971
concerning the recipient. The letter, from a "social worker assistant,"
notes and endorses a physician's "strong" recommendation that the
recipient be provided housekeeping services. NY brief, App. I, Tab 4,
p. 4. The next item is a memorandum (22) dated August 5, 1971, noting
the need for a housekeeper and "approving" a payment of $29 weekly
to a
certain housekeeper. Id., p. 5. The next item is an "authorization to
vendor agency for housekeeper service" for the recipient covering a
period of roughly one year, from September 1972, through August 1973.
Id., p. 6. The next item is a similar form for approximately the
subsequent year. Id., p. 7. The latter forms are Form W442.
It appears clear that the recipient did receive the benefit of vendor
payments before and after the vital month. There is nothing in the
record, and particularly nothing on the first W442, which indicates
anything out of the ordinary. As noted earlier (see discussion under
Case 7 above), regulations required assistance for housekeeping services
once a need had been determined. The need in this case was clear at the
beginning, and clear over a period of years subsequently; it would seem
reasonable to anticipate that the need continued in the interim. The
first authorization noted that the recipient was to be paid "directly
until W442 ret'd signed." Id., p. 5. This apparently would not make the
payment not a vendor payment, since the money still could, and probably
did, go directly to the housekeeper identified on the same page, and we
have no reason based on the record before us to determine that this
necessarily would have made the case appear on the Composite Roll; at
the same time, it does state that payment would continue until a W442
was executed, which would cover the month in question. We note that the
two W442s in the record are quite similar; one might expect such
similarity if both were extensions of an earlier obligation, and might
not expect it if the first W442 was the first authorization of payment
after a hiatus.
In summary, we find that the record gives rise to a reasonable
inference that vendor payments were made in January 1972 to this
recipient. The State was closer to the process than SSA, and in the
absence of anything to the contrary, the State's affirmation of what
happened is entitled to some weight. SSA's primary basis for its
determination on this case was rebutted in a sworn affidavit. The
records strongly suggest a continuity of vendor payments through the
period in question. Overall, therefore, we conclude that New York
properly excluded Case 4 from the APL calculations.
(c) The remaining "other income" cases.
The cases analyzed above fell into groups with common issues. The
remaining four cases will be discussed individually. We conclude that
SSA was correct on two (3 and 24) and that the State was correct on two
(8 and 23).
(23) Case 24:
New York argued as follows: the recipient in Case 24 was in poor
health and lived alone, and so needed a telephone for emergency use. A
nephew paid for the telephone. An allowance for a telephone could be
approved in a case of medical necessity. Since the nephew paid for the
telephone, the State was spared the necessity of a special needs
allowance; thus, the imputed income (the nephew's payment) was
"considered" in determining the money payment to the recipient. NY
brief, p. 61.
SSA's position basically was, as with the rent cases discussed above,
that there was no evidence whatsoever to support the proposition that
anyone considered a special needs grant for telephone services. SSA
brief, p. 91.
This is what the record shows: the evidence does indicate that the
recipient was in bad health and that the nephew was paying the telephone
bill in 1966 and 1968. SSA brief, Vol. IV, Tab 24(A), pp. 1-4. The
record shows that the recipient had an "available telephone no." in
March, 1972. Id., p. 5. However, the record does not indicate whether
the nephew continued to pay for the telephone after 1968, nor whether
the telephone the recipient could use in March 1972, was hers or someone
else's or was being paid for by her or someone else. The State's
witness said in his affidavit:
An allowance for a telephone could be approved as a special needs
grant under appropriate circumstances, including medical necessity.
Although the case folder does not contain sufficient information to
permit a definite conclusion (for reasons noted in P28) above), it would
appear that the recipient here may have qualified for a telephone
allowance on the basis of special need, which the State would have
furnished but for the fact that a relative was already paying for the
service.
Burdick Affidavit, P30.
The cross-referenced P28, already mentioned, noted that often special
needs requests went no further than the caseworker if other resources
were available.
A number of facts in this case, each by itself perhaps insufficient
to support a determination, combine to lead us to be unpersuaded by the
State's arguments. First, as noted, there is roughly a four-year gap
between the time the file indicated that the nephew was paying for
telephone services and the month involved in the APL calculation. The
evidence concerning the availability of a (24) telephone in March 1972,
is too scanty to be useful. While a gap of several months might be
dismissed as insufficient to overcome a reasonable presumption that the
nephew's help would have continued absent some contrary indication, the
gap here is much more substantial. Second, although Mr. Burdick stated
that a special needs grant could be made for telephone services, the
State does not cite support for this proposition (in contrast to the
State's argument in Case 7 above), and our review of New York's
regulations in the record does not disclose anything authorizing such
grants. NY brief, App. II, Tab N. New York's regulations did state
that "Supplemental allowances and grants may not be made other than as
authorized under the regulations," apparently referring to the
regulations in Tab N. Id., P C.1.(b) on p. C-2. Third, if we were to
accept as correct Mr. Burdick's statement that such grants were
authorized by something we have not seen in the record (and common sense
seems to suggest that a telephone allowance would be appropriate in some
cases), then we note Mr. Burdick's reference to "appropriate
circumstances, including medical necessity." As in Case 7, this would
appear to refer to the need for additional decisions before a special
needs grant could be awarded; and there is no evidence that such a
determination was made or even considered. Furthermore, although the
recipient in Case 24 was in poor health, we are not prepared to say,
based on the record before us, that her health condition justified a
special grant for a telephone.
In short, Case 24 involves much speculation based on much ambiguity.
While the case has elements that make it a closer call than some others
here, on balance we conclude that SSA was correct in determining that
the evidence was too insubstantial to support the State's decision to
exclude the case.
Case 3:
New York argued that the recipient in Case 3 received imputed income
in the form of tuition at a trade school paid by a relative, and that
this income obviated the need for a special grant which could have been
awarded for trade school tuition. NY brief, pp. 61-2. The case is
complicated by a question of whether the income was attributable to
January 1972. Id., p. 62., fn. 1.
SSA noted, and New York did not dispute, that the case file for this
recipient was lost after the federal audit. SSA's brief, p. 92; NY
brief, App. I, Tab 3, p. 1. The record thus contains only one document
from the file on this recipient, other than the auditors' findings and
the argument. NY brief, App. I, Tab 3, p. 2. This document, a single
case history sheet, indicates that in August 1972 the recipient was
attending a school for medical and dental assistants, that she was in a
ten-month course which (25) began in February 1972, and that her sister
paid the tuition. The record does contain other material prepared
during the review of the cases which appears to reference facts gleaned
from the missing file, but there is nothing material to the issue here.
SSA brief, Vol. II, Tab 3. Our review of the audit reports does not
indicate any additional illumination.
The State's affiant, Martin Burdick, stated that payment for "tuition
for vocational training could be approved as a special needs grant under
appropriate circumstances." Burdick Affidavit, P31. New York's
regulations do state:
for any of such persons who may because of their case circumstances
require any of the following items in accordance with applicable
provisions of law, Board rules and Department regulations, the standard
of need shall include the cost of the required item or items in
accordance with such provisions: . . . occupational training. . . .
NY brief, App. II, Ex. N, p. B-2.
And the regulations also state:
(e) Occupational training
(1) . . . allowances shall be made for the costs of tuition, books,
supplies and other essential items required . . . to obtain suitable
occupational training from a trade school . . . provided, however, that
such person could not otherwise obtain such training without cost and
demonstrates to the satisfaction of the social services official that he
possesses the talent, aptitude and ability necessary to benefit from the
proposed training. . . .
(2) Recipients in an approved occupational training program,
including the Work Incentive Program, shall receive upon enrollment
therein a one-time allowance of $55 to meet additional clothing needs
incident to such training.
Id., p. C-10.
There is nothing in the record to indicate that the circumstances, in
Mr. Burdick's words, were "appropriate" for award of a special (26)
needs grant. There is nothing to show that the recipient's
circumstances met the requirements of the regulation quoted above (or,
if it was not applicable, whatever other criteria would have been
applicable to determine what was "appropriate.") Furthermore, the
explicit requirement in the regulation for a determination that training
could not otherwise be obtained without cost raises questions about
whether the sister's help could be imputed income in the first place:
if the regulations provided specifically that New York would not pay for
tuition if other assistance was available, then the present of the other
assistance would not obviate the need for a grant in the same way as
contemplated by the policy on imputed income we are dealing with here.
New York argued that it was not unreasonable to assume that the
recipient would have had to pay in January in order to reserve a place
for a course beginning in February. It would appear as equally
reasonable to assume that the recipient might have had to pay in
December, or November, or even earlier. There is no evidence other than
that the course began in February. The inference New York would have us
draw is entirely too speculative, and the circumstances are exacerbated
by the loss of the records which the State and/or the City had custody
of and for which the State bore some responsibility in order to support
its APL claim.
Thus, there is no evidence that the recipient was eligible or
qualified to receive the special needs grant for tuition. It is even
arguable, because of the regulation, that payment of the tuition by a
relative might not be imputed income in any event. While the lack of a
January payment record might not otherwise be dispositive, it joins the
other weaknesses in the case to further call into question the inference
the State would have us draw. Finally, we cannot overlook the fact that
the absence of any further information to support the State's case is at
least in part the State's responsibility.
Based on the foregoing, we conclude that SSA correctly determined
that the State erred in excluding Case 3.
Case 8:
This case involved an imputed income issue and an issue concerning an
alleged budget reduction due to residence in a medical facility. /9/
Shared living was also an issue, substantially overlapping imputed (27)
income, and the case is covered in another section. Although subsumed
under the "other income" heading, the case was primarily presented
on a
somewhat different basis, discussed here.
It was not disputed that a ground for exclusion of a case from the
APL was a budget reduction due to residence in a medical facility for
part of the month. Presumably, exclusion was appropriate because such
cases generally would involve reduced AABD payments, so that their
inclusion would artificially deflate the average.
The record shows the following. The recipient had been living with
another person, apparently the recipient's landlady, whom we refer to as
"Ms. P." The recipient was hospitalized from January 13 to 24, 1972,
and
then returned to live with Ms. P. There is no indication of a reduction
in assistance for January, despite the hospitalization. The recipient
was described as needing extra care for several weeks after discharge
from the hospital. Apparently, Ms. P. asked for additional assistance
on behalf of the recipient (the record indicates that Ms. P. was an
intermediary for the recipient on several occasions). By letter of
January 27, 1972, the State wrote Ms. P. that "For the month of
February, we will allow $110 for room, board, and care for this lady."
NY brief, App. I, Tab 8, p. 10 (emphasis added). This was a substantial
increase over the amount awarded for January. The recipient was also in
and out of a hospital in February; however, unlike January, Ms. P. was
asked to refund an amount of room and board payments attributable to the
February hospitalization. Later, the recipient's caseworker simply
offset the amount due against March's payment.
Based on these facts, the State argued that the recipient's January
1972 budget was, in effect, reduced because of her residence in the
medical facility for part of January. The State reached this conclusion
as follows:
Rather than actually reducing her grant because of her ten-day stay
in a hospital during January, the State used this money as an offset to
compensate (Ms. P.) for the extra care she provided after the
recipient's release from the hospital on January 24, 1972, until the end
of the month . . . the State used the money it would have gained by
reducing (the recipient's) budget (because of her hospital stay) to
compensate her landlady for the extra care she provided . . . for the
last week in January. Accordingly, (the recipient's) grant for January
was in effect reduced for her ten-day hospital stay . . . because the
amount saved was used (28) for the additional compensation to her
landlady for the extra care at the end of the month.
NY post-hearing brief, App. B, p. A-8.
The State also argued that circumstances also separately indicated
imputed income to the recipient. As stated, we discuss this under the
shared living section.
SSA's response essentially was, as with the other cases in this area,
that the State's position depends on speculation and that there is
nothing substantial in the record to support it.
When examined in the general context we discussed above, we conclude
that the inferences the State would have us draw about Case 8 are
reasonable.
The recipient's grant for January could have been, and arguably
should have been, reduced. The most reasonable suppositions concerning
why it was not are these: error or inadvertence; to preserve the
recipient's right to her residence while she was briefly in the
hospital; or because the case worker bent rules to solve a problem
simply rather than achieve the same result through a technically-correct
approach which involved more time and trouble. We find that the
circumstances are more consistent with the third proposition than the
other two, and our finding is supported by the fact that New York, which
we reasonably may assume is more familiar with how caseworkers behaved
in New York than SSA is, affirmatively promotes the third approach.
A budget worksheet states, "ret's to (Ms. P.'s) from hsp. 1/24/72.
Extra care required, for a few wks. at least." SSA brief, Vol. II, Tab
8(B), p. 7. That entry apparently was made on or after January 24.
Then, by letter dated January 27, a caseworker stated that "we have made
a decision regarding your request for some additional payment for the
additional care (the recipient) requires, after her discharge from the
hospital." NY brief, App. I, Tab 8, p. 10 (emphasis added). The
decision was to increase the grant for February, but the letter, and the
earlier budget worksheet, clearly identified a need for care for the
last week in January. In our view, it is much more reasonable to assume
that Ms. P. provided this care than that she did not, particularly so
close to the recipient's release from the hospital, when her need
probably was greatest.
The caseworker who recorded the need and the response on several
documents was also the caseworker who, only a month later, recorded the
"overpay't in Feb." Id., p. 15. And it was the same caseworker (29)
who
suggested, as opposed to an earlier letter from another official who
demanded repayment, that "Rather than have you send us this money, we
are deducting it from March's room and board. . . ." Id., p. 15. Had it
been a different caseworker, or a greater distance in time, an error
might be a more reasonable assumption, but it seems unlikely here; we
find this same case-worker using a short-cut of offset, rather than
recouping the February overpayment as another worker had stated would be
done (see p. 14). All of this reasonably indicates that a deliberate
choice was made not to recover overpayments made in January.
While it might not be unreasonable to continue to pay rent to
preserve a right to occupancy, /10/ board was involved here too, and
board would not have continued during hospitalization; furthermore, the
caseworkers were quick to recover both board and room in February.
Nothing in the record indicates Ms. P. made an issue of this. We do not
find it reasonable to assume that room and board had to be paid in
January solely to protect the recipient's right to reoccupy her
residence.
We therefore are left with the third proposition, the one pressed by
the State: that a caseworker simply met the recipient's
clearly-identified care needs by not recouping an overpayment in
January. This is the most reasonable of the obvious explanations, is
consistent with closely subsequent short-cut action by the same
caseworker, and, after all, is the explanation of the State, which was
closer to the administrative process involved than SSA. The inferences
we draw are much more than speculation; they appear to be the only
reasonable explanation. We conclude that the State reasonably excluded
the case.
Case 23:
As the appeal developed, the State began to rely on the argument that
this case was one in which the recipient was "non-federally
participating," a separate basis for exclusion which we discuss below.
However, New York also argued that there was imputed income due to
receiving meals from an unknown source, and due to an early closing of
the case based on recoupment of funds attributable to January 1972. We
conclude that there is insufficient (30) evidence that the recipient was
receiving imputed income in the form of meals, but we conclude that the
State reasonably excluded the case based on the facts surrounding the
early closing.
We deal first with the meals. Documents from October 1969 showed
recipient renting a room in a hotel with cooking facilities for $21 a
week. NY brief, App. I, Tab 23, pp. 4, 5. A visit in March 1970
indicated a rent of $14 a week and no cooking facilities. Id., p. 7.
Budget plans from December 1969 and July 1971 have an entry for
"restaurant" under "needs-income." Id., pp. 6, 8. But a
budget plan
with the date January 16, 1972, does not show the "restaurant" entry.
Id., p. 9. There is no explanation of how the recipient ate thereafter.
The State argued that "if he received (meals) free from some unknown
source," there was imputed income because the State did not pay a
restaurant allowance despite the apparent absence of cooking facilities.
We find that the State's position requires entirely too much groundless
speculation. An equally reasonable (and equally speculative)
alternative explanation is that the recipient sometimes changed hotel
rooms: note the implication that between October 1969, and March 1970,
there was a move, since rent was reduced and cooking facilities
disappeared. That also might have occurred in January 1972. Whatever
the case, there is insufficient information to support the State's
argument, and one can speculate endlessly on what happened.
The State's position on the recoupment issue is more persuasive. The
recipient was made eligible for Social Security retirement in February
1973, with ful retroactive effect through June 1971. Id., p. 10. In
August 1973 an AABD administrator learned of this, and offered the
recipient the option of repaying AABD funds or of closing his AABD case
for "at least three years" (the administrator had identified needs
which
would have been eligible for continuing assistance at a reduced amount).
Id., p. 11. The recipient opted for the latter. The record also
contains evidence that SSA and the State agreed that cases which were
presumptively determined eligible for AABD assistance, and later not
confirmed as eligible, were excludable. Id., pp. 14, 15. Strictly
speaking, this agreement does not apply here, but the retroactivity
principle in the letter is important. When coupled with the general
principles articulated above, this leads us to conclude that this case
fairly should be excluded. The point is that this recipient received
AABD funds to which he clearly was not entitled because of the "income"
from retirement benefits, and, when this was recognized, the AABD
assistance was, in effect, retroactively reduced. The case would have
been more sharply defined if the grant had been reduced retroactively
through recoupment, but it was unreasonable to expect this destitute
person to repay the several thousand dollars involved. Instead, the
administrator did something which (31) was, for all practical purposes,
just as good: he stopped further AABD payments (presumably until the
deficit was made up).
This is a problem of focusing to absolutely on the January 1972 time
frame that the purposes of the measurement exercise were lost to view.
In fact, the amount of the retirement income was very specifically
"considered" in determining the grant amount, albeit retroactively.
When the payment error was recognized, recovery was promptly undertaken
by reducing the amount of AABD assistance the recipient would have
received to zero. It is clear that this was done to offset, or recover,
or recoup--however one wishes to describe it--the payments (including
January 1972 payments) made in error. In short, there was an effective
reduction in the amount of the January 1972 assistance amount because
the income from retirement benefits was ultimately "considered" in
computing the grant. This was done retroactively, and one might even
call the reduction a constructive one; but it was done, and it is clear
that in the final analysis, Case 23 became one of the kind of cases
which should not fairly be included in the APL calculation. We conclude
that the State should have been allowed to exclude the case.
In summary, then, we have concluded the following with regard to the
11 "imputed income" cases: SSA correctly determined that six should
not
have been excluded on the basis of imputed income (nos. 3, 7, 14, 16, 18
and 24), and the State correctly determined that five should be excluded
(nos. 4, 8, 9, 13 and 23).
B. "Non-federally participating" cases.
It does not appear to be disputed that cases which were
"non-federally participating" (i.e., were not receiving federal AABD
funding) in January 1972 should be excluded. NY brief, p. 63; SSA
brief, p. 97; TR. II, p. 269. The State argued that three cases (nos.
5, 18 and 23) fell in this category. NY brief, p. 63. SSA argued that
New York's basis for exclusion consisted of unpersuasive evidence
improperly developed outside the scope of the parties' agreement about
how to determine whether a case would be included or excluded. SSA
brief, pp. 94, 97.
A major problem here is this: the parties, in their early
negotiations over classifying cases, agreed to something that amounts in
practical terms to an indirect or surrogate determinant of federal
participation--i.e., the point when a case was reclassified from
"presumptive aid to disabled" (PAD) to "aid to disabled"
(AD). During
the appeal, apparently for the first time, New York reviewed records to
determine directly whether federal funds actually had been claimed, and
produced affidavits that no (32) records could be found of such claiming
for January 1972 for the three cases involved here; on that basis, New
York argued the cases should be excluded. /11/ NY brief, p. 62; App.
I, Tabs 5, 18 and 23; App. II, affidavits of John Reilly and Samuel
Wachtel. SSA argued that New York said years ago that it could not
perform the search it did for these three cases, which was one reason
for the agreement to review the PAD/AD determinations. SSA argued that
reviewing the latter determinations, in accordance with the agreement,
requires the conclusion that these cases are includable. SSA brief, pp.
94, 97; Vol. II, Tab 5; Vol. III, Tab 18; Vol. IV, Tab 23.
1. The parties' pre-appeal "agreement".
Both parties referred to the pre-appeal agreement as supporting their
respective positions, with little analysis. There are several letters
in the record dealing with the issue here, although the most definitive
is an October 22, 1979 letter to the State from an SSA official. SSA
brief, Vol. II, Tab 5. That letter states in partinent part:
It was our understanding, as well as that of the HEW Audit Agency,
consistent with agreements reached with New York in October 1974, that a
case presumptively eligible on 1/1/72 or earlier would be included in
the APL recalculation if the case was finally determined to be eligible
for the entire month of January 1972, and the 1/1/72 grant was claimed
for FFP at any time. Further the claim must have been approved by SRS
and not subsequently reclassified FNP.
In your letter of May 24, 1979, your point out that it was not
possible to determine the existence of federal funding on a case-by-case
basis in New York City during the data collection phase of the APL
project. As a result, you found it necessary to include presumptive
cases for which FFP was available in accordance with federal
regulations. As we understand your instructions for New York City,
presumptive cases were to be excluded if: 1) not subsequently
determined AABD, or 2) the date of eligibility was after 1/1/72, or 3)
the date of determination was not in January 1972, and was made more
than 90 days after the date of application. Further, we understand that
this issue does not apply to districts outside of New York City, since
those (33) cases were coded in such a manner that only cases in which
FFP was claimed for 1/72 were included.
I trust that the aforementioned is a true representation of New
York's position. With the assumption that there are no inaccuracies in
any of the details related, SSA prepared to accept New York's
methodology for inclusion of New York City presumptive cases in the APL.
. . .
(emphasis added)
The State responded to the foregoing letter by a letter dated
November 8, 1979. Id. That letter stated in part:
This is to confirm your letter of October 22, 1979 concerning the
treatment of presumptive cases in the APL. Your outline of the
methodology used by the State to determine the includability of
presumptive cases is correct. As you stated in your letter, the problem
of determining Federal participation for January, 1972 payments to
presumptive cases occurred only in New York City. In the Upstate
counties, the case records contained sufficient information to allow the
data collectors to determine when FFP was claimed, and the Upstate APL
questionnaires were completed to ensure that we would include only those
presumptive cases for which FFP was claimed in January, 1972.
(emphasis added)
The record also contains New York's letter of May 24, 1979, referred
to in SSA's letter above. The letter said the State did not in
principle oppose the idea, attributed to SSA, that a case should be
included if federal funding was "actually claimed," but the State
said
"it is not possible to determine the existence of federal funding on a
case by case basis" in New York City and that it therefore included
cases "for which federal reimbursement was available," which the State
described as the "only logical alternative." Id.
The gist of the foregoing is this: apparently, SSA originally wanted
the State to include cases only if federal funding was actually claimed.
New York agreed that that was best, but said it was impossible to do in
New York City based on available records. New York did what it
considered the best alternative: it included a case if the case could
have been funded under standards applicable in January 1972. A case
would be excluded if the PAD determination was not changed to AD (other
criteria were applicable which have not been brought into issue here).
SSA agreed to this approach.
(34) It should be noted also that the record contains a letter of
April 9, 1982, in which SSA agreed to give the State more time to
develop its arguments to SSA, prior to this appeal, if the State agreed
to review the cases which auditors had identified as improperly excluded
--
. . . with the same methodology used in the 100 percent review of the
APL. Folders should not be examined more intensely to find reasons for
exclusion in low payment cases than they were for high payment cases.
Otherwise, the review of the 48 cases would be biased and lead to
unacceptable conclusions.
NY appeal file, Tab 13.
2. New York's supplementary search.
The supplementary New York search of city records is dealt with in
the affidavits of Samuel Wachtel, an attorney for the New York State
Department of Social Services who appeared as one of the Counsel for the
State in this case, /12/ and John Reilly, a New York City official. NY
brief, App. II. The affidavits essentially say that the records
disclosed that the three cases apparently were eligible for federal
participation, but that records which would disclose a claim for federal
funding, if a claim had been made, show nothing. The Wachtel affidavit
explained a search of several kinds of records which, while they
confirmed PAD and AD classifications, did not show a claim for federal
funds for January 1972. Reilly--who apparently did not participate in
the record search with Wachtel--stated essentially that if Wachtel did
what his affidavit says he did, he should have found a record of a claim
if one had been made. The Reilly affidavit stated that in the absence
of the records, "there was a substantial doubt" whether federal funds
were claimed or received.
Wachtel's affidavit refers to several things Reilly told Wachtel
which are not in the Reilly affidavit. One such thing is as follows:
. . . in reality, a large number of cases were never claimed for
federal participation. One reason was that (35) the Division of Field
Audit often was not informed of the reclassification. If the caseworker
did not report the change and the Income Maintenance specialist did not
enter the change into the central computer, there was no way federal
participation would have been claimed. Even assuming the Division of
Field Audit did receive notice, there was no certainty that federal
participation was claimed. The office was short on staff (as were local
welfare centers), and the claims for federal participation sometimes
were not filed.
P10.
Based on the foregoing, we conclude that the State may not reasonably
claim that these cases were "non-federally participating" in January
1972, for the following reasons:
* It is not disputed that the cases were presumpatively eligible in
January 1972, and were classified as finally eligible. The dates of
final eligibility are not matters at issue for purposes of the present
discussion, and we have not reviewed the record concerning that matter.
* It is clear that it was the State's idea originally to judge
whether a case was federally participating by examining eligibility, and
that this affirmative agreement existed for years. Of course, the
fundamental issue is whether cases should be included or excluded, and
it might not be appropriate as a general principle to interpose a
collateral agreement on data-gathering to stop inquiry into better
quality data. But it is not clear that Wachtel produced better data,
and it is clearly unfair for the State, in view of having agreed to one
methodology (in fact, having established it), to come in after many
years, during the appeal, and propose the exclusion of three cases based
on a methodology the State rejected for thousands of other cases. There
is no reason for New York to seek exclusion of these cases other than
that they are low-value cases which depress the APL, and there is merit
to SSA's observation that bias and unfairness would arise from allowing
the State to use a different methodology only for these cases.
* The best that the affidavits show is that Mr. Wachtel found no
records to impeach his client's position. While we have no reason to
question the zeal with which he searched the records, the affidavits
still present (36) evidence of questionable substantiality. We have no
idea how complete the records are after more than a decade, whether
there are other records which should have been searched, how carefully
complete the search was, and how well the records were developed in the
first place. The negative proposition that no records were found, in
the midst of what must be many thousands of papers, is inherently more
questionable than some more affirmative evidence, of any nature, showing
that funding was not claimed. In view of the State's own original
argument that the PAD/AD determination was a good alternative measure,
we are inclined to find the latter evidence more persuasive.
* Mr. Reilly's affidavit clearly discloses that a major reason why
the records might not be present is the failure to make a claim in the
first place, due to inefficiency or staff shortages. It is highly
questionable whether Congress intended that an APL be raised, increasing
the State's ultimate federal funding, based on such failures.
Based on the foregoing, we conclude that SSA correctly determined to
include Cases 5, 18 and 23 in relation to the "non-federally
participating" issue, despite the State's subsequent evidence.
C. Other cases.
Three other cases (nos. 2, 10 and 26) also were disputed for reasons
other than those covered above.
Case 2:
The case file in Case 2 was destroyed in Chenango County, New York,
after the Original APL questionnaire was completed. The State
maintained that the file was destroyed even before the federal audit, so
that federal auditors had no substantial basis for challenging New
York's decision to exclude the case. SSA argued that a review of
questionnaire forms showed reference to the case file and documents
therein, so that it was reasonable to conclude that the auditor had
reviewed the file in making his determination. New York submitted an
affidavit of a Chenango County social worker to whom a record indicated
the auditor had spoken. The affidavit indicated that she believed the
case file was destroyed before the conversation, and that any
information she might have given the auditors without the file would
have been unreliable. NY post-hearing brief, p. A-7; SSA brief, Vol.
II, Tab 2.
(37) This case was argued to the Board only on the basis set forth
above. Our review of the information available about the case, which is
the information SSA relied on to include the case, shows that SSA
considered this case includable because of certain "shared living"
aspects. Although the case was not presented by the parties as a
"shared living" case, we have discussed it under that heading and
have
concluded that, if the facts were as presented by SSA, the State
correctly excluded the case. Therefore, since the case becomes
excludable if we assume the facts to be as SSA argues they are here, we
find it unnecessary to determine whether the case should also be
excluded because the records were lost.
Case 10:
Originally, there was a dispute concerning SSA's determination of
whether New York had included this case in its APL calculation. See,
e.g., NY brief, p. 63; TR. II, pp. 151-70. Based on further review of
documents, New York withdrew its objection to SSA's determination. NY
post-hearing brief, p. 25.
Case 26:
Here, there was a dispute over whether the correct figure for
inclusion was $221 (as the State said) or $174 (as SSA determined).
/13/ The State alleged that the $47 difference represented a proper
partial restaurant allowance. NY brief, p. 64. Alternatively, in its
post-hearing brief, the State said that if one could not determine the
actual amount of the (38) recipient's January 1972 grant, the case
should be excluded on that basis. NY post-hearing brief, p. A-9. The
Wachtel affidavit, previously discussed, contains a statement that there
were records showing that for January 1972, payments totaling $221 were
made to the recipient in this case. NY brief, App. II, Wachtel
affidavit, P5. SSA confirmed this. TR. II, pp. 424, 426, 439. SSA
argued that the recipient received an erroneous overpayment of $47 in
January 1972, which was recouped in March 1972, so that the correct
amount to be included in calculations for January 1972 was $174. SSA
brief, p. 100. We conclude that SSA was correct.
It is not disputed that the restaurant allowance was $47. TR. II, p.
427. Coincidently, half of the normal "food and other" grant--$94
a
month--is also $47. What basically is in dispute here is whether the
recipient was simply overpaid half the "food and other" grant, or
received a restaurant allowance of $47 for January.
It does not appear to be disputed that the record shows an approved
budget for January 1972 of $174. TR. II, pp. 430, 436; SSA brief, Vol.
IV, Tab 26, pp. 1-4. this amount was made up of $80 for rent and $94
for "food and other." Id.
The record shows that in December 1971, the recipient received part
of his January grant, but that it was $40 short; when the welfare
center made a correcting payment shortly after the beginning of January,
they paid him $87 (for a two-week period) with no explanation of the
difference of $47. Later in January, he received payments of $47 and
$40, so that the overall payment for January was $47 more than $174.
TR. II, pp. 437-9; SSA brief, Vol. IV, Tab 26(A), pp. 5, 6.
It is not disputed that in March, a check for $47 to the recipient
was cancelled, and it is not disputed that the cancellation was
unexplained. TR. II, pp. 424-6, 429. The record contains no document
showing that the recipient was ever awarded a restaurant grant.
We find that on balance, it is clearly more likely that the recipient
was simply overpaid an amount equal to half of the "food and other"
grant, and that the overpayment was corrected, than that the recipient
was awarded an extra restaurant allowance for the one month of January
1972. The most reasonable explanation is that the case worker
incorrectly assumed that since the recipient did not get his
semi-monthly rent payment of $40, he also did not get his semi-monthly
"food and other" payment of $47. The only reason the State has an
argument worth considering is because of the coincidence that the
restaurant payment would be the same amount. But there is clearly no
specific evidence that a restaurant grant was ever awarded, and, given
the coincidence (39) of the $47 amount that recipient routinely was paid
twice monthly, it certainly is unreasonable to infer that the $47
one-time extra amount paid in January 1972 was a restaurant grant. The
evidence that the recipient had no cooking facilities in July 1971
simply does not reach that far.
Having reached the conclusion above, the next step is easy: the
cancelled $47 payment must have related to the overpayment made in
January. Probably, it was not explained because the persons involved
did not feel compelled to dwell unnecessarily on their obvious error.
Whatever the case, it clearly is reasonable to relate the $47
cancellation to the $47 overpayment. Therefore, the amount of the grant
in January 1972 was $174, by the same reasonable logic of retroactivity
which we used in finding for the State in case 23. We therefore
conclude that SSA correctly determined that this case should be included
in the APL calculation in the amount of $174.
D. Shared living cases.
1. Overview of the issues.
The cases in dispute here are cases where there is evidence in the
The parties agreed that, in any case where living was truly shared,
"economies of scale" were present. SSA said that economies of scale
signified the presence of imputed income; New York said that this meant
that the recipient was living in the household of another. Either
finding may be a basis for excluding a case from the APL calculation
under 20 CFR 416.2085(a)(1) (1975). /15/ As (40) discussed below, the
question of whether these persons were "living in the household of
another" under section 416.2085(a)(1) is relevant to the threshhold
issue of whether we should apply any presumptions in examining these
cases.
The record shows that the parties themselves never agreed to an
underlying presumption for these cases. Instead, they agreed generally
to a set of criteria to be applied to resolve some of the cases. The
agreed-upon factors indicating the presence of economies of scale were a
friendly relationship between the recipient and the landlord or prime
tenant, sharing of rent, and sharing in the purchase and preparation of
food. In addition, for a number of New York City cases where there was
insufficient data to apply these criteria, and there was only the fact
that the recipient rented living space in another person's apartment or
house, the parties agreed to a compromise. They split the cases, on a
random basis, between the included and excluded categories. (This
compromise was referred to as the "50/50 split.")
On appeal to the Board, the State advanced a legal theory for why all
shared living cases should be determined to be cases where the recipient
was "living in the household of another," stating that perhaps it
had
gone too far in compromising its position in earlier negotiations with
SSA but that this should not preclude it from prevailing here. The
State asserted that SSA's approach to the cases was inconsistent with
SSA's own regulations for determining SSI benefits. We discuss this
issue first below because it provides a useful context for the reader in
considering the other issues in dispute. We conclude generally that the
State's reliance on the SSI regulations is misplaced and fails to
distinguish the difference in purpose behind those regulations and the
APL calculation.
We further conclude, however, that the standard which SSA employed in
analyzing these cases was inconsistent with its articulated position
that, if file documentation was inadequate to support a response
warranting inclusion, a case should be excluded. We explain below why
we think the proper treatment of these types of cases is to exclude them
when there is insufficient documentation on which to make a reasoned
decision.
We next address the question of what is sufficient as an indication
in a case file that economies of scale were present--or, on the other
hand, that instead there was an arm's-length sublet arrangement. To
answer this question, we discuss the criteria agreed to by the parties,
the dispute over how these criteria were to be applied and what they
meant, and our analysis (41) of what type of evidence is sufficient and
why. This discussion provides a context in which we examine the
individual cases in the various subgroups--New York City cases female
residing with male, upstate room and board, and insufficient data.
We conclude that SSA was correct that 4 of the cases should have been
included. With respect to the remaining 14 cases (including Case 2, not
originally a shared living case), we find that there is insufficient
data to make a reasoned decision on the shared living question. We
therefore conclude that they should be excluded, consistent with SSA's
own policy.
2. The "household of another" regulations as a basis for a
presumption of excludability.
The State took the position that there should be a presumption for
these shared living cases that they were excludable, unless there was
"affirmative evidence of a true arm's-length relationship" between
the
recipient and the landlord or primary tenant (characterized as a
"sublease"). NY post-hearing brief, pp. 6-7. The State said that the
term "household of another" should be "given its literal meaning,
so as
to embrace those situations in which an elderly or disabled recipient
lived in a residence belonging to another private individual, whether or
not that other person was a relative or a friend." NY brief, p. 43.
Pointing to SSI regulations for determining benefits, the State said
that SSA had acted inconsistently with those regulations by employing a
presumption of includability in analyzing the shared living cases in
dispute here.
In support of its argument that the cases here all were cases where
an individual was "living in the household of another" under 20 CFR
416.2085(a)(1), the State cited an early internal SSA policy memorandum
which concluded, based on legislative history, that the APL calculation
was to be an average of payments made to persons "living alone."
Memorandum dated March 7, 1974 from Assistant Director for Policy, SSA,
to Project Manager, State Programs, NY appeal file, Ex. E. The State
interpreted "lives alone" to mean "not live with others in their
household." The State then pointed to regulations governing payment of
SSI benefits to individuals living in the household of another and
receiving support and maintenance in kind. The State said that, since
these regulations applied to a person in the "household of a private
individual" (as opposed to a commercial boarding home), whether or not
that individual was a relative or friend, they covered the living
arrangements in question here. Finally, the State said that "SSA
employed an explicit presumption that recipients fitting (42) this
description were receiving in-kind support and maintenance in the form
of food and shelter." NY brief, pp. 42-43, citing 20 CFR 416.1125(b)(4)
(1978). If SSA had applied these criteria to the shared living cases in
dispute here, the State argued, every one of them would have been
excluded.
In response, SSA denied that it had applied any presumption of
includability in analyzing these cases. We discuss below the approach
that SSA said it did take. SSA further responded that the State's
interpretation of the term "household of another" the absurd, as well
as
inconsistent with the mutual agreements and understandings the parties
had reached regarding shared living cases. SSA argued that we should
not adopt the "literal meaning" of the term "living in the household
of
another" proposed by the State because that phrase is a term of art with
its own special connotation in the social welfare context. In SSA's
view, that connotation "is based upon the recognition that it is quite
possible for individuals to reside under the same roof, but in different
households, and thus not enjoy the benefit of economies of scale." SSA
brief, p. 79. SSA said that it did not automatically assume that
economies of scale were present when an SSI claimant was living in a
private household with at least one other person, but sought to develop
the situation further to determine whether the recipient was indeed
receiving support and maintenance from within the household.
We conclude that there is nothing in the "literal meaning" of the
term "household of another" or in the SSI regulations that requires
us
to presume that the living situations here are tantamount to "living in
the household of another," merely because the recipient lived under the
same roof with another person.
While SSA did not provide any direct support for its statement that
"household of another" has a specific connotation in the welfare
context, neither did the State provide any support for the proposition
that, read literally, the term "household" is equivalent to residence.
/16/ We do not agree with the interpretation advanced by the State for
the following reasons:
(43)
* As SSA pointed out, it did not consider an SSI claimant to be
living in another person's household, even if living in that person's
residence, if certain conditions existed, such as that the claimant was
liable to the landlord for rent or paid at least a pro rata share of
household operating expenses. See, e.g., SSA brief, Vol. I, Tab 36; 20
CFR 416.1125(b)(3) (1978).
* The record shows that the State itself, in administering its AABD
program, did not equtate living under the same roof and living in the
same household. Indeed, in many of the cases in consideration here,
even though the recipient lived in an apartment or house owned or rented
by another person, the State concluded that the recipient constituted a
single person household. (We discuss the effect of this below.)
* While we would not necessarily hold the State to the terms of its
agreement with SSA concerning shared living situations if we found it to
be inconsistent with the statute and regulations, we think that the fact
that the State did not contend during its negotiations with SSA that all
of these cases should automatically be considered "household of another"
cases shows that the State understood the difference between the
concepts of residence and household.
Moreover, we do not think that merely because the SSI benefit
regulations presume that a person living in the household of another is
receiving in kind support and maintenance, we should presume that these
shared living cases are excludable. The presumption in the regulations
presupposes a conclusion that there is a "household of another"
situation. One issue here is whether such a conclusion is warranted.
SSA has acted consistently with the presumption in the regulations by
recognizing that shared living may result in income to the public
assistance recipient.
There is a more fundamental flaw in the State's argument, however.
Even if we were to find that SSA employed in its benefit regulations a
presumption that living situations like the ones we are confronted with
here were "household of another" situations, this would not require
that
we apply the same presumption here. The circumstances are entirely
different. Where there is reason to (44) believe that an applicant for
federal SSI benefits is in a living situation which may give rise to
income it is appropriate to employ a presumption that such income
exists, absent convincing evidence to the contrary. The applicant
properly has a burden to show that his needs warrant the granting of a
higher benefit amount. Here, it is the State which is seeking federal
reimbursement, and the same rationale simply does not apply.
Thus, we conclude that the mere fact that a recipient was living in
another person's residence does not afford a basis for excluding the
case as "living in the household of another."
This conclusion does not, however, mean that we agree with the
implication in SSA's arguments that the recognition in the regulations
that persons could live under the same roof and still constitute
separate households somehow supports SSA's position that these were in
fact separate households, and, therefore, the recipients lived alone.
Under the parties' agreement, a conclusion that recipients such as these
lived along would be made only after applying the agreed-upon factors.
Moreover, the presumptions in the SSI regulations are not wholly
irrelevant to the issues presented here. Underlying them is a
recognition by SSA that when a person rents a room in a private home, as
opposed to a commercial establishment, "it is less likely that the
alleged landlord and the individual/tenant are actually separate
households (two units) instead of a sharing household (one unit)." See
SSA Handbook, section 00835.120(c)(3), quoted in NY post-hearing brief,
p. 17, n.1. The view that there is some likelihood of sharing in these
types of living arrangements is further supported by testimony,
presented by the State's witness who was experienced in casework, that,
typically, sharing would take place in these situations. TR. I, p.
101.
Finally, we note that the concept of "shared living" embodied in
the
parties' agreement is broader than the concept of "living in the
household of another." For example, "pro-rated rent" was an agreed-upon
basis for determining that a recipient was not living alone; this
reflected the fact that, if an AABD recipient was "living with" a
relative or friend (even if in a separate household), the State would
not pay actual rent for that recipient but only a pro-rata share of the
total rent on the apartment or house. See NY appeal file, Vol. II, Ex.
N, p. C-6. Further, both parties connected sharing with economies of
scale. SSA said that economies of scale signified imputed income.
Although living in the household of another can give rise to imputed
income (see TR. II, p. 313), SSA never contended that one could find
economies of scale (a form of imputed income) only if one could also
conclude that the recipient was living in the household of another.
Since (45) a case is to be included in the APL calculation under 20 CFR
416.2085(a)(1) only if the recipient was not living in the house-hold of
another and had "no other income," we view our inquiry here to encompass
not only the question of household composition but also the question of
imputed income through other forms of sharing.
3. SSA's approach to shared living cases.
Having concluded that the SSI benefit regulations do not directly
apply here, we examine in this section what SSA's approach was to shared
living and how to treat cases in which there was insufficient data.
SSA argued that the State had improperly attributed to SSA the
position that all shared living cases were presumed to be includable
unless evidence was found justifying exclusion. Instead, SSA described
its position as follows:
SSA's basic premise throughout the course of the 100% case review was
that for each decision made either to include or exclude a case, there
was required adequate evidence in the welfare case file substantiating
that decision. If a response on the APL Survey Questionnaire was
developed warranting exclusion, the case was required to be excluded;
if a response was developed warranting inclusion, then the case file
reviewer was to continue to the next question; and if documentation
adequate to formulate a supportable response was not found in the case
file, then pursuant to mutual agreement, the reviewer was to circle the
question and exclude the case.
SSA brief, p. 82 (emphasis in original, footnote deleted).
SSA also stated: "Cases which were truly indeterminable were
excluded." SSA brief, p. 83.
SSA did not adhere to this position in a consistent manner, however,
either in other statements that it made during the course of this appeal
or in is actual analysis of individual cases. For example, SSA also
said that "if there is no indication in the file that payment of rent or
food purchase and preparation were being shared, then there is no reason
to believe that such sharing was indeed taking place among individuals
who shared living arrangements. Thus, absent any indication of sharing,
or any other reason to exclude, such a case was to be included." SSA
post-hearing brief, p. 14.
(46) Saying that there is no indication of sharing is far different
from saying that there is indication of no sharing, that is, that there
is evidence in the file sufficient to substantiate a decision to include
the case. That SSA confused these two standards is also apparent from
comparing the conclusions reached by the SSA analyst after he examined
the State's arguments in response to the audit findings. The analyst,
in agreeing with the State that two cases should have been excluded
(rather than included as the auditors said) gave as the reason
"insufficient documentation." SSA appeal file, Vol. III, Tab 15. Yet,
in other cases the SSA analyst agreed with the auditors that the cases
should be included, stating merely that there was no specific mention in
a case file that sharing took place in the purchase and preparation of
food, without making any specific finding that there was evidence to
support a conclusion that no sharing was taking place.
Based on our analysis of the record and the purposes underlying the
APL calculation, we conclude that it is proper to include any of these
cases only where there is some information in the file to substantiate a
finding that no sharing is taking place. Thus, where there is
insufficient data on which to make a reasoned decision concerning
sharing (truly indeterminable cases), a case should be excluded. This
approach is consistent with what SSA said it did and with SSA's
agreement that a reviewer was to circle any question on the APL
questionnaire and exclude a case if the reviewer could not find
documentation adequate to respond to that question. The following
reasons also support our adoption of this approach:
* It is consistent with the parties' agreement on shared living cases
(discussed in detail below). Under that agreement, the mere absence of
factors which would justify exclusion on the basis of sharing was not
given as a reason justifying inclusion.
* It is consistent with how the auditors treated cases where they
determined there was insufficient documentation to make a reasoned
decision. See SSA brief, p. 33.
* It is consistent with the parties' agreement to exclude any case
where the case file was missing.
More important, we think that this approach recognizes that, under
the statute and regulations, the relevant inquiry is whether the
criteria for inclusion in the APL calculation have been met. The (47)
APL figure is to be derived from the average money payment to persons
who meet those criteria. Thus, while the process of determining the APL
involved identifying factors which justified excluding certain cases,
the purpose was to isolate those cases that met the rather restrictive
criteria for inclusion. /17/
This is not to say that the State, in advocating exclusion of a case
on some specific basis, does not have to substantiate its decision, but
merely that, in cases such as these, where the very living arrangement
gives rise to some likelihood that the inclusion criteria are not met,
it is appropriate to exclude them unless there is a basis for concluding
that in fact there was no sharing. /18/
Of course, the practical effect ofall of this for this group of cases
is the same as if we employed a presumption of excludability. We are
not, however, adopting the State's presumption that these recipients
were living in another person's household; we are instead excluding
cases which are indeterminable on the question of sharing because we
cannot affirmatively say that these recipients were not living in the
household of another or otherwise enjoying economies of scale from
shared living.
(48) We next discuss the parties' agreement concerning what factors
are relevant for shared living cases, as well as the dispute over how to
interpret that agreement and what kind of evidence is sufficient to show
that any of the factors existed.
4. The parties' agreement on shared living.
(a) What the agreement was.
Included in the "criteria file" used by the auditors is a set of
coding instructions which the State gave to its data collectors to
assist them in determining how to fill out the APL questionnaire in
cases potentially involving shared living. Criteria file, C-28-1,
19-21/25. These coding instructions had been transmitted to SSA by
letter dated March 22, 1976. The SSA analyst who developed SSA's
position on the individual cases disputed here (and who testified at the
hearing) viewed these coding instructions as the embodiment of the
parties' agreement on how to treat these cases. TR. II, pp. 334-7. The
State contended that the agreement actually consisted of two letters: a
letter dated October 30, 1975 from the Deputy Commissioner, Division of
Income Maintenance, NYDSS, to the Regional Planning Officer, Region II,
HEW (NY brief, App. I, Ex. F) and a letter dated February 9, 1976 from
the Regional Planning Officer to the Deputy Commissioner (NY brief, App.
I, Ex. G). The State's position on this is generally supported by the
record. The March 22, 1976 letter transmitting the coding instructions
to SSA states specifically that it is based on the letters of October 30
and February 9. This finding is relevant to some questions of
interpretation of the coding instructions which we discuss below.
The October 30 letter from the Deputy Commissioner set forth her view
of what the parties had agreed to in a prior meeting, stating that the
following types of cases were among those which were excludable:
* Where the public assistance recipient lives with a person
identified in the case folder record as a friend unless there is
information that sharing is not taking place.
* Where the public assistance recipient lives with another person,
not identified as a friend in the case folder, but where there is clear
evidence that the rent is shared.
* Where the public assistance recipient is living with another
person, not identified as a friend (49) but there is indication that
they are taking their meals together or there is some indication that
food is purchased and prepared in common.
and that the following types of cases were includable:
* Where the public assistance recipient is subletting, e.g., "client
rents a room from the prime tenant, has kitchen privileges."
* Where the public assistance recipient lives with a friend or other
person but there is a full restaurant allowance and no indication that
the rent is shared. If a partial restaurant allowance is made, the case
may be excluded if it meets one of the exclusion categories . . .
above.
NY brief, App. I, Ex. F.
SSA's initial response to the October 30 letter was to question
excluding a case where the file of an individual living with a friend
lacked documentation to show that rent was shared or prorated. Letter
of November 19, 1975 from Regional Planning Officer to Deputy
Commissioner, NY brief, App. I, Ex. H. However, the February 9, 1976
letter responded to a subsequent request by New York for confirmation of
SSA's position on shared living arrangements by stating, "The criteria
set out in your 10/30/75 letter for inclusion/exclusion of cases is
acceptable to us." NY brief, App. I, Ex. G.
The coding instructions merely set out the possible combinations of
factors listed in the October 30 letter, indicating to the data
collectors how to code the APL questionnaire form, and defining the
terms used. /19/ Yet, as discussed below, the SSA analyst applied these
instructions in a rather mechanistic way, without considering the
underlying agreement or the purpose behind the APL calculation.
(50) With respect to the "New York City" and "female residing
with
male" cases in dispute here, the State's primary argument was that there
was evidence in the file from which it was reasonable to infer that the
relationship between the recipient and the landlord (or the primary
tenant in an apartment) was a friendship or other dependency
relationship and, therefore, these cases should be excluded under the
criteria agreed to by the parties. In advocating this position, the
State challenged SSA's interpretation of the agreement in two respects:
the quality of evidence needed to establish a friendship; and the
effect of a finding that a friendship existed where no other factor was
present to justify either including or excluding the case. The State
also questioned the manner in which the SSA analyst applied part of the
coding instructions (section II.C.). We discuss these issues next.
(b) SSA's view of "friend" cases.
At the hearing, the SSA analyst testified that he would conclude that
there was a friendly relationship between a recipient and the landlord
or primary tenant only if there was "objective evidence" in the file
of
that relationship, that is, a notation in the file that "the landlady
and the individual are friends or something to that effect." TR. II, pp.
357-62. In its post-hearing brief, SSA pointed to the NY Deputy
Commissioner's October 30, 1975 letter, referring to the phrase
"identified in the case folder as a friend" (translated in the coding
instructions to "specifically identified as a friend in the case
folder"). SSA said that the parties had agreed "that the actual word
'friend' must appear in the case folder." SSA post-hearing brief, p.
19. SSA did not provide any evidence that this was the parties' actual
intent, but explained that this "satisfied the parties' desire to be
objective and to avoid the kind of subjective and arbitrary analysis
that would result if many different case reviewers attempting to draw
distinctions or inferences about human relationships from a few sketchy
facts contained in the welfare case folders." SSA post-hearing brief, p.
19.
The State said that, in refusing to draw inferences with respect to
the existence of friendship (or any other agreed-upon factors), SSA had
employed an unjustifiably narrow standard of file review. The State
argued that nothing in the agreement of the parties suggested that
reasonable inferences could not be drawn from evidence in the files, and
that it was not justified to require explicit notation of the factors
since the files were developed many years before the APL criteria were
established. The State further pointed out that SSA itself had drawn
inferences to support judgments on the includability of some cases.
(51) We agree with the State that SSA took too narrow a view of what
the agreement required. We think that a person can be identified in a
case file as a friend even though there is no notation using the word
"friend" or some equivalent word. SSA did not support its assertion
that the parties had agreed to require such a notation, but merely
speculated that this would avoid the problem of having different
reviewers reach different results on the same facts. But there is
nothing in the record to indicate that the reviewers in fact required
the kind of evidence concerning friendship that the SSA analyst was
looking for. Moreover, as we discuss below, we do not agree with SSA in
general that findings can never be based on reasonable inferences.
On the other hand, we think that, in the context of the State's AABD
program, the mere inference that a friendly relationship existed is not
as definitive on the question of sharing as is evidence that a
caseworker made a specific determination that a friendship existed. The
State's witness who was experienced in casework described in his
affidavit how a friendship might develop between a recipient and a
family from which the recipient rented a room, stating that this would
not normally be reflected in the case file. The reason he gave why this
kind of relationship would not be labeled as a "friendship" by the
client was that then a pro-rata limitation on rent would have applied.
NY brief, App. II, Burdick Affidavit, p. 10. This statement refers to a
State policy for its AABD program to pay only a pro-rata share of rent
for any recipient who was residing in the home of a relative or friend.
See NY brief, App. II, Ex. N, p. C-6.
Indeed, many of these files contain forms with a space to check if
the recipient was living with friends. Thus, the question of friendship
was germane to the State's AABD program and could limit the amount of
the rent allowance. Considering this factor, we think that a finding of
friendship based on inference must be based on evidence which gives rise
to an inference that a friendship did exist strong enough to overcome
any evidence that the case was treated as though no friendship existed.
While we recognize the disincentive a recipient would have to
acknowledge friendship, and the possible reluctance of a caseworker to
raise the issue (see NY brief, App. II, Burdick Affidavit, P10), the
failure to consider this factor could have resulted in the State
allowing more for rent than if the factor had been considered. Since a
friendship might still give rise to some other less specific economies
of scale, it is relevant to our inquiry, but we give this factor less
effect where there is evidence that the rent was not pro-rated as a
result of the friendship.
(52) We also note that, in its analysis, the State tried to equate a
"dependency relationship" with friendship. We could find no indication
in the record that a dependency relationship could have the direct
consequences for the State's payments to a recipient that a friendship
would have and the parties' agreement did not explicitly recognize such
a relationship as a factor in determining whether living was shared.
Yet, SSA did not directly challenge the State's assertion that such a
relationship could give rise to economies of scale (possibly through
food sharing), and it is reasonable to presume that an aged or disabled
recipient might develop a dependency on a landlord or prime tenant
precisely because that person was providing some assistance. On the
other hand, it is also possible that the landlord or prime tenant could
take advantage of a dependent recipient (for example, by charging them
higher rent than was reasonable), whereas this would not be generally
expected of a friend. Thus, we think that a dependent relationship
should not be treated as equivalent to a friendship here but that a
dependent relationship may provide some support for excluding a case
where there is other evidence that sharing was, in fact, taking place.
(c) The effect of a finding of friendship.
As mentioned above, the State also challenged the SSA analyst's
approach to cases where a friendly relationship existed but there was no
basis for deciding whether the other factors requiring inclusion or
exclusion existed. The analyst thought that such cases would be
included because the coding instructions said to code such a case as an
"E-6" on the APL questionnaire, and that code is listed on the
questionnaire as a "continue" condition--that is, it indicates that
the
collector should continue to fill out the form to determine whether
there is a basis for excluding the case (and, if there is no other basis
for excluding the case, should ultimately include the case). TR. II,
p. 345. What the coding instructions say, however, is that if a client
is initially identified as living with friends and it cannot be
determined whether or not rent is shared and there is no restaurant
allowance, "Code such a case E-6 and stop." Criteria file, C-28-1,
19-21/25 (emphasis added). A "stop" condition meant that a case was
to
be excluded. The Deputy Commissioner's October 30 letter setting out
the parties' agreement said that a case would be excludable "Where the
public assistance recipient lives with a person identified in the case
folder record as a friend unless there is information that sharing is
not taking place." That the SSA Regional Planning Officer understood the
implications of this statement for cases where there is no information
that sharing is not taking place is clear from the November 19 letter
(53) initially responding to the October 30 letter and objecting to this
statement. Moreover, the copies of the coding instructions in the
record contain markings which were made by an SSA official and which
indicate that the official making the markings knew that these cases
would be excluded. See also SSA brief, p. 85. Thus, we conclude that
the proper treatment, under the parties' agreement, where (1) a client
is living with a person identified as a friend, (2) it cannot be
determined whether rent is shared and (3) there is no indication of no
sharing, is to exclude the case.
(d) How the SSA analyst applied section II.C. of the coding
instructions.
The State also challenged during the hearing the manner in which the
SSA analyst applied the coding instructions. Specifically, the analyst
based many of his determinations of includability on section II.C. of
the coding instructions. That section states:
II. Where a client is initially identified as living with another or
others and:
* * *
C. The case does not involve the sharing in the purchase and
preparation of food (i.e., 0-2) and the client rents a room from another
or others who may or may not be the landlord. Code such a case E-4 and
continue; . . .
The State's major problem with the analyst's application of this
instruction was that the analyst always applied it whenever there was no
specific notation that the client was living with a friend and, as
discussed above, the State thought that such a notation was not
required. (Section I of the coding instructions applied where the
client was initially identified as living with friends.) Section II
should apply where the client is residing with someone "who is not
specifically identified as a friend," since that is the definition of
"other" in the coding instructions. But, as we have stated, this does
not mean the mere absence of a notation containing the word "friend"
or
its equivalent.
The record shows a more fundamental problem with how the SSA analyst
applied section II.C., however. The analyst made decisions to include
cases after finding that the case met the description in II.C., without
considering whether there might be another basis for exclusion, i.e.,
shared rent. Under section (54) II.D., if a case involved the sharing
of rent, it was to be excluded. /20/
Only instruction II.A. applies to a case where rent sharing is
unknown to give a "continue" answer on the APL questionnaire with
respect to the recipient's shelter living arrangement (Question G).
Section II.A. says to continue such a case if it "does not otherwise
involve rental of a room."
The SSA analyst's failure to address the issue of shared rent is
significant because, the some of these cases, the State's original basis
for exclusion was shared rent and even the auditors initially said that
there was insufficient data to determine whether rent was shared. For
these cases, since they all involve rental of a room, the mere lack of
an affirmative finding that rent was shared was not a sufficient basis
for including a case (see general discussion above).
We also note that the SSA analyst did not, in most cases, provide any
support for his findings that there was no food sharing. Rather, he
thought that section II.C. of the coding instructions automatically
applied to include a case if there was no specific notation that food
was shared, other than a mere reference to the recipient having the use
of kitchen facilities. While we agree, (55) and New York acknowledged,
that mere use of a kitchen is not sufficient to support affirmatively
the conclusion that a case should be excluded on the basis of food
sharing, the parties' agreement required only an "indication that they
are taking their meals together" or "some indication that food is
purchased and prepared in common," regardless of whether friendship
existed. NY brief, App. I, Ex. G.
While the SSA analyst's misinterpretation of the agreement on these
points is understandable given the complexity of this area and the
complicated evolution of the parties' agreement, we think that this
misinterpretation led him to include cases without considering all of
the relevant evidence in the files. /21/
5. How we considered the evidence, in general.
(a) The evidentiary standard SSA applied, in general.
In analyzing these cases, SSA took the position generally that
application of the relevant factors had to be based on "objective
evidence" in the case file. The State objected to this strenuously,
arguing that there was no reason why a decision could not be based on
reasonable inferences and that relying on inferences was often necessary
because these records were developed for the State's AABD program,
rather than for the specific purpose of calculating an APL.
We agree with SSA, in principle, that findings should be based on
"objective evidence." But we think that, in practice, SSA applied
too
rigid a standard concerning what constitutes objective evidence of
sharing. Where an inference is based on specific facts which are noted
in a case file, and is an inference that any reasonable person would
draw from those facts, a finding based on that inference does have a
degree of objectivity. Requiring that subjectivity be totally
eliminated, as SSA advocated, is (56) unwarranted under the
circumstances here, where we are examining records compiled for a
purpose other than our specific inquiry. /22/ The State's expert
testified that recording by caseworkers in the AABD program was apt to
be light (TR. I, p. 103), and the files we have before us bear this out.
Fair review of such files is virtually impossible if no finding can be
made except one based on a specific file notation.
Yet, it is important to note here that (contrary to what the State
would have us believe) not all of the types of information which might
help us to decide these cases were totally irrelevant to the State's
AABD program. While some of them were, some were not. The State's
expert testified that a caseworker would be unlikely to record in a file
information about whether a recipient had access to the full apartment
in addition to kitchen privileges, because "that was not germane to the
basic eligibility of the case nor the payment." TR., p. 103. This
testimony was uncontradicted. On the other hand, the reason he gave
concerning why he would not generally expect a file to contain
information on sharing of food preparation and purchase was quite
different:
The grant of $84 was a grant for a one person household, and by
recording the sharing of purchasing and common preparation of food, it
would have endangered for some future audit the fact that $84 was being
authorized for this individual.
TR. I, p. 103.
In other words, food sharing was germane to the determination of
whether an applicant for assistance constituted a single person
household or should be considered part of a larger household. Indeed,
the manuals for determining AABD benefits in New York in this time
period indicate that the number of persons in a household determined, on
a sliding scale, the amount of the grant for basic needs (including
food), with $84 the highest amount given. NY brief, Vol. II, Ex. N, p.
C-6. Thus, New York did officially take food sharing into account, at
least in some circumstances, by awarding a lower basic grant amount.
(57) In this respect, the question of food purchase and preparation
in common becomes like the question of friendship. If we accept the
State's reasons as to why there would not have been a specific notation
of food sharing in an $84 grant case or of friendship in a case where
rent was more than the recipient's pro-rata share, we must also consider
the other side of the coin: in these cases, the State acted as though
the recipient constituted a single person household or that no
friendship existed, or both. Thus, any contrary finding that there was
sharing of food or friendship would have to be based on an inference
that was sufficiently strong to overcome the negative inference arising
from the State's own treatment of the case. /23/
The same analysis applies in a more general way. A reasonable
inference that sharing was taking place is not sufficient to exclude a
case on the basis of any of the agreed-upon factors if there is also
information in the file from which it is reasonable to infer that no
sharing was taking place. The inference of sharing must be strong
enough to overcome any inference to the contrary. /24/
(58) In view of the discussion above, we note here that we do not
agree with SSA that an inference of no sharing arises from the mere lack
of a specific notation that one of the agreed-upon factors for exclusion
exists. Despite the SSA analyst's assertion that he based his
conclusions on "facts" in the files, it is clear from examining the
reasons he gives for including individual cases that generally it was
the absence of any specific notations concerning friendship and food
sharing that led him to apply section II.C. of the coding instructions.
In other words, he frequently inferred solely from the absence of a
specific notation about a condition that the condition simply did not
exist. We find nothing in the parties' agreement or in the underlying
purposes of the APL calculation which warrants this type of inference.
(See our discussion above on why we think a conclusion of no sharing
must be substantiated by evidence which outweighs any evidence that
there is sharing.)
Finally, while we agree with the State that it is acceptable (and,
indeed, necessary) to base decisions about sharing on reasonable
inferences, we note that we do not consider an inference to be
reasonable if it is based on pure speculation or a highly attenuated
analysis. As we discuss next, however, we do give some weight to
testimony by the State's caseworker in considering whether an inference
is reasonable.
(b) The weight we give to the State expert's testimony.
The final question which we must address before analyzing individual
case submissions regards the weight we should give to the testimony by
the State's expert on AABD casework. SSA did not present any expert
witness to rebut his testimony on the cases but challenged it on the
grounds that it was based on his own subjective reaction, rather than
objective application of the agreed-upon factors. /25/
(59) As we have discussed above, however, this is not a situation
where subjectivity can be avoided altogether. Moreover, the expert's
reaction was based on his extensive experience in working with New
York's AABD cases. SSA itself recognized during the course of the
parties' negotiations that a caseworker might be better able to make a
determination about these cases than an inexperienced reviewer.
Criteria file, C-28-1, 13/25.
SSA did not allege any bias on the witness's part, although SSA did
challenge his qualifications in two respects. SSA pointed out that he
did not have any experience in real estate appraisal, or similar
experience, which would qualify him to make judgments about whether rent
was low. There is some merit to this challenge. But we find on balance
that his background and testimony show that he had a fair amount of
knowledge of rents actually paid by recipients. Further, SSA presented
no evidence to the contrary.
SSA also questioned the expert's knowledge of conditions in upstate
New York (relevant for the category of cases called "upstate room and
board"), since he had worked exclusively in New York City. We agree
with SSA that the witness would not have particular expertise in upstate
cases, except to the extent they dealt with matters which were uniform
in the AABD program throughout the State.
Thus, we give some weight to the expert's testimony in determining
whether the inferences drawn by the State were reasonable, except that
we give less weight to his evaluations of rental value and no weight to
his testimony regarding matters that were peculiar to upstate New York.
Finally, we do not find the State's analysis defective merely because
it went beyond the scope of the factors specifically used in the coding
instructions. We do not think that we are limited to applying the
coding instructions in the mechanistic way the SSA analyst did, nor do
we think the underlying agreement justifies the analyst's view that the
factors could be established only if a case file specifically noted
their existence. Many of the facts discussed by the caseworker are
relevant to the question of whether the agreed-upon factors existed and,
therefore, to our consideration.
6. Summary of our conclusions on how to analyze shared living cases.
For all of these cases, by operation of SSA's regulations and the
parties' agreements, there is some likelihood of sharing which arises by
the very fact that the recipient was living in (60) the apartment or
house of another. This, in effect, is a threshold over which all of
these cases have passed to reach the disputes discussed below. Thus,
based on our consideration of the approach SSA said it took and the
underlying purposes of the APL determination process, we have decided to
include a case only where there is information in the file to
substantiate a finding of no sharing and that information gives rise to
an inference sufficiently strong to overcome any inference of sharing
from other information in the file. Stated differently, and simply, an
indeterminable case should be excluded.
Since a finding that a case is indeterminable justifies exclusion of
a case, we have not found it necessary in every case examined below to
decide affirmatively whether some agreed-upon factor for exclusion is
met; we need only find that the evidence of no sharing does not
outweigh the evidence of sharing.
We have also concluded that:
* Findings may be based on reasonable inferences drawn from specific
facts noted in a case file.
* An inference is not reasonable if based on pure speculation or a
highly attenuated analysis.
* The State witness's testimony carries some weight in determining
whether the inferences the State drew were reasonable, except that his
evaluations of rental value carry less weight and his testimony
regarding matters peculiar to upstate New York carries no weight.
* The fact that the State awarded a recipient an $84 (single person
household) grant is some evidence that there was no sharing in the
purchase and preparation of food.
* If the State determined that a recipient was living alone rather
than with a friend, this is some evidence that rent was not pro-rated.
* Evidence that there was a dependent relationship between a
recipient and others does not carry the same weight as evidence of a
friendship.
* No inference of no sharing can be drawn from the mere absence of
specific mention of the agreed-upon factors.
(61) * A case cannot be included based on a finding that there was no
food sharing (i.e., that section II.C. of the coding instructions
applied), unless there is also evidence to substantiate a finding that
rent was not shared.
With these conclusions in mind, we next turn to our examination of
the individual cases disputed on the basis of shared living.
7. Analysis of the 18 cases.
In this section, we analyze the individual cases in the shared living
category. For this purpose, we use the subgroups used by the parties
(except that the subgroup the State called "shared living" we call
"New
York City cases"). We introduce each subgroup by discussing the general
arguments about that subgroup and then analyze the individual cases in
that subgroup to the extent necessary.
(a) New York City cases (nos. 14, 18, 6, 19, 21).
These are all cases of residents of New York City who received $84
grants but rented a room in an apartment or house owned or rented by
another person. The State's expert testified that, based on his
experience with living arrangements typical of AABD recipients in New
York City, he would conclude that these cases reflected shared living.
The expert said that the manner in which most people found their way
into apartments of others was "because there was an introduction of that
person to the primary tenant in the apartment either because they had
known them for some period of time or there was someone else who knew
them and was recommending them." TR., Vol. I, p. 101. He described a
typical scenario to support his conclusions regarding Case Nos. 14 and
18 (discussed in detail below):
These cases appear to represent a fairly common situation where an
individual AABD recipient (often an elderly woman) lived with a family.
While the recipients would have their own bedrooms in many such cases,
they would share the common living areas with the family. The rent paid
by the recipient would help the family to defray a portion of the
household expenses. In addition, the recipient would often help out by
doing light cleaning and babysitting for the family. From the
recipient's point of view, she would have people to help her during
periods of ill health or when she needed transportation to the clinic or
welfare center. Thus, even if no relationship existed at the outset,
the (62) recipient would tend to become involved in the household
affairs and receive certain benefits from living within another person's
household.
NY brief, App. II, Burdick Affidavit, p. 10.
With respect to the three other cases in this subgroup (Case Nos. 6,
19, and 21), the witness stated: "The cases reflect a common situation
where a recipient, who would be unable to maintain herself or himself
without the assistance of others, was being helped by the family he
lived with." Id., p. 11. He acknowledged that these relationships were
"more probably" based on dependency rather than friendship in the
normal
sense of that term, but stated that these are certainly not examples of
individuals maintaining separate households. Id., p. 11.
We give some weight to the expert's testimony that recipients could
benefit from the kinds of relationships described, since we do not agree
with SSA that it is based on pure speculation. The witness had
considerable experience in working with AABD recipients in New York
City. However, in determining whether the described scenarios actually
fit the cases here, we have not merely deferred to the expert's
judgment, but have gone further and examined the record. The fact that
such scenarios were typical adds little to the notion that sharing is
likely in these kinds of arrangements, unless we can find in individual
cases that the facts fit the scenario.
Case 14:
This case involves a woman (probably in her seventies, although she
apparently had some difficulty proving her age), renting a room in a
6-room apartment from a Mrs. W., who is variously referred to as the
landlady or primary tenant. The recipient had lived in the apartment
since August 1966. The first reference to the living situation says she
had "full use of the kitchen and bathroom." SSA brief, Vol. III, Tab
14,
p. 25. There is a notation in the file made in 1967 that the recipient
was dissatisfied with her living conditions, stating that she preferred
to cook on a hot plate in her room because the landlady acted "very
funny" about her using the kitchen. Id., p. 26. However, there is no
further mention of this and notations in later years indicate that she
had use of the kitchen and access to the rest of the apartment,
including the living room. The State's expert testified that this was
beyond what you would expect if someone had merely rented a furnished
room. TR. I, p. 104. The rent for the room was initially $48 a month,
and there was no increase between 1966 and 1972 except to $12 per (63)
week, which is nominal. The State's caseworker noted that this was
"rather unusual in a time of escalating rents." Id. n26
The State's expert also concluded that the recipient was providing
some housekeeping services in the apartment. Based primarily on these
three factors (access to the full apartment, nominal rent increase, and
provision of services), the State's expert testified that he would
"suspect that were was a relationship that developed over a significant
period of time," warranting the conclusion that the recipient lived in
the household of another. Id. The State also based exclusion on a
finding that the recipient lived with Mrs. W.'s family. Although there
is some reference in a caseworker's notes in 1969 to two other tenants,
notes made in 1966 indicate that the other occupants were Mrs. W.'s
husband and daughter and that the recipient was the "only boarder."
SSA
brief, Vol. III, Tab 14, p. 25.
On cross-examination, the State's expert conceded that perhaps a
finding that the recipient was providing housekeeping services was not
warranted, when it was pointed out to him that a reference to her doing
"her housework" at the time of a home visit by a caseworker is followed
by a reference to her cleaning her room. See id., p. 22. The expert's
testimony was consequently flawed by his possible misreading of the
file. /27/
(64) On the other hand, SSA's analysis of this case was very
conclusory. In response to the State's statement that it could not be
determined whether rent was shared because the file did not indicate
total rent, the SSA analyst stated the "total rent is not needed because
of the subletting arrangement." Id., p. 1. The analyst cited audit
workpapers which said that a sublet was an "include" case, but did
not
support his finding of a sublet by citing to any specific evidence in
the file. /28/ The SSA analyst also found that there was no sharing in
the purchase and preparation of food, but based this finding on the lack
of a specific notation in the file that there was food sharing.
We conclude that the case should be excluded. There is very little
in the file to indicate that no sharing was taking place. The recipient
was receiving the $84 grant for a single person household, but there is
no documentation to show that rent was not shared. Although the
recipient apparently preferred to cook in her own room in 1967, there is
no indication that this situation continued, and the fact that she
remained in the apartment and gained increasing access seems to indicate
that the problem between the recipient and Mrs. W. over cooking was
resolved.
On the other hand, in addition to the rent payment remaining
substantially the same and the recipient's increased access to the
apartment, there is some further support in the record to show that the
State's scenario was reasonably applied to this case. Mrs. W. had
agreed to wait for the rent in September 1966, when the recipient could
not pay it right away although the recipient had just moved in in
August. Id., p. 25. Moreover, although the recipient initially got
rent receipts from Mrs. W., she apparently stopped doing so in 1969,
which could indicate increased trust. Finally, the recipient was
described by several caseworkers as being a friendly person, who said
she got along with the other tenants quite well. None of these facts
alone would show conclusively that the recipient and Mrs. W. were
friends, or that the recipient was living in Mrs. W.'s household, but,
taken together with the State witness's testimony, at the very least
they balance the slight indications that no sharing was taking place.
(65) For the reasons we have discussed above, this analysis warrants
the conclusion that the case is indeterminable and should be excluded on
that basis.
Case 18:
This recipient was a woman in her late fifties who suffered from
Bell's Palsy, as well as psychological disorders. She rented a
furnished room in the 5-room apartment of Mr. and Mrs. R., paying $15
per week. She had been living there at least since August 1969.
The State submitted that the recipient was living in the household of
Mr. and Mrs. R., arguing that the following supported that conclusion:
the recipient had "access to the entire apartment"; the case file
records that she "does a little cleaning around the house"; she did
not
speak or understand English and suffered from nervousness and other
psychological problems, yet had lived with the R.'s for at least two and
one-half years, so one could conclude that she had established an
amicable relationship with the R.'s. NY brief, App. I, Tab 18, pp. 1-2.
In addition, the State expert found this case to be similar to Case 14
above.
The SSA analyst's decision was conclusory, based primarily on the
finding that there was "no notation in the case that she shared in the
purchase and preparation of food." SSA appeal file, Vol. III, Tab 18, p.
2. The analyst did not cite to anything in the file to support his
findings.
We conclude that the case should be excluded. There is very little
in the file to substantiate a conclusion of no sharing. The recipient
did receive an $84 grant, and one form in the file identifies her as
"head of household." Id., p. 23. Yet, the same form contains an entry
of a name under "other in household," and what appears to be "St"
under
"relationship." There is nothing in the file which sheds any light
on
why this entry was made.
The file is less than definitive regarding facts which might justify
the affirmative conclusion that sharing was in fact taking place. The
State did not state why it had found that the recipient had access to
the entire apartment, and the only indication we could find was that a
caseworker noted that the recipient "lives with" Mr. and Mrs. R. Id.,
p.
39. Other than the time period we know the recipient had rented the
room (which was shorter in this case than in Case 14) and the mere
possibility she did some cleaning around the apartment, there is little
to enlighten us on her relationship to the R.'s. The facts that she had
language and other problems does not necessarily support a finding of an
(66) "amicable relationship" with them as the State suggested. However,
there is nothing to suggest the State's scenario did not fit this case.
Moreover, there is no indication that rent was not shared. Indeed, the
PDC and Arthur Young found that rent was shared, and the auditors
initially found that there was insufficient data to determine whether
rent was shared since total rent was unknown.
Thus, although we cannot agree with the State that the data is
sufficient to conclude definitively that the recipient was living in the
R.'s household, the evidence of sharing is certainly sufficient to
balance the slight indications of no sharing. This justifies
classification of the case as indeterminable and warrants exclusion for
the reasons we have articulated above.
Case 6:
This recipient was a male in his mid-twenties who had psychological
problems. In January 1972, he was living in a 2-room furnished
accommodation on the second floor of his landlady's house (she lived on
the first floor). He was sharing kitchen and bathroom facilities with
another tenant. His rent was $20 per week.
The State originally excluded the case on the ground that the
recipient shared the kitchen and bath with the other tenant, but later
conceded that this, by itself, was not a valid basis for exclusion on
the ground of shared food. Before us, the State relied on file
notations which state that the landlady's daughter, who accompanied the
recipient to the welfare office, was a "friend," and that the "landlady
seems to be very protective of him and looks after him - worker doubts
that client could adequately care for himself without this help." The
State also mentioned, as additional factors, that in July 1970 the
recipient had not paid his rent for over a month because he was
hospitalized, and that the recipient was described as a psychologically
disturbed adult in need of protection. TR. II, p. 356. The State said
that, in view of the relationship between the recipient and his landlady
and her daughter, this was "not an arms-length sublet arrangement."
NY
brief, App. I, Tab 6, p. 2. /29/
(67)
The SSA's analyst determined that the case was includable since the
client rented a room and there was no notation in the case file of
sharing in the purchase and preparation of food. SSA brief, Vol. II,
Tab 6, p. 2. At the hearing, the analyst said that the State's scenario
was "absurd," and that he had based his conclusion on facts, which
was
what the criteria required. TR. II, pp. 357-8. The facts he pointed to
were that one of the forms in the record (which he referred to as the
"S-2") gave the answer "none" to the question "others
in household" and
"For January of 1972, there was no reference as to being friends."
TR.
II, pp. 357-8.
We cannot agree with the State that we should conclude affirmatively
that this recipient lived in his landlady's household. The friendly
relationship with the daughter cannot be determinative here where the
record shows that the daughter lived elsewhere. Tab 6, p. 19.
Moreover, although the record does indicate that the recipient may have
depended on the landlady for protection, there is a substantial question
of whether the recipient can be considered as living with the landlady.
The caseworkers described the recipient as living in an "apartment"
on
the second floor of the landlady's house, sharing kitchen facilities and
bath not with her but with another tenant. Tab 6, p. 29. Thus, we do
not have the same likelihood of sharing between the landlady and the
recipient as where persons are actually sharing living space.
On the other hand, we find nothing in the file to substantiate a
conclusion that no sharing was taking place. There is simply no
information on the relationship between the recipient and the other
tenant, or on the total rent for the second floor. Moreover, unlike
several other cases, there is no form in the record here showing that
the caseworker determined that a client was not living with a friend, so
there is a possibility that rent was pro-rated. Thus, this case should
be excluded on the grounds of insufficient data to determine whether
there was a shared living situation.
Case 19:
The recipient in this case is a woman in her fifties who was
described as a "chronically psychotic and extremely rigid woman who
functions marginally because of her grandiose delusions and obsessional
defenses." SSA brief, Vol. III, Tab 19, p. 20. She paid $54.20 a month
for a room "in a modern house with all facilities" and had "kitchen
privileges." Id., p. 19. The house was rented or owned by a minister
and his wife.
The State said that this case fit its scenario of a dependent
recipient being helped by the family she lived with. The SSA (68)
analyst included the case on the basis that there was a room rental with
no notation of food sharing.
There is very little information for this case, perhaps because the
recipient had not applied for assistance until August 1971. What little
indication there is of no sharing is quite weak. The State awarded her
the $84 grant for a single person household, but for the reasons
discussed above we give this little weight. The SSA analyst pointed out
that the recipient had responded to a questionnaire in 1973 by
indicating that she lived alone (at the address of the minister's
house). But the question was raised in a context which suggested to the
recipient that a cwrong" answer might jeopardize her grant, so the
answer is hardly definitive. A caseworker responsible for filling in an
application form left blank the space for identifying the recipient's
living arrangement (i.e., living with family, relatives, or friends, or
alone).
On the other hand, there are several notations in the file that
support the conclusion that the State's scenario is a more reasonable
explanation of the living arrangement in this case than an arm's-length
sublet. The caseworker, noting the client's unatable mental condition
at the time she applied for assistance, said that the worker had tried
to persuade her client to get a medical and psychiatric evaluation and
the minister's wife had stated that she would go with her. Id., p. 13.
The psychiatric consultant who ultimately evaluated the recipient, using
the description quoted above, also said: "She presently is a boarder
with a minister and his wife who although recognizing she is a difficult
person, feel sorry for her and continue to keep her in their home." Id.,
p. 14. Although he also found that she was especially tidy and found
"fault with the way her hosts maintain their home," she was still
living
there in December of 1973.
While this evidence does not show conclusively that the recipient was
sharing with the minister and his wife, this is not a case where we can
say that the inference of no sharing is sufficiently strong to outweigh
the inference of sharing. Thus, the case should be excluded.
Case 21:
This is a case of a male, approximately 30 years old, who apparently
had a problem of drug addiction. In April 1971, he rented a furnished
room in an apartment occupied by Mr. C. and his family. He paid $20 per
week to Mr. C. He received the $84 basic needs grant.
The State said that this case fit its scenario of a dependent
recipient being helped by the family with which he lived. The (69) SSA
analyst found that this was a room rental with no notation of food
sharing and determined that section II.C. of the coding instructions
applied to include the case.
This is another case where there is sketchy, and possibly
conflicting, information. What information there is indicates the
following. In September 1971, the recipient requested a restaurant
allowance, stating that he did not know how to cook, that he had to eat
in restaurants, and that he did not have kitchen privileges. SSA brief,
Vol. III, p. 26. In response to this request, the caseworker consulted
with a medical social worker who said that there was no medical reason
to substantiate the approval of a restaurant allowance, so the recipient
was advised to get a letter from his doctor. Id. There is no
indication that he ever did this or that the State ever paid him a
restaurant allowance. No restaurant allowance appears on the budget for
January 1972. On December 27, 1971, the recipient came into the welfare
office and said that he had not received his check due in mid-December.
The caseworker noted:
He further stated that he was ill and this is why he could not come
in sooner to report non-receipt of PA grant. . . . Since he is living
with a family, they helped him sustain himself until he gained enough
strength to come into center. . . .
Id., p. 28
A form apparently filled in in January or February 1972 described the
recipient's living conditions as follows:
Patient is living with family (Mr. C.). He has cooking facilities
but states he has his meals in a restaurant. He does not know how to
cook. He states his living arrangements are adequate.
This form has a space for indicating if a client is living alone or
living with family, relatives, or friend. None of the boxes is checked.
We do not think that there is persuasive evidence here to conclude
that the State's scenario fit and the recipient was living in the
household of Mr. C. Contrary to some of the other cases, there is in
this case evidence (in addition to the $84 grant) that the recipient was
not sharing in food purchase and preparation. While one could infer
that the recipient had dropped his request for a restaurant allowance
because he was receiving food assistance from the C.'s, it is just as
reasonable to infer that he simply did (70) not qualify for a restaurant
allowance. He was not incapable of cooking; he just did not know how.
He said in early 1972 that he continued to eat his meals in a
restaurant. /30/ Nor is the single instance of assistance from the C.
family very meaningful here, since it apparently took place around
Christmas time. In addition, the record shows that the recipient had
rented the room in April 1971 and moved out some time prior to July
1972. Thus, there is not the sustained relationship here that is
evidenced in other cases.
In summary, we conclude that these five cases (14, 18, 6, 19, 21)
should be excluded.
(b) "Female residing with a male" cases (nos. 5, 12, 25).
The State grouped together these cases because in each case a female
recipient was residing with a male in an apartment which he rented. The
State caseworker explained that cases of this kind were not unusual,
although caseworkers were often unaware of the details of the
relationship. NY brief, App. II, Burdick Affidavit. Implying that
common-law marriages existed, the State said: "For (71) APL purposes,
if a logical inference can be drawn that in each of the three cases the
AABD recipients was living with a man as a couple sharing the apartment,
the cases should be excluded from the APL calculation, which by its
terms applies only to individuals and not to couples." NY brief, p. 52.
At the hearing and in posthearing briefing, the State did not press the
argument that these cases represented "couples," and we would not
agree
in any event that these cases should be excluded for that reason. The
State did not treat these individuals as "couples" in January 1972,
but
awarded the female recipients the $84 grant for single person
households.
The State also argued, however, that these cases should be excluded
as shared living cases because of the relationship of the recipient with
the primary tenant. SSA challenged this, asserting that the only factor
which permitted the State to infer a relationship was the fact that the
recipient was a woman and the primary tenant a man. In response to this
point, the State's caseworker testified that he would have found shared
living in these cases, based on the sustained relationship of two people
in an apartment without much room, even if the two people had been of
the same sex. TR, I, p. 143.
Although we give some weight to the State caseworker's testimony that
it was not unusual to have a common-law marriage situation even if it
was not officially recognized as such, we do not think it is necessary
to reach this issue. The parties agreed that friendship could indicate
sharing. While it is speculative to infer that the nature of the
relationships here was a common-law marriage, we think it is reasonable
to infer that there may be at least a friendship, precisely because
there was a female residing with a male: as a general proposition, we
think it is less likely that a woman would have moved into an apartment
with a man she did not know, than with a man she knew and trusted.
This does not end our inquiry, however. We must examine the
individual records to determine whether there is any indication of no
sharing which outweighs the reasonable inference of sharing. We do this
below.
Case 5:
This female recipient was about 40 years old, with a disability
diagnosed as follows: "Drug Addiction - Heroin and Others, Alcoholism
with Cirrhosis, Sickle Cell Traits." SSA appeal file, Vol. II, p. 29.
She had a husband, but left him some time in 1970. On September 2,
1970, she came into the welfare office with a note from a Mr. E.,
described by the caseworker as "the primary tenant (72) where she is now
living." Id., p. 20. The note said that the recipient had rented a room
from Mr. E. on August 20, 1970, for which she paid $18 per week. The
caseworker confirmed this by phone call to Mr. E., and noted: "Client
states she is planning to live at this address from now on and will not
return to husband. . . ." Id., p. 20. The only description of the
living conditions in the apartment is the cursory notation on a form:
"all modern conveniences." Id., p. 30.
The State said that this case should be excluded based on the
relationship of the recipient with Mr. E. The SSA analyst determined
that the case should be included under section II.C. because the
recipient rented a room and there was no notation in the file about food
sharing.
Unlike the other cases we have discussed, there is some basis here
for concluding that rent was not pro-rated. On one of the forms in the
record, under the category of "living arrangements," the caseworker
checked the box for living "Alone," rather than the one for living
with
a friend. Also, the file shows that in June 1972 the total rent on the
apartment was $78.50 monthly, and the recipient was budgeted for $72.00
monthly. Id., p. 33. /31/ allocation of the rent cost.
(73) This means that the recipient had no substantial benefit from
her relationship with Mr. E. as far as rent is concerned (although she
paid less than if she had rented the entire apartment for herself). It
does not necessarily mean that she did not benefit in other ways,
however. The file indicates that the recipient had some difficulty
functioning and may have needed care. Id., pp. 31, 29. While this
would be just as consistent with a dependency relationship as with a
friendship, we find additional support for the idea that she had
probably known him before moving in with him in the fact that she had
previously lived at an address on the same street as Mr. E.'s apartment,
at most a few doors away. Id., p. 27. /32/
This evidence is not definitive. But, considered in light of the
State expert's testimony, the inference of sharing is sufficient to
counter the slight inference of no sharing from the lack of pro-rated
rent and the fact that the recipient was determined by the State to be
living in a single person household. Thus, this case should be
classified as indeterminable and excluded on that basis.
Case 12:
This recipient was a woman in her early sixties who resided in a
furnished room in the basement apartment in a brownstone building,
sharing kitchen and bath with her landlord. She had lived there at
least since April 1969, when she applied for assistance because her
restaurant had failed. SSA brief, Vol. III, Tab. 12, p. 13.
In support of its position that the recipient must have had a
common-law marriage, or at least a friendship with the landlord (Mr.
G.), the State submitted an affidavit by an experienced investigator who
said he had spoken with Mr. G. and inspected the basement apartment in
April 1983. NY brief, App. II, Carr Affidavit. The investigator said
that Mr. G. affirmed that he had shared the apartment with the recipient
in 1972. The affidavit does not indicate whether the investigator had
questioned Mr. G. about the nature of his relationship with the
recipient. The main point of the affidavit is that there is only one
area in the apartment (which had not been renovated since 1972) where
anyone could have slept.
(74) SSA challenged the affidavit on the ground that it contained
hearsay and moved that it be striken from the record. SSA also relied
on its general argument that the State's theory was purely speculation
and on the SSA analyst's position on the case. The analyst had
concluded that there was no indication in the file of food sharing and
no notation that the recipient was a friend or common-law wife of the
landlord. SSA brief, Vol. III, Tab 12, p. 2. As usual, the analyst did
not cite to any specific evidence that no sharing was taking place.
We conclude that the case should be excluded, although we do so based
on the case file rather than the investigator's affidavit. /33/
The only indications of non-sharing here are the $84 grant, a
description (based on a psychiatric consultation) that "she lives
alone"; and the fact that a box for "Alone" is checked on one
form to
indicate living arrangement. Id., pp. 15, 19. This form, however, was
for the purpose of summarizing social factors for determining disability
in April 1969 and, unlike some of the other forms we have discussed,
does not have a box specifically identifying the arrangement of living
with "friends." At the bottom of the form, the response to the question
"How does patient get along with people with whom he lives?" is "well."
There are also other indications that support the conclusion that it is
unlikely this woman would have moved in with her landlord if she did not
know him: she suffered from hypertension and nervousness and had
difficulty sleeping, yet she lived with him in a small living space.
Id., pp. 15, 19. While this evidence is less than definitive on the
question of friendship, we think it is sufficient to counter what little
contrary evidence there is and to render the case indeterminable.
Alternatively, the case could be classified as indeterminable on the
question of shared rent. The total rent for the apartment is unknown,
so there is no indication that the recipient was paying more than her
pro-rata share. There is a notation in the file (75) that a rent
increase, apparently in 1973 (when she was still living in the
apartment), was "because of increase in utilities." Id., p. 23. This
is
somewhat more consistent with the notion of sharing than with an
arm's-length sublet.
Accordingly, we conclude that the case is indeterminable on the
questions of friendship and rent sharing and therefore should be
excluded.
Case 25:
This is the case in this subgroup with the strongest evidence of
sharing. The recipient, a woman in her sixties, had lived with a Mr.
S. in his 6-room apartment since 1955, sharing kitchen and bath. In a
note given to the caseworker in 1957, he said he had not charged the
recipient rent initially and even supplied her food, but was no longer
able to do this. Her rent in January 1972 was $12 per week, which is
what it was in 1957 (including utilities). NY brief, App. I, pp. 4-5.
/34/
The SSA analyst included the case because there was no mention of
food purchase and preparation in common.
We think that there can be little doubt that the recipient and Mr.
S. were friends. It is highly unlikely he would have provided her
assistance in the 1950s or that they would have stayed together over
such a long period of time if they had not been. This is especially so
since the recipient was a nervous person, who, at least in 1968, "had
the door tightly shut with a police lock." Id., p. 18. We think these
factors are sufficient evidence of friendship to outweigh the fact that
the recipient was receiving an $84 single-person household grant. Thus,
there is some basis here for concluding that the recipient did not live
alone.
Alternatively, this case could be excluded on the ground that it is
indeterminable with respect to the question of rent sharing. There is
no indication of total rent on the apartment, nor any indication that
the recipient had been determined by a caseworker to be living alone.
Indeed, one could say that rent sharing was (76) likely here, given that
Mr. S. had assisted the recipient in the past and that there was no
increase in rent over a 15-year period. Thus, we conclude that this
case should be excluded.
In summary, we conclude that all three of the cases in this
subcategory (5, 12, and 25) should be excluded.
(c) Upstate room and board cases (nos. 8, 20, 28 and 2).
In these cases, each recipient lived in upstate New York and received
a grant for room and board, rather than the $84 basic needs grant. The
question of shared living arises because the recipients were boarding in
private homes, rather than in commercial boarding establishments.
The SSA analyst determined that Case Nos. 8, 20, and 28 (2 is
discussed separately below) should be included because they were
"boarding" cases, citing (at least for Case 28) notes by SSA of a
meeting between the parties in July 1974. The notes had been included
by the auditors in the Criteria File. These notes said: "Boarding
situations - Include unless individual is living with relative or friend
and amount of board is low." C-28-2(a) 3/25. The analyst apparently
thought that this statement represented a criterion agreed to by the
parties. For two of the cases, the SSA analyst referred to "NYS APL
Question Log, question 1, inclusion of room and board cases." The
Question Log is included with the SSA submission and responds to the
question "How should Domicialiary Care cases be coded; specifically
those Domiciliary Care cases budgeted for their room and board under
shelter?" by instructing that a case should be coded as "lives alone"
(C-4) and "domiciliary care" (F-8).
The analyst did not specifically find that these were domiciliary
care cases, but said that boarding situations were includable because
they were similar to domiciliary care cases.
At the hearing, the State questioned the SSA analyst on whether he
had relied on a State memorandum summarizing meetings between the
parties in July 1974. That memorandum said:
It was agreed by HEW and NYS to cinlude the following shelter
arrangements if such cases had not already been screened out by the
(mutually agreed upon) screening criteria:
Boarding cases
Sublet arrangements
Criteria File, C-28-1, 5/25.
(77) The analyst said that he had relied on this document, as well as
others, for including the upstate room and board cases. TR. II, p.
371.
The State asserted that, contrary to the analyst's interpretation,
the parties did not agree to include all boarding cases. The State said
that it was clear from both the federal and the State notes of the July
1974 meetings that, although persons living in commercial boarding
houses were considered living alone, that did not apply to recipients
boarding in private homes. NY post-hearing brief, pp. 17-8; see also
TR. II, pp. 372-9. Rather, the State argued, these cases should be
excluded under the shared living criteria: specifically, the statement
in the October 30, 1975 letter that cases should be excluded if the
recipient and other person were "taking their meals together." NY
brief,
App. I, Ex. G. The State noted that the initial response by SSA to this
phrase was that it might be thought to encompass "commercial boarding
homes." Id., Ex. H. At the hearing, the State also implied that any
agreement regarding boarding cases reached in the July 1974 meetings was
contingent on the parties agreeing to a rent screen, which they never
did. TR. II, pp. 372-9.
In cross-examining the SSA analyst on these cases, the State also
asked him whether he had considered a letter dated March 24, 1975 from
the SSA Regional Planning Officer to the New York Deputy Commissioner
for Income Maintenance. Criteria file, C-28-1, 12/25. This letter
explains SSA's review of 15 cases where New York said there was
insufficient data, stating that the one case that was properly excluded
on the basis of sharing was one where an individual was living with
friends in a room and board arrangement. The letter said that the grant
amount of $108.30 for room and board was "obviously below norm. . . ."
Id., p. 2. The SSA analyst said that he concluded that this letter did
not apply because the upstate room and board cases did not have a low
room and board rate and the letter was talking about a New York City
case. The SSA analyst indicated that he had judged that the upstate
room and board cases here, involving payments of $100, $95, and $80,
were not low because there was no indication in the file that these
amounts were below norm. TR. II, p. 383. He also said that in Case 20
an increase in the room and board rate requested by the landlady was
denied because they were already paying the maximum. Id.
On redirect, the SSA analyst testified that he had some job
experience making field visits throughout New York State and that
through personal knowledge he could say that rental values are
significantly lower in upstate New York than in New York City. TR. II,
P. 434. Other than eliciting this information from the analyst, SSA did
not address these cases further despite the (78) clear challenge by New
York to the applicability of the "criteria" relied on by the analyst.
Based on our review of the record, we conclude that these cases
should be excluded on the basis that the recipient and landlady in each
case were "taking their meals together." We do not agree with SSA's
analysis because:
* The notes of the July 1974 meetings do not appear to represent an
agreement of the parties. While it is not clear as New York suggested
that everything at the meetings was contingent on an agreement about a
rent screen (which was never reached), the notes hardly evidence
agreement when the parties' descriptions of how to treat boarding cases
are different. Moreover, it is unclear whether the parties were
referring to all boarding situations or solely commercial boarding
cases. The later correspondence in 1975 concerning shared living
indicates a basis for distinguishing private boarding cases from
commercial boarding cases.
* The APL Question Log (Question #1) relied on by the SSA analyst
applies to domiciliary care cases and even the analyst recognized that
these cases do not strictly fall into that category.
* Even if we assume that SSA is correct that generally upstate
charges for room and board could be expected to be lower than such
charges in New York City, it is unclear from the March 24, 1975 letter
that the 15 cases discussed there were all from New York City. The only
reference to New York City in that letter is a mention of a problem "NYC
reviewers" were having because of an improper coding instruction. The
letter also mentions, however, that a similar instruction was found in
procedures for upstate counties. Thus, the case where SSA itself found
$108.30 per month to be "obviously" below norm for room and board
may
have been an upstate case. In any event, $108.30 is more than eight
percent higher than the highest amount for room and board here ($100),
and significantly higher than the lowest rate ($80), so we would take it
as some evidence that the charges here were low, even if the case
discussed in the letter was clearly a New York City case.
Thus, we conclude that Cases 8, 20, and 28 should be excluded on the
basis of a finding of shared living under the criteria agreed to by the
parties.
(79) In addition, we have concluded that one additional case (no. 2)
is a non-commercial boarding case and should be excluded. We,
therefore, discuss next the parties' arguments about this case and our
analysis of it.
Case 2:
The State challenged SSA's determination to include this case on the
ground that the determination was based solely on a "purported"
conversation with a Chenango County social worker, rather than on the
case record. NY brief, App. I, Tab 2. The State submitted an affidavit
by the social worker, stating that it appeared the case record had been
destroyed by the county in 1979. Id., App. II, Benedicto Affidavit.
The social worker further stated that she could not recall any
conversation with an auditor and that any information she would have
given the auditor in 1980, without the file to refresh her memory on the
recipient's living situation in 1972, would have been unreliable. Id.
SSA responded that the auditors must have had the case record and
reviewed it because audit workpapers refer to record documents and the
auditors could not have filled in their copy of the APL questionnaire on
this case without the record. SSA brief, p. 98. SSA further pointed
out that HHS auditors adhere to a strict set of ethics and standards.
Id., p. 99. In addition, the SSA analyst had commented that it was the
State's responsibility to maintain the integrity of the case files. SSA
brief, Vol. II, Tab. 2, p. 2.
In its post-hearing brief, the State said that SSA had not presented
any evidence to rebut the social worker's affidavit and that therefore
this case should be excluded, consistent with the auditors' treatment of
State-included cases when the case files were unavailable for subsequent
audit agency review. NY post-hearing brief, App. B, p. A-7.
We agree with SSA that the audit workpapers show that the auditors
did have the file to review in making their determination. The
auditors' notes on the APL questionnaire refer to file documents, as was
the auditors' practice in these cases generally. SSA brief, Vol. II,
Tab. 2. Thus, inclusion of this case is not patently inconsistent with
treatment of cases where the State lost the files prior to the audit.
We also agree generally that the State had the responsibility for
maintaining the files and ought not to profit unfairly from any failure
on its part to meet this responsibility.
This analysis does not end our inquiry, however. The APL
questionnaire forms and the auditor's notes of his conversation with
(80) the social worker (which we consider trustworthy since it appears
that the auditor had the file at the time) show the following: the
recipient was paying $90 a month for room and board to a Ms. G., with
whom the recipient resided in an apartment in upstate New York. Id.
The auditors included the case based on their understanding of what the
parties' criteria required for room and board cases. For the reasons
which we have discussed above, this conclusion was not warranted for
non-commercial room and board cases such as this one.
Accordingly, while we do not agree with the State that this case
should be excluded merely because the case file is now missing, we
cannot uphold SSA's decision to include the case since that decision was
based on an improper premise. Thus, this case should be excluded from
the APL calculation.
In summary, we conclude that these four cases (nos. 2, 8, 20, and 28)
should excluded.
(d) Insufficient data cases (nos. 1, 11, 15, 17, 22, 27).
For these six cases, New York did not argue that there was
affirmative evidence of sharing but said that there was insufficient
data to determine whether sharing was taking place, so the cases should
be excluded. SSA's position essentially was that there was no
indication of sharing, so the cases should be included. As we have
discussed above, this was inconsistent with SSA's assurance that it
would include cases only if there was evidence to substantiate a
conclusion to include the case. Although as discussed above we do not
agree with the State's presumption, based on the reasoning set out
above, we decided that, where a person is residing in the house or
apartment of another person, a case should be excluded absent evidence
of no sharing sufficient to outweigh any evidence of sharing. We have
examined these cases based on that standard.
Based on our analysis below, we conclude that four of these cases
should be included and two of them should be excluded.
Case 1:
This recipient received the $84 single-person household grant, which,
as we have concluded above, is some evidence (however slight) that there
was no food sharing. In July 1971, the recipient (who was apparently
disabled since he received AD (81) payments) moved into a "room with no
cooking facilities" for which he was charged $20 per week. /35/ SSA
brief, Vol. II, p. 18.
Since he had already paid July rent for the room he had moved from,
he was denied a further rent allowance until he returned the first rent
check and was threatened with eviction, apparently as a result of not
being able to pay for his new lodgings. Id. The information about his
new residence is somewhat conflicting; a form filled out in July 1971
shows the following:
* Under "Kind of Accommodation," the category "Rooms in Private
Apt."
is circled but it also had a question mark after it.
* The category "Rooming House" has a slash through it and a question
mark after it, followed by the handwritten notation: "There are three
other rent paying tenants."
* In a space for "Landlord" the name of a Ms. G. is written, but
there is nothing to indicate whether she is also considered a tenant.
* No apartment or room number is given; the recipient's room is
described as second floor front.
Id., p. 19.
In December 1971, immediately prior to the month we are primarily
concerned with here, the recipient failed to pay his rent and was
threatened with eviction. Id., p. 26. As a result, the caseworker took
action to have the rent paid by 2-party check to the recipient and Ms.
G. starting February 1, 1972. Id. We do not know the total rent on the
accommodations, nor do we know whether a caseworker determined that the
recipient was living alone or with friends (i.e., pro-rated the rent).
(82) We conclude that this case should be included for the following
reasons:
* The $84 grant is some evidence that there was no food sharing and
we find nothing to contradict that evidence.
* Although we do not have a specific notation finding no friendship,
nor do we know the total rent, what we do know is more consistent with
an arm's-length sublet arrangement than with rent sharing. The
recipient was threatened with eviction twice, and could not be trusted
to pay his rent. The caseworker's ambiguity about the accommodations
appears to have arisen because, although technically an apartment, the
residence was run by Ms. G. as a rooming house. Moreover, although we
do not have information about the quality or location of the residence,
the fact that the weekly rate here is higher than in most of these cases
provides some tangential support for the conclusion that rent was not
shared. /36/
Thus, we disagree with the State that this case is indeterminable,
and we include it on the basis of evidence we find sufficient to support
a conclusion of no sharing.
Case 11:
This recipient was a woman in her fifties who suffered from a chronic
respiratory condition. The recipient had moved to an apartment in
December 1969, but a form noted that she was "not the primary tenant."
SSA brief, Vol. III, Tab 11, p. 19. A caseworker attempted a home visit
in May 1970 and "was advised that the lady . . . did share the apartment
with another lady." In June 1970, a caseworker noted that the recipient
was "sharing an apartment with a Mrs. (G.) and has two room front of
this apartment and the use of the kitchen and they share the bathroom."
Id., p. 13. A report of a home visit in August 1970 states: "The
worker found our client home in her apartment." Id., p. 17.
(83) Prior to March 1971, the recipient received a rent allowance of
$21.65 semi-monthly, but a note from Mrs. G. said, "Rent was increased
to $30.00 semimonthly as of March 1, 1971." Id., p. 15. The recipient's
rent allowance was not actually increased to account for this until the
end of May 1971, but there is no indication that she was threatened with
eviction for non-payment or anything of the sort. Id., p. 16. The
recipient was living in the apartment and paying $30.00 semi-monthly in
January 1972 and continued to live in the apartment even after Mrs. G.
moved out August 1, 1973. Id., p. 20.
The recipient received an $84 single-person household grant. A form
summarizing social factors for determining disability in June 1970 shows
a check by a box for living "alone," leaving unchecked a box for living
with a friend. This form describes living quarters as 1st floor, 2
rooms and "share kitchen." Id., p. 18. The total rent on the apartment
in January 1972 is unknown, although there is some indication in the
file that total rent in August 1973 was at least $129 per month.
We conclude that this case should be excluded as indeterminable. The
$84 grant indicates no food sharing and there is nothing specific to
contradict this. But with respect to the question of rent sharing we
have only the fact that the box for friendship was not checked on one
form. As we have noted with respect to other cases above, however,
since this type of form focused on the disability determination rather
than grant amount, it does not as clearly represent a determination by a
caseworker that rent was not pro-rated. Moreover, there are some
indications to balance the implication of no rent sharing from how this
form was filled in. Unlike some of the other cases where a caseworker
says merely that a roomer has use of a kitchen or access to other parts
of an apartment, this recipient is described as sharing the kitchen, and
even sharing the apartment. The rent is less than Case 1 (when figured
on a weekly basis), although this recipient had two rooms rather than
one. Also, even allowing for some increase between January 1972 and
August 1973, it appears that the recipient was only paying about half
the total rent on the apartment.
Case 15:
This recipient was a woman in her fifties, suffering from arthritis,
heart condition, and other ailments (including (84) obesity). She
resided in a 3rd floor room in a building described as a house. She
paid $39 semi-monthly for rent, and received the $84 grant. Notes from
a home visit in January 1969 state:
The furnished room on 3rd floor rear is well heated, adequately
furnished and is compact, . . . Kitchen and bath shared. The bldg is
locked, and most of the inhabitants resemble (the client) in situation;
she knows her neighbors and they get along well.
SSA brief, Vol. III, p. 14.
Later notes, from a July 1969 visit, state:
. . . she cannot negotiate the stairs more than once a day because of
shortness of breath; this led to the following topic:
(recipient) is dissatisfied with current conditions because of the
flights of stairs, the isolation and small size of her room, and the
general lack of privacy she has in such a large house with a community
kitchen. She also feels, as is well known, that the neighborhood is
dangerous and she is a prime target for burglars, although the house is
at least locked and secure. . .
. . . Wkr spoke to supt. (Mrs. H.) on 1st fl., who verified rent
(had been paid).
Id., p. 15.
A form summarizing social factors for determining disability in
October 1969 shows a check for lives "alone", leaving lives with
"friend" unchecked. Handwritten notes on the second page of this form
state: "Client cannot perform more than immediate self-care and
cooking; it is nearly impossible for her to negotiate stairway, . . .
." Id., p. 17. There are also indications in the record that she
required a special diet. Id.
We conclude that this case should be included. The inference from
the $84 grant that she was not sharing in purchase and preparation of
food is strengthened by the indications that she did her own cooking and
was on a special diet. While the recipient got along well with her
"neighbors," we do not know if this meant the other boarders in the
house or neighbors along the street; even if it was the former, the use
of the term "neighbors" would imply separate living arrangements rather
than sharing. Moreover, the recipient (85) apparently complained of
isolation and lack of privacy in "such a large house." There is no
indication in the file affirmatively indicating food sharing, except the
notation that she shared the kitchen, which is also described as a
"community kitchen."
Similarly, with respect to the question of shared rent, the whole
situation is, from what little we know of it, more consistent with the
notion of an arm's-length sublet. Besides the choice of the "alone"
box
rather than the "friend" box on the one form, we have a building with
a
superintendent and, apparently, a number of "inhabitants." The rent
is
relatively high for these cases, although the recipient had one small
room and had to negotiate three flights of stairs. We do not know any
"total rent" figure, but then we are not even sure that the building
is
rented, as opposed to owned, by the landlord.
In this case, we think that the inference of no sharing sufficiently
outweighs any possible inference of sharing so that this case properly
should be included.
Case 17:
This recipient, a woman about 70 years of age in good health, had
moved to a "furnished room" in March 1968. SSA brief, Vol. III, Tab
17,
p. 13. Caseworker notes of a visit in June 1968 state:
(Recipient) occupies a nice furnished room with cooking facilities
available in an old house in Flushing. The room is adequately furnished
for (her) needs, and she seems to be happy with her present living
arrangements. She stated that she gets along well with her landlady and
that she minds her own business and does not bother any body.
(She) is an extremely good manager. She states that she is very,
very careful with her money and is always able to take care of her needs
adequately. She states that she shops carefully, and tries her best to
get bargains at the shop when things are on sale. She states that she
sees no reason why everyone on welfare cannot manage as well as she
does.
Id., p. 14.
Notes of a visit six months later say:
Living conditions remain unchanged. (She) continues to occupy a
furnished room. She has cooking facilities (86) available to her.
(describes furnishings in room) . . . she is happy with her present
living conditions.
* * *
(She) is a very good manager and never has any problems in this area.
. . . She shops wisely. . . .
Id., pp. 14, 15.
In October 1972, she was still residing at the same address and
volunteered the following information regarding her household status:
"Alone - one person." She received the $84 single-person household
grant
and $43.33 semi-monthly for rent.
We find nothing in this file which could possibly indicate that the
recipient shared food with her landlady other than that they apparently
got along well and the cooking facilities were "available" to the
recipient. On the other hand, in addition to the $84 grant award, the
file gives a picture of a woman who was proud of her self-sufficiency,
did her own shopping and budgeting, and did not bother anyone.
With respect to the question of rent sharing, we have again a
situation where the recipient rented a room in a house which may have
been owned by the landlady, rather than an apartment rented by another
tenant. The rent she paid was relatively high, although she only had
one room and there is no indication she had access to the rest of the
house, other than the kitchen.
In this case, it is more reasonable to infer that no sharing is
taking place than that it is; thus, we conclude that the case should be
included.
Case 22:
This man in his fifties, suffering from liver disease and other
physical problems as well as depression, moved in December 1970 to "a
furnished room with use of a kitchen . . . roomer in a private
apartment." SSA brief, Vol. IV, Tab 22, pp. 21-3. Most of the
information in the file pertains to his health status; what little
information there is on living arrangements does not pertain to the room
he moved into in December 1970 and continued to occupy in January 1972.
We find this case to be indeterminable on the question of shared
rent. The total rent on the apartment is unknown and we find no
indications here of an arm's-length sublet arrangement. There is (87)
no form in this file to indicate whether rent was pro-rated. /37/ Thus,
this case should be excluded.
Case 27:
This man was probably in his sixties (although he had trouble
verifying his age) and had lived in a room in a private apartment since
1954. SSA appeal file, Vol. IV, Tab 27, p. 18. He was apparently
friends with the primary tenant and she helped him out in time of need,
but she lived in New Jersey, not with him, so this is not a sufficient
basis for a finding of sharing here. Id., pp. 14-9, 20-1. He paid rent
of $26 semi-monthly to the primary tenant's sister.
The caseworker notes from an intake interview in October 1967 are
largely illegible, although they appear to say at one point, "lives
alone in a furnished room. . . ." Id., p. 17. Notes from a home visit
in October 1967 say that the recipient had stated "that he has use of
cooking facilities and bathroom." Id., p. 20. Notes from a visit in
December 1969 refer to "other tenants" and that the recipient "shares
kitchen and bathroom." Id., p. 22. They also say that he "has no close
relatives and few friends" and that he cis a lonely, sick old man who
spends too much time by himself." Id. A form in the file which
identifies the recipient as the "Tenant in Occupancy" and a landlord
who
lived in Brooklyn (not New Jersey) apparently directed the landlord
effective October 1969 to reduce the rent on the apartment to about $65
per month. Although the recipient remained at the same address for a
long time, there is no file notation to indicate whether the "other
tenants" did so as well.
The more reasonable inference from these facts is that this recipient
was not sharing rent or food with other tenants. It appears that he
remained in the apartment because of his (88) relationship with the
"primary tenant" rather than any other tenants who may have lived
in the
apartment with him. (He received an $84 grant and there is no
indication of food sharing other than the mere notation that he shared a
kitchen.) While we do not know whether the total rent remained at $65 in
January 1972, it is reasonable to assume it did since the recipient
continued to receive the same rent allowance throughout that period.
The rent allowance amounted to $52 a month. This is substantially more
than half of $65 so it could not be a pro-rata share unless he occupied
substantially more than one-half the apartment. Yet, he only had one
room to himself: there must have been at least one other bedroom to
accommodate another tenant and here there is a reference to other
tenants. Although there is no specific determination that the recipient
was not living with friends, there is the notation by a caseworker that
he lived alone.
The evidence that there was no sharing here is based on a reasonable
inferences, countered by nothing except the mere possibility that the
recipient shared with the other tenants. Thus, we conclude that this
case should be included.
In summary, we conclude that four of these "insufficient data" cases
(nos. 1, 15, 27, and 17) should be included and two (nos. 11 and 22)
should be excluded.
II. The challenge to SSA's APL calculation.
A. Background and summary of holding.
In addition to challenging some of SSA's individual case
determinations, New York also questioned the propriety of SSA
calculating an alternative APL, as well as the validity of the
statistical processes SSA used and the way in which SSA used the
results.
Briefly, our decision is as follows: review of the record indicates
that generally, SSA's sampling and statistical methodology were sound
and reasonable, but the way in which the data were used involves two
important issues which should be further addressed by the parties. The
issues are: whether the "point estimate" should be used rather than
a
more conservative figure closer to the confidence interval limit; and
whether the "difference estimate approach" should be used to calculate
an APL rather than the method SSA used. Since the record is incomplete
concerning these two important elements, and given that the APL
calculation is subject to substantial revision in any event because of
our conclusions above on individual cases, we concluded that it would be
in the best interests of the parties, and in accord with the (89)
cooperative nature of the SSI program, to remand this portion of the
case for further review in light of the analysis below. If the parties
are unable to agree, we are prepared to take the case back, after the
further review, on an expedited basis.
B. Summary of facts.
There were about 712,000 cases on the State's January 1972 assistance
rolls, of which 248,000 were identified as AABD cases. New York used a
computerized process to eliminate cases which clearly were inappropriate
for inclusion in APL calculations, leaving 133,470 for a manual review.
New York found documentation for 112,110 of these cases. By agreement,
the 21,360 for which documentation was not found were excluded. New
York and a consulting firm used a questionnaire to examine the 112,110
cases.
Of the 112,110, New York determined that 28,071 should be included in
the APL calculation, and the rest, 84,039, should be excluded. The
former came to be called the "included Stratum" or "Stratum I"
cases,
and the latter the "excluded Stratum" or "Stratum II."
New York contracted with the accounting firm of Arthur Young and
Company to do a quality assurance review of the data collection process.
The firm examined a sample of 4,426 cases from the overall universe, and
concluded that the efforts were adequate.
HHS auditors used the 4,426 Arthur Young cases as a basis for their
APL determination. The auditors developed a sample from the Arthur
Young cases which consisted of 666 cases for Stratum I (New York's
"included" cases) and 508 for Stratum II (New York's "excluded"
cases).
For Stratum I, the HHS auditors excluded 36 cases for which they
determined there were incomplete or missing files, and found 30 cases
which they determined were erroneously included. For Stratum II, SSA
excluded 130 cases on the basis of incomplete or missing files, and
determined that 4o cases had been erroneously excluded. The auditors
then computed a mean value for the 600 includable Stratum I cases and a
mean value for the 43 includable Stratum II cases. Then, the auditors
multiplied each such mean value by the related population for the
respective stratum, added these figures, and divided the sum by the
total population to produce a weighted composite mean. The mean was
adjusted for certain factors outside the computation process so far
described. This, then, produced a figure of $191.56 +/- $3.28 (i.e., a
midpoint or "point estimate" of $191.56 and a confidence interval
of
from $188.28 to $194.84 at the 95% confidence level).
(90) C. The sampling process.
As we stated, we find that the sampling methods used by the auditors
generally are reasonable and reliable, although, further below, we
discuss our reservations about the use of the results of the sampling.
In this section are findings and conclusions on the methods.
1. Use of a sample vs. use of the universe.
New York questioned SSA calculating an APL based on a sample when New
York had calculated an APL based on all 28,071 cases. NY post-hearing
brief, pp. 27-8. New York argued essentially that a sample has an
inherent sampling error, while the universe has no sampling error
(although it may be affected by measurement error). /39/ Id.; TR. II,
pp. 519-23. This argument amounts to a truism, and was acknowledged by
SSA's statistical expert Ms. Joanne Greer. TR. III, pp. 716-9. SSA's
position essentially was that the universe had measurement errors, while
SSA's sample had less measurement and low sampling errors. SSA brief,
pp. 49, 60; TR. III, pp. 718-20.
One authority on statistics has referred to the "seemingly
paradoxical situation" that, while in theory measurement errors can be
eliminated and sampling errors cannot, "in practice sampling error can
often be made to be a much smaller part of the total error in an
estimate than the non-sampling measurement errors." Mandel, Id., p.
214. This is because sampling error can be readily measured and
reduced, but measurement errors generally cannot without substantial
effort. Id.
(91) It is clear that there were measurement errors in New York's
Stratum I universe of 28,071 cases. New York acknowledged this. NY
post-hearing brief, p. 28; see also, TR. III, pp. 520-1. The State
said it "has not challenged the (auditors') findings in the 'include'
cases where errors were found." Id. Rather, New York primarily relied on
the argument that there was a better alternative to SSA's recalculation
of the APL (discussed below). Id. In this decision, we have also
confirmed the SSA determination that a number of Stratum II cases were
wrongly excluded. Thus, overall, there were many measurement errors in
the universe. /40/
There are also other indications of error. There was testimony that,
during the State's review, which took several years, the agreed-upon
rules for classifying cases sometimes changed, and the changes were not
always implemented for cases already reviewed. TR. I, pp. 81-2. There
was evidence, albeit not dispositive, that the 100% review was performed
by clerical personnel including temporary clerical personnel (TR. I, p.
73) who arguably were not ideally accurate in dealing with the complex
records and criteria. There was testimony that a massive undertaking to
review many thousands of cases inherently was subject to more error than
the closer, more professional review of the much smaller group of cases
in the sample. TR. IV, pp. 841-4. We conclude, therefore, that there
clearly were enough reasonable questions about the accuracy of the
universe to justify SSA's looking closely at a sample of cases, rather
than merely relying on an APL based on the universe.
Concerning the sample, it is not contested that the auditors
"reviewed each and every state welfare case file and made an independent
determination of inclusion/exclusion as to each of these cases. . . ."
SSA brief, p. 56; TR. II, pp. 264-5, 388-9. The auditors'
determinations concerning erroneously included Stratum I cases were not
contested by the State. The State did (92) contest some of the
determinations on Stratum II, and we have upheld the State in regard to
some of those challenged; however, as we noted above, the corrections
effectively remove the measurement error. /41/ Overall, then, there is
no meaningful measurement error in the auditors' sample, as corrected.
Concerning randomness, we note that although New York argued strongly
that Stratum II contained evidence of statistical bias (an argument
which we consider, and reject, below), one of the State's witnesses said
he had "no reason to believe that stratum 1 is not a random sample of
the included cases." NY brief, App. II, Heiner Affidavit, P8. Both
Arthur Young and Company and HHS used random number generators to select
the cases reviewed in the sample. TR. III, pp. 700-2 (we discuss use of
the Arthur Young cases further below). Based on the foregoing, and on
our analysis of certain issues regarding Stratum II which we discuss
below, we conclude that the sample cannot reasonably be impeached on the
basis that it was not random.
2. Sampling error in the sample.
Of course, there was sampling error in the sample. This is not
disputed. This does not impeach the sampling process per se, as we have
discussed. The important issues are whether SSA validly accommodated
the sampling error statistically--which also does not appear to be
disputed (SSA computed a confidence interval using a confidence factor
of 95%)--and whether SSA then validly applied the results of the
sampling process. The latter is much disputed, and we have determined
below that there are some legitimate unresolved questions about the
latter point.
3. Use of the Arthur Young sample.
A point relied on by the State early in the dispute, and less so
later, was that it was improper for SSA to have relied on a sample drawn
from a sample (as noted, HHS auditors drew their 1,174 cases from Arthur
Young's 4,426 cases).
There was substantial expert testimony that the Arthur Young study
was random and thorough. TR. III, pp. 687-8; TR. IV, 826-7; Greer
Affidavit, PP12-15. The testimony of SSA's statistical (93) experts
Drs. Arkin and Greer, who generally we found to be more persuasive than
the State's witnesses on the technical issues involved here, /42/ was
more than ample to rebut the conclusory charge of the State's witness
that drawing the sample from the Arthur Young sample was a "weakness."
NY brief, App. II, Heiner Affidavit, P8.
It is common sense that if the subsample is a random sample from the
prime sample and is itself sufficient in size to project to the
universe, and the prime sample was a random sample, then the fact that
there is a sample drawn from a sample is immaterial. Stated another
way, if the subsample is random and large enough in relation to the
universe, it makes no difference that it was drawn from an interim
sample. "The selection of a random sample from an already existing
random sample is a perfectly acceptable statistical approach and in no
way impeaches the results of the sample survey." Greer Affidavit, P18.
We have found based on the record that the subsample was randomly drawn
from the Arthur Young sample, and we find that the Arthur Young sample
itself was randomly drawn from the universe. The remaining issue is
whether the subsample size of 1,174 was statistically adequate. This
has not been disputed. /43/ Therefore, we conclude that it was
reasonable and valid to use a subsample from the Arthur Young and
Company sample.
4. The "representativeness" of the sample.
New York initially challenged the auditors' sampling methodology on
the basis that the sample was not randomly drawn. In the face of the
evidence which we have discussed above which showed that Arthur Young
had employed a random number generator in constructing its sample, New
York later focused instead on its argument that in any event the sample
was not representative of the universe from which it was drawn. The
State based this argument on an analysis of Stratum II performed by Dr.
Heiner based on information presented in two charts: Chart 1 compared
the frequency of certain (94) exclusion categories (derived from the
codes assigned to the cases on the APL questionnaire, e.g., E-5 & G-3,
N=1, 0=1) in the sample with their frequency in the universe; Chart 2
compared for six exclusion categories the "average value" of sample
cases in each category with the "average value" which the State had
calculated for all the cases in that category in a "universe subset."
With respect to Chart 1, Dr. Heiner stated that a few categories
where the auditors had found exceptionally high rates of error were
"overrepresented in the sample," that is, significantly more cases
in
these categories appeared in the sample than would be expected given the
ratio of sample cases to cases in the universe (1:203). Heiner
Affidavit, p. 11-2. On cross-examination, however, Dr. Heiner stated
that looking at all of the categories and considering the small number
of sample units in some of the categories, the ratios in Chart 1 were
not that startling. TR. III, pp. 581-5. /44/
Essentially, Dr. Heiner acknowledged that he would not be concerned
about the representativeness of the sample based on Chart 1 alone. Id.
Thus, the critical inquiry here is whether Dr. Heiner correctly
concluded based on Chart 2 that there was reason to question whether the
Stratum II sample was truly representative of the universe from which it
was drawn, and, if so, how this affects use of the sample results.
As indicated above, Chart 2 was presented by the State as a
comparison of the average dollar values of cases in the sample with the
values of cases in a universe subset. The State explained in a footnote
to the chart that it could not present figures for the entire universe
of each ofthe six categories listed because, in arriving at the figures,
it could not retrieve (95) (from its computerized data base) the values
for cases which were ultimately excluded from the APL calculation both
by an "Exclusion Program" and by a "Computation Program."
/45/
The State said that the editing done by the computer should not
affect the representativeness of the average value used in Chart 2 for
the universe subset of each of the six exclusion categories. Dr.
Heiner's analysis of Chart 2 was based on the premise that this was
correct and, more generally, that the State's figures were accurate.
Dr. Heiner said, and Dr. Carroll (the State's other statistical expert)
agreed, that the fact that Chart 2 showed for every one of the six
exclusion categories listed that the average value per case in the
sample was significantly lower than the average value per case in the
universe subset evidenced a distortion in the sample and therefore the
sample results were unreliable.
In attacking the validity of the comparison in Chart 2 and the
conclusions the State's experts drew from that comparison, SSA relied
primarily on the following:
* Testimony by Dr. Greer that the State's presentation was based on a
"post-hoc" analysis, which is disfavored in statistics because it
is
subject to bias, and which, (96) in any event, should only be used if
certain "penalties" are applied in the analysis, a step which Dr.
Heiner failed to take here.
* Testimony by Dr. Arkin that the results of a sample cannot be
rejected or modified simply because of an after-the-fact "feeling"
that
the sample is not representative and that "the sample has to be accepted
unless you can demonstrate that there has been indeed a failure in the
execution of a proper sampling plan."
TR. IV, pp. 820-1. /46/
New York's response on these points was essentially as follows:
* That a limitation on the use of sample data is the distinct
possibility that the "true" figure sought to be estimated does not
lie
within the confidence interval; specifically, that there is a five
percent chance here that the true universe figure is not within the
confidence interval, NY post-hearing brief, p. 33, citing Heiner
Affidavit, P3, Greer Affidavit, P26;
* That, even if a sample is properly designed and implemented, it is
still possible that the sample was by chance skewed toward extremely low
value cases, Id., p. 33, citing TR. III, pp. 647-8, 658-9;
* That it was undisputed that the six groups of sample cases were
significantly lower in value than the cases for which they were a proxy;
and
(97) * That this distortion should not be ignored simply because the
State's analysis was "post-hoc," especially since the State did not
participate in developing the sampling plan so that its analysis
necessarily occurred after the sample was drawn.
NY- post-hearing brief, pp. 33-6.
Based on our consideration of the record, and the purpose behind our
examination, we find that we cannot agree with either party entirely on
this issue. We cannot adopt Dr. Arkin's proposition that the results of
a sample must be accepted so long as there is no failure in the
execution of a proper sampling plan. While this requirement may apply
to a statistician or auditor who has decided to proceed by use of
statistical sampling techniques, the question relevant to our
decision-making where statistical sampling evidence has been introduced
is whether that evidence is reliable. Where, as here, the sampling
methodology is shown to be proper (see our discussion above), we
consider the evidence to be presumptively reliable based on the proven
validity of such techniques in general for achieving a high degree of
accuracy. /47/ This does not mean, however, that we are precluded from
considering other evidence which may show that in a particular case the
results of the sample cannot be relied on. It is clear that, even when
a proper methodology is used, having a 95 percent degree of confidence
in a result still permits a five percent chance that the "true" value
sought is outside of the confidence interval.
On the other hand, given the validity of statistical sampling
techniques in general, the State here (as the proponent of rejecting or
modifying the sample results) has a burden to produce evidence
sufficiently strong to overcome the presumptive reliability of the
sampling evidence. Here, the State's evidence simply does not meet that
burden.
(98) We find the State's evidence in Chart 2 and the testimony based
on that chart to be defective, not merely because it was "post-hoc"
analysis (any analysis performed by a party other than the one which
performed the sample necessarily has that characteristic), but for more
substantive reasons. The record provides a reasonable explanation of
the discrepancies shown in Chart 2 between the universe subset average
values and the corresponding sample values. In addition, the State's
analysis may have been biased.
In support of the figures used as the average value of the universe
subset for each of the six exclusion categories listed, the State
submitted a computer printout for each of the categories. NY brief,
App. II, Ex. R. These printouts are identical to the printout showing
computation of the State's submitted APL figure, except with respect to
the figures shown. NY appeal file, Ex. 5, p. 7. Thus, it is reasonable
to conclude that the "average value" figures the State gave for the
six
exclusion categories were computed using the Computation Program
designed for the APL calculation. Indeed, the "average values" shown
on
the printouts are identified as "adjusted payment levels" and the
"universe subset" figures shown on Chart 2 appear on the printouts
as
"total included." Ex. R (emphasis added). This explains the
discrepancies on Chart 2 when considered with one additional fact shown
in the record: a flowchart describing the Computation Program logic
directs that, after determining an average payment level for includable
cases, the $10 food stamp bonus should be added to the average payment
level to find the adjusted payment level. In other words, the figures
identified by the State as the average values for the universe subsets
may be expected to be higher than the sample values (which did not have
the $10 food stamp bonus added in). Thus, it can be reasonably
concluded from the record that Chart 2 does not represent a fair
comparison; the difference in values between the universe subset and
the sample items has a logical explanation other than distortion of the
sample. The testimony by the State's experts was premised on the
accuracy of Chart 2 and cannot be a basis for impeaching the sample
where the record shows that, in fact, the figures were inaccurate as a
basis for comparison.
Moreover, there are other reasons why we are not persuaded by the
State's analysis based on Chart 2. There were 32 exclusion categories,
yet the State provided information regarding only six. The State
explained that this was because AABD budgets (and, consequently, total
payments) would not have been filled in for these cases. SSA challenged
this, stating there was no reason for distinguishing these six
categories from the remaining categories on that ground. Nothing in the
record indicates that the reviewers would be more apt to fill in budgets
for the six categories than for some other ones.
(99) In addition, the State's experts premised their testimony on the
State's assurance that there was no reason to believe that the values of
the edited cases varied significantly from the values of the cases in
the subset. However, we think there is such a reason: one of the
computer edits in the Computation Program excluded cases where the total
dollar value was less than $94. See note above. Deleting such cases
from computation of the "average values" for each category could be
expected to raise the average value. The State appeared to recognize
this itself when it submitted a revised Chart 2 at the hearing; the
revised chart compared only sample cases drawn from the universe subset
with the universe subset. Yet, the State did not call to our attention
why it made the revision, nor did the State revise its statement in the
footnote. By mentioning this, we do not mean to imply that the State
acted in bad faith, but merely that, in actively pursuing its case, the
State may have tended to see the evidence here through a somewhat
jaundiced eye.
Given the defects we find in the State's evidence, we cannot say that
the State's evidence was sufficient to overcome the presumption of
reliability the sample results have.
D. REMAND ISSUE No. 1: use of a "difference estimator."
New York argued essentially that SSA, if it was going to second-guess
New York's APL, should at a minimum calculate an alternative APL by
using a "difference estimator" approach rather than merely figuring
its
own APL ab initio. n48 A State witness described how to use the
"difference estimator" approach as follows:
Take the New York State's original census figure of 191.59 and adjust
that downward, based on your estimate of what SSA auditors would have on
the average found to be the discrepancy in each case. TR. III, p. 530.
An SSA witness described the approach as follows:
If you take a difference estimator, the result of your work will be
an estimate of how far they were wrong. That is not the analysis which
was done (by SSA). The (100) analysis which was done was an adjusted
value analysis in which the corrected values were projected and they are
equally acceptable." TR. IV, p. 813.
Both of SSA's expert witnesses testified that the "difference
estimator" approach is as valid statistically as SSA's "direct
projection" approach. TR. III, pp. 694-5; TR. IV, pp. 813, 819, 824-6,
862. In fact, one of these witnesses, Dr. Arkin, testified that the
approach "probably" or "generally" would be more precise.
TR. IV, p.
862. The other SSA witness, Dr. Greer, appeared to have reservations
primarily because the auditors did not originally set out to follow such
an approach, so that to do so now would be inconsistent with the
original plan. TR. III, pp. 694-5. A New York witness, Dr. Carroll,
testified that SSA's approach was not invalid, but was less precise (TR.
III, p. 632) while New York's approach was "fairer" (id., p. 636),
"more
precise" (id., p. 637), and "preferable" (id., p. 657). Another
New
York witness, Dr. Heiner, testified thast the "difference estimator"
technique was superior to both SSA's "direct projection" approach
and
the use of New York's whole universe figures (which earlier the same Dr.
Heiner had promoted). TR. III, pp. 529-33; see also Heiner Affidavit,
P11. He stated:
I prefer it to the SSA approach because they don't account for any
sampling error. I prefer it to New York State's approach because they
don't account for any measurement error. This accounts for both.
TR. III, p. 533.
According to Dr. Arkin, under the "difference estimator" approach
and
SSA's "direct projection approach," the "point estimate"
(or mid-point)
of the distribution of measured items would be about the same. TR. IV,
pp. 824-5. Why, then, the dispute? Primarily because while the point
estimates of the two distributions may be about the same, the confidence
intervals may be substantially different: according to Dr. Arkin, the
sampling error is different for the "difference estimator" approach.
Id., p. 825. Using the State's figures, which we assume for the purpose
of illustration are correct, we see that this reduction in sampling
error using the "difference estimator" approach produces confidence
interval figures for Stratum I (using the 95% level of confidence) which
are quite different: SSA's figures for Stratum I /49/ were $188.73
(101) +/- $3.56, while New York's figures for Stratum I using the
"difference estimator" were $191.47 +/- $1.06. NY hearing exhibit
3;
TR. III, p. 713. It should be borne in mind that, according to SSA's
counsel, each penny in the APL has about a $50,000 magnitude. TR. III,
p. 574.
SSA's response to the "difference estimator" approach was as follows:
Since the use of the direct projection approach is statistically
valid, commonly accepted and popularly used in audits, its use in the
present audit must be upheld by the Board. It is submitted that it
should not be the function of the Board to substitute its judgment for
that of the Office of Audit as to which of several valid statistical
methodologies should have been used in the audit . . . (a contrary)
decision would establish the broader precedent that complete reliance
upon commonly accepted statistical textbook procedures is no guarantee
that an audit will withstand scrutiny by a reviewing body.
SSA's post-hearing brief, pp. 22-3.
Beyond the foregoing, essentially collateral, argument, SSA did not
attempt to justify not considering the "difference estimator" approach.
Presumably, this was due to SSA's position using the "point estimate"
as
its APL: if the "point estimate" was the same for both calculations,
and if we upheld use of the "point estimate," then SSA probably would
be
correct that we should not look behind the method used. The method used
would not make any difference. But there are two problems in the case
before us: first, it was not established dispositively that the "point
estimates" of the two distributions indeed are identical (a difference
of even a penny would be meaningful); and second, more important, we
have questioned the proposition that the point estimate is the correct
APL (see discussion below). We believe that SSA is quite wrong in
arguing that we should give deference to an unexplained choice of
alternative approaches which could result in a difference of millions of
dollars. While the two methodologies may be equal, their results are
not, and it is clear that SSA has a responsibility to justify to New
York SSA's selection of the one which would result directly in a
substantial loss to the State. It would be indefensibly arbitrary to
shield a huge differential impact behind the merely incidental fact that
the methodologies used are both statistically sound.
At the same time, we are by no means prepared to decide for the State
on this issue based on the record before us. As best we can (102) tell,
over the many years that this long and complicated dispute has lasted,
New York meaningfully raised the issue of use of the "difference
estimator" for the first time only at the hearing in this case. There
may well be reasons, presently not clearly articulated, why SSA's
approach is justified. The record simply is too incomplete for us to
render a well-considered opinion, and it is incomplete not in small part
because of New York's tardiness in raising the issue. Furthermore, the
issue may become moot as a matter of practicality because of future
discussions the parties are likely to have in response to other holdings
above and below. Therefore, we remand this issue to SSA for further
consideration in light of our discussion. If the issue is not resolved,
SSA should give New York a substantive explanation of the justification
for the differential impact of use of a "direct projection approach"
in
relation to the facts of this case. New York may return to the Board if
it disagrees with that determination.
E. REMAND ISSUE No. 2: use of the point estimate.
The "point estimate" is the mid-point between the upper and lower
limits of confidence in a frequency distribution. TR. IV, p. 819.
These limits describe a range of error due to sampling called the
"confidence interval." Mandel, p. 251; Greer Affidavit, P26. Using
the
95% confidence level (common and customary), as was done in this case,
means that statistically we can say there is a 95% probability that the
"true" universe value lies between the lower confidence interval and
the
upper confidence interval. SSA's "bottom line" confidence interval,
before adjustments, for the average money payment was $178.44 to
$185.00, with a mid-point of $181.72 (the weighted composite mean,
before adjustments). Greer Affidavit, P26. This can be shown as
follows:
(See Illustration in Original)
(103) What can also be seen from the illustration is that there is a
50% chance that the universe value is higher than $181.72, and a 50%
chance that the universe value is lower. Thus, using the mid-point
effectively means that New York and SSA are equally sharing the risk
that the mid-point is an incorrect approximation of the true universe
value.
The State took the position that this sharing is unfair. NY
post-hearing brief, pp. 30-1. The State argued, in effect, that (a) a
greater risk ought to be borne by SSA, which did the sampling and chose
the sampling size; and (b) that if the State's own universe figure
falls within the 95% confidence interval, then the sampling process
ought to be viewed as confirming the State's figure, rather than as
setting an alternative figure. See, e.g., TR. III, pp. 519, 639.
Concerning the first argument, we find there are the following
factors to be considered:
* Dr. Arkin, SSA's statistical expert, testified --
HHS won't like this, but, again, in my opinion, one should never use
a point estimate unless the sampling error is relatively negligible.
Otherwise, you've got to make some allowance for it.
TR. IV, p. 868; see also, pp. 865-6.
* The Board has in its possession, from a different case, a
memorandum of April 22, 1980 in which an Acting Assistant Inspector
General for Auditing of HHS stated, among other things, that, in using
sampling for determining disallowances, a point estimate "may only be
used . . . when the sampling error at the 95% confidence level is less
than 10 percent of the point estimate." /50/ The memorandum, of course,
does not speak to APL calculations, and we do not know how the quoted
and other provisions of the memorandum would (104) apply in the
recalculations that will be done in this case. Nevertheless, the policy
in the memorandum should be considered by the parties.
* The Board has dealt with invalid uses of point estimates in two
cases which the parties should have an opportunity to review. Cf., Ohio
Department of Public Welfare, Decision No. 226, October 30, 1981;
University of California--General Purpose Equipment, Decision No. 118,
September 29, 1980. We do not suggest that these cases are dispositive
here, but only that the cases show that use of a point estimate, while
it may be appropriate, ought to be affirmatively justified in the facts
of a given case.
The thrust of SSA's arguments in this case were directed at the
validity of the statistical sampling process assuming the validity of
the individual case determinations, and SSA made no substantial
arguments related to the principle of use of the point estimate. While
use of the point estimate may be valid--after all, it reflects the most
probable value, and sharing the risk of sampling error arguably might be
reasonable in the facts of this case--we think it inappropriate to
decide the issue in the absence of more focused argument, and, since our
individual case determinations will produce further calculations, we
believe that the parties ought to have the opportunity to consider this
matter further.
Concerning the State's second and overlapping argument--that the
State's APL should be considered affirmed if within the confidence
interval calculated by SSA--we note the following:
* The State's admissions regarding its measurement error in Stratum I
cases requires some adjustment of the State's submitted figure, but, as
discussed above, there are at least two reasonable approaches as to how
this can be done.
* Since recalculations based on our individual case determinations
for Stratum II likely will substantially change the proposed figures,
the nature of this element of the dispute is no longer as well defined.
* New York argued that SSA had affirmed the APL figure calculated by
Massachusetts because that figure was within the confidence interval for
the lower figure calculated by federal auditors based on a sample of
Massachusetts AABD cases. NY brief, (105) p. 29; see also App. I, Ex.
C. If SSA did, indeed, do this for another state, SSA may be willing to
do the same for New York based on recalculations pursuant to our
decision. Otherwise, SSA has an affirmative obligation to explain why
it would not.
As with the "difference estimator" issue, we have concluded that
the
parties should have the opportunity to consider this matter further. Of
course, if these matters remain in dispute, the Board is prepared to
provide further review on an expedited basis.
CONCLUSION
Based on the foregoing analysis, our conclusions are as follows:
* We conclude that New York was correct in 18 of the individual case
determinations, and that SSA was correct in 10 of these determinations.
A chart summarizing our findings is at Appendix A.
* Generally, we conclude that SSA's statistical methodology was
valid. There are two issues which we remand for further examination:
the use of the "difference estimator" approach; and the use of a "point
estimate." We anticipate that our individual case determinations and
other matters discussed above will prompt discussions which will resolve
or moot these issues, but if not, the Board will provide further review
on an accelerated basis.
(106) APPENDIX /1/ Exactly how much is not clear. New York said that
the "impact of the adjustment" would be "some $97,000,000."
NY
brief, p. 4. At the hearing, a State witness agreed with SSA's counsel
that "every penny in the final computation of the APL carried with it
about a $50,000 magnitude." Transcript of hearing, Vol. III (Tr. III),
p. 574. /2/ We have not explored how the five cases mentioned by the
auditors relate to the 11 cases in issue here, because the parties both
deal with the 11 and because the auditors' determinations generally
relate to the "income". issues in dispute. Actually, there were issues
regarding imputed or in-kind income in many of the cases under scrutiny
in the dispute; the State chose not to challenge some of the SSA
determinations. /3/ The State also reasonably can be said to
have a burden to facilitate SSA decision-making by providing information
which SSA needed to make a decision under section 416.2085( a)(2).
There is no evidence or argument that the State failed to do so.
/4/ Testimony of Martin Burdick, TR. I, pp. 134, 142-3, 147; Burdick
Affidavit (NY brief, App. II), P6, 16, 23, 25. /5/ However, although we
agree with New York generally that the cases with "no other income"
will
tend to be higher in value than cases with income, we do not agree that
only the highest value cases should be included in the calculations. It
seems reasonable to conclude that Congress intended to include all cases
where the recipient was totally dependent on AABD payments, even if
those payments might be lower in a given case because the recipient had
less needs due to factors other than lack of income; e.g., lower rent
payments or less costly care. /6/ Both parties' appendices contain
considerable material which was not paginated. We show the
numbers the pages would have had if they had been numbered (starting
from "one" in each tab). /7/ It is not quite as clear as in Case 9
that
the circumstances here remained the same in 1972 as in 1970, which is
the date borne by some of the documents in the record here showing that
the recipient was paying $35 a month to her mother. We note an
indication in the record that "Recipient remained at the above address
throughout January on (sic) 1972 and continued to report her monthly
rent as $35.00." SSA brief, Volume III, Tab 13(A), p. 1. SSA did not
dispute this. SSA brief, pp. 89-90. /8/ Mr. Reilly's affidavit
contained some details on processing of three other cases, and
cross-referenced another affidavit dealing with those three cases in
relation to the Duane Street Composite Roll. Nonetheless, only paragraph
6 of Reilly's affidavit is relevant to Case 4. /9/ The State
presented its argument on this case more clearly in the post-hearing
brief than in earlier materials, so the Board conducted a post-hearing
telephone conference to discuss the case and give SSA a chance to
respond. A tape of the conference is in the record. /10/ The
record contains an indication that rent could be paid during short
hospitalization to protect a right of occupancy. NY brief, App. II, Ex.
N, p. D-2. No one argued this in this case, and the record contains no
indication that it happened here. /11/ Earlier, New York had
maintained, as it continued to maintain in this appeal, that these three
cases also were excludable on other grounds discussed elsewhere in this
decision. /12/ TR. I, p. 4. SSA alluded to a question of
impropriety of Mr. Wachtel's participation as both a witness and counsel
in this case. SSA brief, p. 97. We do not reach this issue since we
find against the State on these cases in relation to the basis pressed
here. /13/ Before the appeal, New York apparently had argued
that this case should be excluded on the basis of imputed income, but on
appeal did not press that argument. SSA brief, App. IV, Tab 26; cf. NY
brief, App. i, Tab 26; SSA brief, p. 100; and NY post-hearing brief,
pp. 24-5, A-9. The imputed income argument rested on an indication in
July 1971 that the recipient did not have access to cooking facilities
in his apartment, and the argument that this required an inference that
someone, therefore, must have been giving him meals if he was not
getting a restaurant allowance (as he was not, except arguably for
January 1972, and we reject this argument). The record contains no
indication that the recipient got meals from anyone. Although we need
not decide the issue, since New York abandoned this position on appeal,
on the basis of the record here it does not appear that we could agree
that there was imputed income. Unlike the other imputed income cases,
there is no indication whatsoever of any assistance from another person,
and it is just as reasonable to speculate that the recipient eventually
got access to a kitchen. /14/ We also discuss with the "upstate
room and board" subgroup one case (no. 2) which we believe is related
although the parties argued the case on other grounds. /15/ This
regulation provides that the average money payment to be used in
calculating the APL is to be determined generally "by computing the
average of the cash payments made for basic needs and special needs, . .
. to individuals who were not living in the household of another, whose
support and maintenance was not contributed to by another, and who had
no other income." /16/ The dictionary definition of the term
"household" is "those who dwell under the same roof and compose
a
family. Webster's Third International Dictionary, 1976 Ed. (emphasis
added). A similar definition of the term "household" as "individuals
living as a family unit" appears in the SSI regulations, but in a
different subpart than the regulations cited by the State. 20 CFR
416.1003(b) (1975). /17/ The APL Determination Guide issued by SSA in
January 1974 stated: The definition of cases that are to be included in
developing the average money payment is rather restrictive. This means
that only a relatively small percentage of cases will actually be
considered in this calculation. Therefore, since the State does not
have the systems capability to identify only those cases that meet this
definition, it should use any means it can to automatically exclude as
many cases as possible that do not meet the definition. The remaining
cases will have to be manually screened to isolate those cases that do
meet the definition. NY appeal file, Tab 1 (emphasis in original).
/18/ This does not mean that we agree with the State's proposal that we
should also exclude all of the cases which were included under the
parties' 50/50 split compromise. See NY post-hearing brief, p. 11, note
1. That treatment was not necessarily inappropriate, and we see no
reason why New York should not be considered bound by its agreement
regarding those cases. /19/ The parties subsequently agreed to
revise the definition of "shared rent." The revised definition of
"shared rent" was "the total rent for the apartment is known
and the
client is budgeted for a pro-rata share or other reasonably equal
allocation of the shelter cost." Letter of July 12, 1976 from the State
Community Relations Specialist, SSA, to Director, SSI Unit, NYDSS,
enclosure to SSA's February 13, 1984 submission. We discuss below the
definition of "non-shared rent" and its significance for this dispute.
/20/ When the parties agreed to a revised definition for "shared
rent," they did not specifically revise the definition of "non-shared"
rent in the coding instructions. That definition was: 1. The total
rent is known or verified and the client is budgeted for the full
amount; or 2. The case folder record says specifically that the rent
is not shared. We do not adhere strictly to this definition below
because it is inconsistent with the revised definition of "shared rent."
A corresponding revision would have stated that "rent was not shared if
total rent was known and client was budgeted for more than a pro-rata
share or otherwise reasonable allocation." Moreover, the October 30,
1975 letter indicated that, if there was no friendship involved, "clear
evidence that rent was shared" was needed in order to exclude a case. NY
brief, App. I, Ex. G. We also note that the possibility of shared rent
arises only where both the recipient and another occupant of a dwelling
unit do, in fact, pay rent. /21/ The SSA analyst apparently also
refused to exclude New York City cases on the basis of "insufficient
data" because he said that all of the cases where a definitive
determination could not be made had already been identified as part of
the 50/50 compromise split. This ignores the fact that New York would
not have identified these cases as insufficient data cases because the
State initially thought that it had definitively determined the cases
were excludable on one of the agreed-upon grounds. /22/ Nothing
in the parties' agreement indicates that reviewers' decisions regarding
shared living could never be based on inferences. Indeed, the October
30, 1975 letter from the State setting out the factors for deciding
"shared living" cases made these decisions dependent on "indications"
or
"information" concerning the factors (except that "clear evidence"
that
rent was shared was needed when the recipient lived with a person not
identified as a friend). NY brief, App. II, Ex. F. /23/ We note
that SSA never argued that an $84 grant, no-friendship case was per se
excludable, and that position would not be consistent with the parties'
agreement. Implicit in the agreement is a recognition that sharing
might warrant exclusion even if the State did not lower the basic grant
and rent allowance as a direct result. Moreover, the State's expert
testified that a caseworker might ignore factors indicating sharing in
order to avoid reducing the recipient's benefits, so long as the rent
requested was reasonable. Thus, we do not consider any inference drawn
solely from State determinations about these matters to carry much
weight. /24/ The State's expert testified that, if there was a true
sublet arrangement (i.e., no sharing) it was likely that this would be
reflected in the file through an award of a restaurant allowance or
other notations. The point seemed to be that absence of a restaurant
allowance strongly implied sharing. We disagree. The example given in
the NY Deputy Commissioner's October 30, 1975 letter of a case
includable as a subletting arrangement was "Client rents a room from the
prime tenant, has kitchen privileges." This is significant because at
one point in the case review process New York was excluding cases on the
basis of food preparation and purchase in common where there was no more
than a notation in the file that the recipient had use of kitchen
facilities. SSA objected, and the State in effect conceded, that mere
use of kitchen facilities did not indicate food sharing. The
correlative of this is that mere absence of a restaurant allowance does
not necessarily indicate food sharing. Thus, we think there is little
significance to absence of a restaurant allowance. /25/ SSA also
pointed out that it was based on excerpts rather than the entire case
files. We agree with the State, however, that this is meaningless in
the absence of a showing that the excerpts were selective. SSA alleged
that it had included with its individual case submissions more documents
than New York submitted. This is true. But many, if not most, of these
documents are not case file excerpts. Our comparison of the submissions
leads us to conclude that, while New York omitted file documents, there
is no basis for concluding this was materially misleading. More
important, as discussed below, the thrust of the State expert's
testimony concerned patterns of living arrangements typical of AABD
recipients, and specific cases were discussed primarily in that context.
We give greater weight to his testimony on patterns than his testimony
on specific cases. /26/ Although the lack of increased rent
could be explained as plausibly by rent control as by the relationship
between the recipient and Mrs. W., the auditors decided that the shelter
was not rent controlled. There is also some indication in the file that
the apartment building was "rundown." However, the caseworkers' notes
from home visits over the years consistently describe the apartment and
the recipient's room as well-furnished and nicely kept. /27/ On
the other hand, it is not impossible that the recipient did some
housekeeping for Mrs. W. over the years: after another visit, a
caseworker noted merely, "she was at home doing her housework" (p.
29);
she had been formerly employed as a domestic; although she had a heart
condition and other ailments, the State had in 1967 reclassified her
from disabled to Home Relief (in 1972, she was receiving assistance due
to old age), so she might have been capable of doing light housework.
While we do not consider this possibility sufficient as a basis for
concluding that Mrs. W. kept the rent low in exchange for housekeeping
services (see our discussion in the imputed income section above), it at
least prevents us from concluding definitively that no such sharing was
taking place. /28/ Arthur Young had excluded this case on the
basis of shared or pro-rated rent, although we do not know why. The
auditors did not agree with this conclusion, but the audit workpapers
show that, at least, the auditor who initially reviewed the papers did
not find affirmatively that there was no shared rent but said instead
that there was "insufficient data" on this question since the total
rent
was not known. Id., p. 18. /29/ The State further pointed out
that the rent the recipient paid was $7.50 less per week than the
recipient paid when he resided in a hotel for a period and said that
this "gives rise to a presumption of imputed income in the form of below
market rent because the landlady felt protective towards (the
recipient)." Id. The State did not specifically categorize this as an
"imputed income" case, so we take this argument, in context, to merely
be that the recipient received a benefit from his relationship with the
landlady. /30/ This record raises some question of the veracity
of the recipient's statements, however. Not only did he give
conflicting information on the use of kitchen facilities, but he may
have given misinformation about whether he had actually received one of
his AABD checks (possibly the one he sought replacement for in the
December 27 visit). Indeed, the State originally sought exclusion of
the case on the ground of recipient fraud although it did not pursue
this basis on appeal to us. /31/ Apparently the caseworker
originally budget the recipient for $39 bi-monthly, which is slightly
more than two times $18 per week, which is what Mr. E. wrote he was
charging the recipient. This is consistent with how rent was treated in
other cases before us, where the semi-monthly amount is more than two
times the weekly rent, presumably to account for the extra days in the
month. For some unexplained reason, however, the budget for January
1972 for this recipient shows only $36 per month. At an earlier stage
in the case, the State argued that this gave rise to an inference that
rent was shared because the only possible explanations were either that
Mr. E. had decided to let the recipient live "rent free" for part
of the
month or that Mr. E. took in another tenant who was paying part of the
rent. Id., p. 18. We think both of these scenarios are entirely too
speculative and no more reasonable than other possible explanations
(such as that a new caseworker simply multiplied the $18 by two and
forgot to add an extra amount). Moreover, they are contradicted by the
evidence that rent was not pro-rated. /32/ While the recipient
apparently still living in Mr. E.'s apartment in June 1972, we
do not know if she would have continued to live there. Her husband
reported to a caseworker that she died in Harlem Hospital on July 6,
1972. Id., p. 33. /33/ We think this affidavit lends little to
the analysis here because: (a) it was made more than 10 years after the
relevant time period; (b) it fails to address the key question of the
relationship of the recipient and Mr. G.; and (c) the physical
configuration of the apartment (i.e., whether it was renovated) is of
less significance than how the apartment was furnished in January 1972.
Since we do not base our decision on the affidavit, we do not decide
whether it should be striken from the record, as SSA argued it should
be. /34/ In 1957, the recipient was described as the "only
boarder" (Id., p. 6), but caseworker notes from a home visit in November
1968 state: "The apartment belongs to a Mr. (S.) who rented another
room to another man." SSA appeal file, Vol. IV, IV, Tab 26, p. 18. We
do not think the possibility that there was another tenant alters our
analysis, however. /35/ The State had argued to SSA that, since the
room had no cooking facilities, and the recipient received no
restaurant allowance, someone must have furnished his meals (a form of
imputed income). The State did not pursue this argument on appeal, and,
in any event, we would not be persuaded by it. Absent any additional
information, it is more reasonable to assume that, although his room had
no cooking facilities, they were otherwise available to him than to
infer that someone fed him. /36/ We also note that the $86.50
which the recipient was paying per month is close to the $90 per month
rent allowance figure which would trigger the need for a higher level of
State approval. NY brief, App. II, Ex. M, p. VII-16. /37/ A
letter from the family health center where the recipient was a patient
requested in March 1972 that the recipient be given housekeeping
services since "he lives alone and is not able to properly care for
himself, . . . ." Id., p. 24. One could imply from this that he was not
getting any assistance from the other person in the apartment; on the
other hand, it is possible that he had received some help from that
person prior to the request since a caseworker had noted as early as May
1969 that he did not function very well and was unable to fix his own
meals. Id., p. 16. /39/ The term "measurement error," as the
parties used it here, refers to errors in the universe which may be
attributable to, for example, tabulation or sorting errors, wrong
answers, and, generally, erroneous classification of material. NY
post-hearing brief, p. 28; TR. III, pp. 520-1, 715-36. Cf., B. J.
Mandel, Statistics for Management, Dungary Publishing Company, Inc.,
Baltimore, Maryland (1977), pp. 212-4. "Sampling error" refers to
the
difference between the true value of a thing in the universe and the
value of that thing as estimated from a sample. "Sampling error is an
inherent part of sampling systems and cannot be eliminated as long as
the estimate is based on a sample," although it is subject to laws of
probability and therefore can be effectively reduced. Id., p. 214; TR.
III, p. 522. /40/ The error rate for Stratum I was about 5%.
TR. IV, p. 743. This figure apparently did not reflect 36 cases where
the State claimed a case should be included but had incomplete or
missing files, and also does not reflect exclude errors. New York tried
to make a point that since SSA corrected the auditors' determinations by
5 cases, the auditors had a 10% error rate. TR. IV, p. 746. This is
misleading; the sample, when used, had been corrected, in effect, for
measurement error by deletion of the 5 cases, whereas the universe was
never corrected. How meaningful and substantial the error rate in the
universe is appears to be a subjective determination. /41/
Moreover, for those cases where we reverse SSA's determination to
include the case, we are not necessarily disagreeing with the auditors'
determination that the State's original basis for excluding the case was
improper. /42/ In addition to subjective reactions to the witnesses, we
agree with SSA's characterization of Drs. Arkin and Greer as having more
qualifications and experience with statistical sampling issues in the
auditing context we have here. SSA post-hearing brief, pp. 20-1. /43/
There was testimony from a State expert that the sample size for
Stratum I was "quite a large sample size" and "large enough."
TR. III,
pp. 628-30. /44/ SSA also tried to impeach Dr. Heiner's
testimony by implying that the figures given to him by the State for
Chart 1 were not accurate; specifically, SSA suggested that the correct
universe figure for category 0 should be 781 rather than the 571 figure
the State used. However, the exhibit SSA pointed to does not, when read
as a whole, support this theory since the 781 figure appears to identify
any case excluded based on the category 0 (which would encompass cases
coded 0= 0and 0=1), whereas the 571 figure only pertains to cases coded
0=1. NY appeal file, Ex. 5, pp. 9-10. /45/ The State explained this as
follows: After the questionnaires were completed, the data was
transferred to an APL Master File, which was then scanned by different
computer programs. The "Exclusion Program" excluded cases based on
various question-response combinations. The "Computation Program"
scanned the same master file and rejected cases or payments on the basis
of various inappropriate payments amounts and conditions (for example,
if a payment to the recipient was restricted, as indicated by the "Pay
Code," of if all payments for a case totaled less than $94). It
sometimes occurred that the same case would be excluded by both the
Exclusion and the Computation Programs. Such cases were not retrieved
for purposes of comparing the universe to the auditors' sample on a
category-by-category basis. Thus, the figure shown for each "subset"
in
Chart 2 is the average value of the cases excluded solely by the
Exclusion Program. Chart 2, note 1. /46/ SSA also relied on
testimony by Dr. Arkin that the discrepancy between the universe subset
and the sample values would be expected because the State's comparison
was between a universe figure and a figure for sample cases which had
been erroneously excluded, i.e., "misclassified." TR. IV, p. 822.
As
New York pointed out, however, this was based on Dr. Arkin's
misconception that the State had only compared the universe subset
figure with the sample errors. NY post-hearing brief, p. 35, n.1; TR.
IV, pp. 829-33. Chart 2 also compares all of the sample cases in each
category with the universe subset from which they were drawn.
Misclassification would not explain the differences shown in this
comparison. /47/ It is well-established that statistical
sampling evidence is generally reliable. See, e.g., University of
California - General Purpose Equipment, Decision No. 118, September 9,
1980, and cases cited therein; see also cases cited in SSA brief, p.
48. "The use of statistical samples to audit claims and arrive at a
rebuttable initial decision was reasonable where the number of claims
rendered a claim-by-claim review a practical impossibility . . . it was
not unreasonable to place the burden on the challenging party to present
evidence to rebut the statistical sample." Illinois Physicians Union v.
Miller, 675 F.2d 151, 155 (7th Cir., 1982), citing State of Georgia v.
Califano, 446 F. Supp. 404 (N.D. Ga., 1977). /48/ New York
presented the "difference estimator" in relation to Stratum I errors,
but we do not see any reason why its applicability, or inapplicability
in principle, is any different for Stratum II errors; presumably, it was
presented this way because New York believed that it had other
persuasive reasons to impeach Stratum II. /49/ SSA did
computations for Stratum I only as an interim step toward a weighted
mean, and the figures of both parties were subject to various
adjustments; we use the figures here not as final points, but as
illustrations of figures comparable in relative values. /50/
Under Board procedures, we may include in the record of an appeal
materials "such as evidence submitted in another appeal" after the
parties are given notice and an opportunity to comment. 45 CFR 16.21(
a). Since our decision on this part of the appeal is not dispositive,
but amounts to a remand, we are referring to this memorandum, and a copy
will be sent to the parties separately.
NOVEMBER 14, 1984