|
Department of Health and Human Services DEPARTMENTAL APPEALS BOARD Appellate Division |
|
| IN THE CASE OF | |
|
|
DATE: June 24, 2005 |
| |
|
|
|
Docket No. A-01-40,
A-01-41,
A-01-42, A-01-43, A-01-44, A-01-87, A-01-120 Control Nos. HI/01/001/MAP, IL/01/001/MAP, ME/01/001/MAP, LA/01/001/MAP, TN/01/001/MAP, TN/01/002/MAP, LA/01/002/MAP Decision No. 1981 |
| DECISION | |
|
DECISION In this decision, we address disallowances of federal Medicaid funds claimed by five different States. The Centers for Medicare & Medicaid Services (CMS) issued disallowances in January 2001, imposing funding reductions for expenditure periods going back to either October 1, 1992 or July 1, 1993. Each State appealed, and the cases were consolidated for Board consideration. After a lengthy period during which the parties tried to settle the cases, followed by extensive document discovery, the parties briefed the issues and, on October 21, 2004, presented oral argument to the full Board Panel assigned to the cases.
CMS based the funding reductions on a 1991 law addressing state taxes on health care items or services. Under the 1991 law, expenditures for which Medicaid funding is otherwise available must be reduced by the revenues from any "impermissible" tax, for purposes of determining the amount of federal funds to be paid to a state for any quarter. A health care-related tax is "permissible" if it meets certain requirements. Essentially, the requirements are that the tax be broad-based and uniformly imposed (unless these requirements are waived), and that the state not hold taxpayers harmless, in any one of three ways described in the statute. The only requirements which CMS alleges the States' tax programs did not meet are the "hold harmless" provisions, specifically, the "positive correlation" test and the "guarantee" test. The 1991 law required CMS to obtain input from states on the implementing regulations, and this resulted in an interim final rule in 1992 and a final rule in 1993. Meanwhile, the States at issue here enacted programs they say were intended to comply with the new law, imposing taxes on nursing facilities effective between July 1, 1992 and July 1, 1993. Some were taxes on licensed or occupied beds, others on income or gross receipts. Each of the States also enacted a program either of grants to private pay patients in nursing facilities or of tax credits for such patients (although some of them ended their programs as early as 1993). Although CMS regional officials had questioned each of the States' programs by at least December 1994, CMS did not issue formal disallowance determinations until 2001. CMS then determined that the grants or tax credits to private pay patients constituted non-Medicaid payments to the nursing facilities that indirectly held the facilities harmless, violating both the positive correlation and the guarantee tests. The States argue on appeal that their programs do not violate either hold harmless provision, relying on the wording of the provisions and their history and context. The States also argue that they were prejudiced by CMS's lengthy delay in taking the disallowances, in two ways. First, they allege that the delay compromised their ability to challenge the disallowances. Second, they allege that requiring them to return funds that were long ago collected and expended would have a devastating impact on their current Medicaid program operations, given the huge amounts involved and current budget demands. They further argue that they reasonably thought that CMS had ultimately agreed that their tax programs were permissible, in light of CMS's failure over many years either to issue any clear policy interpretation or to formally reduce their Medicaid funding. We do not reach the State's arguments about the delay, because we reverse on other grounds. As we explain in detail below, we conclude that the position CMS advanced in these disallowances is inconsistent with the wording of the 1991 law, as interpreted in the regulations and in the preambles to the regulations. Briefly stated, a hold harmless provision is in effect if a state makes a non-Medicaid payment to a taxpayer that is "positively correlated" to the amount of the tax. The preamble to the final rule stated that the term "positive correlation" is used in its statistical sense. Thus, the term "positive correlation" means a relationship in which one variable increases as the other variable increases. CMS concedes that it did not base the disallowances on any statistical analysis. This failure would not have been dispositive if CMS had focused on the correct variables and had made a finding that a positive correlation between those variables automatically followed from the provisions of State law, thus obviating the need for any statistical analysis. Instead, CMS determined merely that the States' grants or tax credit programs were "related to" or "associated with" the States' nursing facility tax programs, based primarily on factors such as when the programs were enacted and what the legislative intent was. In enacting the 1993 final rule, however, this Department specifically rejected the use of such subjective factors to establish a positive correlation. In its arguments before the Board, CMS seeks for the first time to show a positive correlation in the statistical sense. In spite of having had ample opportunity to obtain information from the States and to develop the record, however, CMS does not rely on the type of correlation analysis used in statistics. Instead, CMS relies on charts that compare the wrong variables. The statute refers to a positive correlation between the amount of a non-Medicaid payment to taxpayers and the "amount of the tax," which the regulation interprets as the "total tax cost" to the taxpayer. In other words, these are the two variables that must be examined to determine if one increases as the other does. Yet, rather than examining whether the amount of the non-Medicaid payment varies as the total tax cost varies, CMS compares the rate of the non-Medicaid payment to the tax rate or compares the per bed amount to the per bed tax amount. These comparisons do not take into account other factors that would affect the relevant amounts (such as variations in the bases to which the rates are applied or variations in the number of beds occupied by private pay patients compared to the total number of beds in a facility). Thus, even CMS's belated analysis does not meet the terms of the regulation. CMS also cites an example (in the preamble to the interim final rule) that mentions grants to private pay patients, as a basis for saying that any grant or tax credit to a private pay patient violates the positive correlation test. The regulatory preambles, however, do not support a conclusion that a grant payment alone constitutes a hold harmless provision, as CMS seems to be suggesting. Given relevant differences between the preamble example and the States' programs, the context in which the example appears, and, most important, the wording of the statute and regulations (requiring both a non-Medicaid payment and a positive correlation), CMS's reliance on the preamble example is misplaced. Thus, we conclude that CMS's position on the positive correlation test is inconsistent with the statute and regulations (by which we are bound), and that the CMS findings are not legally sufficient under that test as a basis for reducing the States' funding. A hold harmless provision is also in effect, however, if a state provides (directly or indirectly) for a payment that "guarantees" to hold the taxpayer harmless. The regulations implemented this provision by adopting a two-prong test for determining whether there is an indirect guarantee provided by a state. Although CMS's findings were that the States' programs contained indirect guarantees (or otherwise indirectly held taxpayers harmless), CMS did not find that any of the States' programs failed the two-prong test (and most of the States' programs met the first prong because the tax rate was considered to be at or less than the average level of taxes applied to goods and services other than health care-related ones). The officials issuing the disallowances apparently thought that an indirect guarantee could be found other than by applying the two-prong test, but this is contrary to the plain terms of the governing regulation. The interim final rule specified that the indirect guarantee test applies if there is no "explicit" hold harmless guarantee. Contrary to what CMS argues, the final rule did not substantively change this approach. As the States point out, their grant or credit programs were structured so that they provided no explicit or direct assurance of any payment to a taxpayer provider. While CMS now asserts that the States did provide direct guarantees, CMS does not point to any wording in the States' programs that could reasonably constitute an explicit or direct assurance of any payment to the provider taxpayer. Instead, CMS argues that the preamble to the interim final rule made clear that a direct guarantee would be found if grants or tax credits for private pay patients were used in a way that guaranteed any indirect payment to the nursing facilities. The preamble statement on which CMS relies, however, was not describing when an explicit or direct guarantee would be found, but was referring to an "indirect guarantee" (to be identified by applying the two-prong test). Moreover, CMS's current view of the guarantee test as a broad catch-all provision is contradicted by the history of the provision and the implementing regulation. Since the statute specifically permits use of health care-related taxes to increase Medicaid reimbursement and uses the term "hold harmless," this Department considered the statutory language in context and chose to adopt an indirect guarantee test that would permit states some flexibility in their tax and payment programs. CMS cannot reasonably create and apply a new and broader indirect guarantee test, outside of the regulatory framework, and then try to justify it under the guise of being a direct guarantee test. Thus, we conclude that CMS's position on the guarantee test is also inconsistent with the regulations (by which we are bound) and that the CMS findings regarding that test are therefore not legally sufficient as a basis for reducing the States' funding. In reaching our conclusions, we reject CMS's characterization of the States' programs as "scams" or "schemes" to improperly increase federal Medicaid funding, since CMS presented no adequate basis for finding that the States' programs violated the terms of the 1991 law and has never even clearly articulated how the States' programs contravene the intent of the law. While CMS had previously expressed concerns that the States were holding taxpayer providers harmless by returning part of the taxes through increased Medicaid payments and the remainder through grants or tax credits, CMS made no findings regarding the States' Medicaid payments here, relying only on the non-Medicaid payments. Other factors CMS cited in questioning the States' programs over the years, such as the intent of the State laws, are largely irrelevant under the 1991 law. Indeed, in implementing that law, the Department specifically rejected a proposal that subjective factors such as intent be used to determine whether a hold harmless provision is in effect in a state. Finally, some of the factual findings on which CMS based its determinations for particular States are not supported by the record. Nor does the record support the position CMS took before us that the taxpayers were the private pay patients, rather than the providers. This claim is inconsistent with CMS's own audit findings, with the wording of the State laws, and with the regulations (which interpret "taxpayers" as "providers or others paying the tax"). The private pay patients may have born the burden of part of the taxes (if the providers increased private charges to cover the related tax costs, which did not always occur), but CMS provided no evidence that the private pay patients were "paying" the taxes at issue. In sum, CMS's position in this case is inconsistent with the statute, as interpreted in the Department's regulations and preambles. The record, moreover, does not support all of CMS's factual assertions, and, in any event, does not contain adequate evidentiary support for the findings the regulations require in order to justify a reduction in Medicaid funding. Accordingly, for the reasons explained more fully below, we reverse the disallowances. Our decision below is organized as follows. We first set out the legal background, specifically, the 1991 law (starting on page 7), the interim final rule with relevant excerpts from its preamble (page 9), and the final rule with relevant excerpts from its preamble (page 12). In the analysis section, we first set out general conclusions. We describe what is not at issue here (page 17). We then explain in detail why we conclude that CMS's application of the "positive correlation" test in these disallowances is inconsistent with the interpretation of the statute in the Secretary's regulation and in the preamble to that regulation (page 18). We next explain in detail why we conclude that CMS's application of the "guarantee" test in these disallowances is inconsistent with the wording of the regulation and its history (page 34). Following that, we discuss the intent of the 199l law (page 42). After explaining our general conclusions, we turn to the individual States. For each of the five States, we set out the State provisions at issue establishing the tax program and the grant program or tax credit program that CMS says rendered the tax impermissible. We describe the history of the disallowance action, including CMS's correspondence with the State and findings regarding the tax program, and then address for each State why we conclude that CMS's case is not legally or factually sufficient as a basis for concluding that the tax was impermissible under either the "positive correlation" or the "guarantee" test. In the heading for our discussion regarding each State, we give the docket numbers, and the disallowance period. The pages on which our discussion of the various States' programs begin (and the total disallowance amounts) are as follows:
Since our decision is based on the statutory and regulatory language of the hold harmless provisions and on the preambles to the interim final and final rule, we set them out in considerable detail in this section.
As the record in this case indicates, the Medicaid Voluntary Contribution and Provider Specific Tax Amendments of 1991 (Public Law No. 102-234) were intended to resolve a lengthy controversy between CMS and states about taxes states impose on health care items or services. (1) CMS believed that such taxes were being used to artificially inflate federal Medicaid funding to states and had proposed regulations to reduce Medicaid funding if states imposed any health-care related taxes and made any payment linked to those taxes. States considered this an interference with their taxing authority and obtained congressional moratoria on CMS's proposals. Ultimately, CMS and the states reached a compromise that was adopted almost verbatim in the 1991 law. CMS viewed the 1991 law as intended to stop state schemes to inflate federal funding, and states argued that the statute protected them from CMS's overreaching by permitting health-care related taxes, with no reduction in Medicaid funding, so long as they met certain requirements. Essentially, the requirements were that the tax be broad-based and uniformly imposed (unless the state obtained a waiver of these requirements by showing that the tax was generally redistributive and met other requirements), and that the state not hold taxpayers harmless, in any one of three ways described in the statute. As amended by the 1991 law, section 1903(w)(1)(A) of the Social Security Act provides in relevant part that--
Paragraph 1903(w)(4) provides:
The provisions of this paragraph shall not prevent use of the tax to reimburse health care providers in a class for expenditures under this title nor preclude States from relying on such reimbursement to justify or explain the tax in the legislative process. Section 5(c) of Public Law No. 102-234 required the Secretary to consult with the states before issuing regulations to implement the legislation. The preamble to the interim final rule states that the agency "met this requirement by conducting a series of meetings with representatives of the National Governors Association, the National Council of State Legislatures, the National Association of Counties, the National Association of State Budget Officers, and the American Public Welfare Association" and, during these meetings, "received written and oral input from these groups concerning the issues involved in developing these rules." 57 Fed. Reg. 55,118, 55,139.
The interim final rule (with comment period) was published on November 24, 1992. The hold harmless provisions were implemented by subsection (f) of 42 C.F.R. � 433.68, as follows:
The preamble to the interim final rule explained this provision as follows:
57 Fed. Reg. 55,118, 55,129. The preamble discussion of the first hold harmless provision (the "positive correlation" test) does not define any of the terms used. Instead, it gives two examples of "the types of situations which might fall under the criteria," describing the first example as follows:
The preamble contains the following general statement about the "guarantee" provision:
Id. The preamble then explained the reason for the two-prong "guarantee" test described in the regulation:
Id. The preamble explained that 6 percent of revenues was chosen for the first prong because it was considered to be "the average level of taxes applied to other goods and services in the States" and that the 75/75 percent test was chosen for the second prong because "we think it strikes a reasonable balance between our need to assure that States do not use Medicaid rates to repay providers for tax costs in a way not permitted under the statute, and our desire to permit States flexibility in the design of their tax and payment programs." 57 Fed. Reg. at 55,129.
The final rule, responding to comments, was published on August 13, 1993, and made the following changes to subsection (f) of section 433.68:
58 Fed. Reg. 43,156, 43,182. The preamble to the final rule noted that "the majority of the commenters urged us to reconsider our positions regarding the hold harmless provisions . . . ." 58 Fed. Reg. at 43,166. In responding to a comment that the regulation should not "dictate how each State should use their hospital tax levies," the preamble stated:
58 Fed. Reg. at 43,157. Comments on the "positive correlation" provision (including the example in the preamble) and the relevant responses were as follows:
58 Fed. Reg. at 43,166-167. (2) Most of the comments on the hold harmless provisions addressed the two-prong test under the "guarantee" provision. The preamble clarified certain aspects of the test, including that "a tax that is broad-based and uniform, and applied at a rate of 6 percent or less, is considered a permissible health care-related tax under the first prong of the two-prong hold harmless test" and not just "presumed" to be permissible. 58 Fed. Reg. at 43,166. The preamble also addressed the change in the regulatory language that explains when the "two-prong" guarantee test applies:
58 Fed. Reg. at 43,167. The preamble noted that "when the regulations impose new requirements, such as the hold harmless provisions, we have extended the deadline for compliance" and that "we have extended the deadline for compliance of the hold harmless requirements to (30 days after publication date of this final rule)." 58 Fed. Reg. at 43,158. Formal policy guidance from CMS on the regulations has focused on other provisions of the 1991 law, providing little to no guidance on the hold harmless provisions. A letter to State Medicaid Directors, dated November 16, 1993, repeated the language of the three general hold harmless tests, and then described the two-prong test, stating: "The regulation at 433.68(f)(3)(i) specifies a two-prong test if an explicit guarantee does not exist." JAF Ex. 54, at 10. A June 21, 1995 State Medicaid Directors' letter only briefly addressed the hold harmless provisions. In response to a question about including a health care-related tax as an allowable cost in determining reimbursement rates, the letter stated:
JAF Ex. 59, at 8. A State Medicaid Directors' letter issued October 9, 1997 did not directly address the hold harmless provisions, although it did clarify that taxes based on occupied beds or patient days would be considered to be uniform. JAF Ex. 64. This letter also stated that CMS was seeking more authority from Congress to work with States "to resolve current tax liabilities," but would move forward to apply the current law if legislation was not enacted by August 1998. Id. at 3. |
|
| ANALYSIS | |
|
At the outset, it is important to note what is not in dispute here. First, CMS does not claim that any of the States' tax programs violated the requirements that the tax be broad-based and uniformly imposed. Second, CMS does not claim that the States were using Medicaid payments to hold the taxpayers harmless, nor that the grants or tax credits were positively correlated to the difference between the Medicaid payment and the total tax cost. Transcript of oral argument (Tr.) at 29. CMS also does not claim that any of the States' programs failed to timely pass the two-prong indirect guarantee test under section 433.68(f)(3)(i). Most of the taxes at issue were applied at a rate that produced revenues less than or equal to 6 percent of the revenues received by the taxpayer (or were timely modified to comply), and therefore were permissible under the first prong of the test. In other words, since the taxes were imposed at or less than the rate that was the average level of taxes applied to other goods and services in the States, the taxes were permissible under the indirect guarantee test regardless of how much of the tax was in fact repaid to the providers in the form of Medicaid payments or otherwise. (3) These concessions by CMS do not end our inquiry since CMS asserts nonetheless that there was a hold harmless provision in effect in each State's program, but, as our discussion below makes clear, these concessions undermine CMS's position in several ways. As mentioned above, we do not reach the issue of whether CMS's delay in formally taking these disallowances precludes it from doing so now. We note, however, that although the delay was not as lengthy as the States portrayed it, the history of the interactions between CMS and the States (discussed in the individual State sections below) undercuts CMS's position that the States should have known from the face of the statute that their tax programs were impermissible. The States question why, if that was so clear, CMS took so long to issue the disallowances. CMS has provided some explanations for parts of the delay, but they are not entirely persuasive. The record shows that CMS officials were aware of questions about what statutory terms such as "positively correlated" and "hold harmless" meant, and were long aware of the States' programs. Yet, as the States argue, the interpretations CMS now advances in support of the disallowances were not set out in any official policy issuance, and are different from the reasons CMS initially gave for questioning the States' programs. Below, we first set out the general reasons for our decision. We then provide a more detailed analysis of each of the States' programs at issue, including an analysis of whether the CMS findings about those programs are supported by the record and consistent with the States' laws.
Under section 1903(w)(4)(A) of the Act, a hold harmless provision is in effect if the Secretary determines that the State "provides (directly or indirectly) for a payment (other than under this title) to taxpayers and the amount of such payment is positively correlated either to the amount of such tax or to the difference between the amount of the tax and the amount of payment under the State plan." As we explain below, the fundamental problem with the CMS position that the States' tax and grant or credit programs created a hold harmless under section 1903(w)(4)(A) is that it fails to give effect to key terms in that provision, as defined in the regulation or interpreted in the preamble to the final rule. Specifically, CMS did not apply the terms "positively correlated" and "amount of such tax" in a manner consistent with the regulation and its history. Instead, CMS decided that the positive correlation test was violated based primarily on the preamble example referring to grant programs. Contrary to what CMS argues, CMS cannot reasonably rely on the preamble example as an independent basis for determining that any grant or tax credit to private pay patients automatically renders a related tax impermissible. The example applies the regulatory language only to particular facts, which differ from the facts here. Any implications drawn from the example must, moreover, give effect to all of the regulatory terms - a standard that CMS's interpretation does not meet.
The term "positively correlated" is not defined in the statute or regulation, but it is not disputed that it is a term of art used in statistics. The States provided the following explanation from a statistics text:
JAF Ex. 2, at 2 (emphasis added). (4) The preamble to the final rule responded to comments that the regulation was vague in describing positive correlation by explaining:
58 Fed. Reg. at 43,167 (emphasis added). CMS describes this preamble statement as "less than precise." CMS Br. at 46. The context, however, indicates that the statement was an attempt to be more precise and to eliminate any vagueness. Such guidance was critical, as the record amply demonstrates, since once states enact tax programs, they cannot change them until at least the next session of the state legislature and since the consequences on the Medicaid recipients of any retroactive disallowance can be a reduction in the services they receive. Moreover, CMS does not deny that the term "positive correlation" is a term of art in statistics and use of the word "correlated" by itself certainly connotes something more than a mere relationship or association. CMS argues that it is not required to engage in a statistical analysis in order to find that a positive correlation exists. This is true (and we reject the States' argument to the contrary) since there may be situations in which a positive correlation in the statistical sense necessarily results from a systematic connection between the two variables in question. See JAF Ex. 1, at 3. If a state provided for a non-Medicaid payment that automatically increased when one of the other variables in question increased, then a systematic connection would necessarily exist. The preamble to the final rule addressed this as follows:
58 Fed. Reg. at 43,167 (emphasis added). If such an automatic (or systematic) correlation between the two variables is not evident on the face of the State law provisions, however, then a statistical analysis of the relationship of the two variables (the "two sets of scores") would be required to show a positive correlation in the statistical sense. CMS also argues that since no mathematical or numerical test is specified in the regulation (unlike for other provisions, such as the indirect guarantee provision), this is evidence that no statistical analysis was ever contemplated. The preamble clearly states, however, that the term "positive correlation" is being used in its statistical sense. Since there are commonly accepted statistical methods of measuring correlation (such as the "correlation coefficient r"), moreover, there would be no need to specify such tests in the regulation itself, whereas there was a need for the regulation to specify the other cited tests, which are unique. (5) Here, CMS neither did any statistical analysis of the correlation between the two relevant variables, nor found that one variable would increase automatically as the other increased. Instead, CMS argues that we should find that the positive correlation test was met because of the "relationship" or "association" between the States' health care-related tax programs and grant or credit programs. See, e.g., Tr. at 22-23, 62-63. To evidence this "relationship" or "association," CMS relies on factors such as the simultaneous authorization of the provider tax program and the grant or credit program, the stated intent that the grants or credits were to benefit the private pay patients who otherwise would bear the burden of the tax, and the fact that the providers were not generally precluded from passing the cost of the tax through to private pay patients. The preamble to the final rule specifically rejected using such subjective factors, however, stating that this "would result in a lack of specific standards by which hold harmless could be measured," and "would be administratively burdensome and virtually impossible to apply fairly throughout the nation." 58 Fed. Reg. at 43,167. (6) Indeed, CMS had informed the States early on in the negotiations over the hold harmless provisions of its position that statistical tests should be used, and this approach was merely reaffirmed in the regulation. See, e.g., JAF Ex. 26, at 4. CMS also points to the following sentence in the preamble: "If a provider is receiving a non-Medicaid payment for its tax cost, there would be positive correlation between these two variables, or a hold harmless would exist." 58 Fed. Reg. at 43,167. CMS argues that, therefore, "if the provider is receiving a payment (directly or indirectly), in the form of a grant or tax credit for the costs of the tax, CMS would find that a positive correlation exists." CMS Joint Brief (JBr.) at 15. Such an unqualified conclusion is not warranted, however. The preamble statement refers to "a provider" receiving "a non-Medicaid payment for its tax cost." (Emphasis added.) Thus, this sentence presumes a one-to-one (or dollar-for-dollar) relationship, where the non-Medicaid amount each provider receives is equal to the amount of its tax cost. When there is such a relationship, one can determine that a positive correlation exists without doing a statistical analysis because, by definition, the non-Medicaid payment to a provider would automatically go up as the provider's total tax cost went up. This statement cannot reasonably be read, however, to mean that CMS would always find a positive correlation to exist whenever any non-Medicaid payment is made, no matter what its relationship is to the tax cost to the provider taxpayer. Moreover, as we discuss below, this sentence focuses on the tax cost to the provider and its relationship to the non-Medicaid payment, whereas CMS here focuses on the rates of the tax compared to the rates for the grant or credit program or on the per bed amounts. CMS does not take into account other variables that could affect either the amount of the tax cost to a provider or the amount of the grants or credits to private pay patients in the provider facility and thus affect whether a positive correlation would exist between the two amounts, that is, whether one would increase as the other increases.
Section 1903(w)(4)(A) refers to a non-Medicaid payment being positively correlated either to "the amount of such tax or to the difference between the amount of the tax and the amount of payment under the State plan." The implementing regulation at section 433.68(f) provides:
(Emphasis added.) In this provision (unchanged since the interim final rule), the regulation interpreted the statutory phrase "the amount of the tax" to mean "the total tax cost." When asked whether the phrase "amount of such tax" in the statutory provision meant the same thing as the phrase "amount of the tax" in that provision, CMS at first conceded that it did. Tr. at 89. When further queried about whether it followed that "amount of such tax" therefore also meant "the total tax cost," CMS attempted to withdraw its concession. Tr. at 92. CMS provided no explanation why the two phrases should have a different meaning, however. Under principles of statutory construction, the phrases would be construed as intended to have the same meaning because the same wording is used. Moreover, while one phrase in the statute refers to "such tax" and the other to "the tax," the preamble to the interim final rule treated them as the same, describing the issue as whether "the amount of the payment is positively correlated either to the amount of the tax or to the difference between the amount of the tax and the amount of the Medicaid payment." 57 Fed. Reg. at 55,129 (emphasis added). Since the regulation interprets "amount of the tax" as "total tax cost," CMS was required to use that as one of the variables for purposes of determining whether a positive correlation existed, but did not do so. The history of the regulatory provision indicates that CMS agreed with the States very early on that the positive correlation test looked at the total tax cost as one of the variables. JAF Ex. 25, at 7. (7) This makes sense as the amount to be compared to the amount of any non-Medicaid payment to the taxpayer to determine whether the taxpayer is being held harmless. Use of the total tax for this purpose is consistent with the use of "the amount of the total tax paid" in the second hold harmless test (not at issue here). See 42 C.F.R. � 433.68(f)(2). The 75/75 percent test, moreover, looks at what percentage of individual taxpayers in a class receive what percentage of the individual taxpayer's "total tax costs back in enhanced Medicaid or other State payments." 42 C.F.R. � 433.60(f)(3)(i). (8) The preambles to the interim final and final rules repeatedly refer to providers (or taxpayers) being held harmless for "their tax costs" or a provider being held harmless for "its" tax costs. Thus, we conclude that the relevant variable is the total tax cost to the taxpayer. That conclusion is clearly suggested by the regulation and the preambles, read as a whole, and CMS has never issued any policy document containing a contrary interpretation. Instead, CMS relies almost exclusively on the preamble example reference to grant programs for its conclusion that the States violated the positive correlation test. As we discuss next, this reliance is misplaced.
We first note that the preamble example does not explicitly define any of the terms in the positive correlation test. Instead, any notice from the example of how to interpret the test must be parsed from what may be reasonably inferred from the example, consistent with the language of the regulation itself and its context and purpose. CMS reads the preamble example as follows:
CMS Jbr. at 14. According to CMS, "in the preamble to the regulation, the Secretary tried to set forth the fact that these . . . . granny grant programs would no longer be permissible." Tr. at 108. CMS's reading is inconsistent with the wording of the preamble, however. First, CMS fails to consider (and in its brief fails to quote) the lead-in language, which describes the example as one "of the types of situations which might fall under the criteria." 57 Fed. Reg. at 55,129 (emphasis added). (9) Second, the example describes a situation in which the variable to which the non-Medicaid payment correlates is the difference between the total tax cost and the Medicaid reimbursement. Specifically, the example is as follows:
(Emphasis added.) Here, however, CMS agrees that it is not now relying on the relationship of the grant amounts to the difference between the total tax cost and the Medicaid reimbursement. Tr. at 29. Even if it were, moreover, the conclusion in the preamble that "[i]f the tax is considered to be levied on the nursing home, the State is using non-Medicaid funds to compensate nursing homes indirectly, for the cost of the tax imposed on private charges" does not specifically state that, in that situation, the compensation is positively correlated to the difference between the Medicaid reimbursement and the total tax cost. The variables might be positively correlated. For example, if the increase in Medicaid reimbursement to each nursing home was equal to the full amount of the tax cost to the nursing home for non-private charges, then the non-Medicaid payment for the cost of the tax on private charges would necessarily increase as the difference between the Medicaid payment and the total tax cost increased. But the variables might not be positively correlated if the increases were distributed differently. Third, CMS relies heavily on the alternative conclusion, following the example, that "[i]f the tax is considered to be levied on the third party, the State is directly providing for a non-Medicaid payment to a private pay patient that is positively correlated to the amount of the tax." This reliance is misplaced. In the example, the tax is on "charges" and the non-Medicaid payment is to compensate the private pay patients "for the tax added to their nursing home bills." In other words, the wording of the example implies a one-to-one correspondence between the amount of a tax (levied on and paid by a private pay patient) and the amount of the grant to that patient. That a positive correlation might exist in this situation does not mean that it would automatically exist whenever grants are made to private pay patients. Moreover, contrary to what CMS now argues, the statement does not clearly imply that CMS will always consider such a tax to be levied on the third party. The "if" makes this conclusion contingent on a determination about the nature of the tax. CMS has provided no guidance on how to make this determination, for purposes of applying the positive correlation test. The regulation, however, interprets the statutory term "taxpayer" as meaning the "providers or others paying the tax." 42 C.F.R. � 433.68(f)(1). (10) Thus, only if the third party (here the private pay patient) is "paying" the tax would the third party be the taxpayer. CMS itself describes the private pay patients here as "shouldering" or "bearing the burden of the tax" rather than as "paying" the tax, however, and further says that it is not suggesting that one needs to view these taxes as imposed on the private pay patients. Tr. at 21-24, 52-53. As the States point out, moreover, the State Medicaid Manual issued by CMS had previously addressed taxes that were considered to be imposed on residents, rather than on facilities, and the only example given was a sales tax. State Medicaid Manual, � 2493. (For a discussion of this provision, see Louisiana Dept. of Health and Hospitals, DAB No. 1109 (1989), aff'd Louisiana Dep't of Health and Hospitals v. Sullivan, 760 F. Supp. 929 (D.D.C. 1991).) The States also assert that the wording of their tax statutes makes it clear that the tax is imposed on the providers, not on the patients. CMS provides no analysis of the language of the tax statutes to support a position that the taxes should be considered to be imposed on the private pay patients. Instead, CMS relies on the fact that the providers were generally not precluded from passing the tax on to the private pay patients (and that in one State the providers were required to inform private pay patients of the amount of tax imposed with respect to payments from them to the nursing facility). CMS also argues that the grant or tax credit programs for the private pay patients were established in anticipation that the providers would increase their private pay rates to cover the cost of the tax. In our view, the States correctly determined here that their taxes were imposed on the providers, not on the private pay patients, in light of the regulation interpreting the term "taxpayer" as the "providers or others paying the tax," the wording of their state laws, and the previous guidance in a related context. CMS itself did not rely on the preamble example to suggest that the private pay patients could be viewed as the taxpayers until at least June 2000, well after most of the questioned programs had ended. Finally, as the States point out, "CMS has not acted as if this was a tax on the recipients of service. They've treated it as a tax on the nursing homes." Tr. at 55. Specifically, CMS reduced the total Medicaid expenditures for each State by the total amount of taxes the State received from the providers, not just the part that was related to private pay patients. (11) We note that, in one respect, the example arguably supports the CMS position here. The States argue that there was no payment to the provider taxpayer, even indirectly, by reason of their grant or tax credit programs. According to the States, since the providers could pass through the tax costs to the private pay patients irrespective of the existence of the grants or credits (and in many cases continued to do so after the grant or credit program ended), the flow of money to the provider taxpayer from the private pay patients cannot be considered to be attributable to the grants or credits. The States support this view by pointing out that nothing required the recipients of the grants or credits to use the funds to pay providers. Those States with tax credits also point out that the patients did not receive the benefit of the credits until the year following the year in which the tax was passed through. The example did at least, however, give the States implicit notice that payments to private pay patients might be considered indirect compensation to providers. In sum, while the preamble example indicates that non-Medicaid payments to private pay patients might give rise to a hold harmless situation, CMS cannot reasonably rely on the example as precluding all grant or tax credit programs for the benefit of private pay patients, since that would be inconsistent with the wording of the preamble example, as well as with the plain wording of the statute and regulations. (12)
CMS takes the position that "if there was any doubt as to using the word might in the interim final rule, that doubt was completely removed in the final rule." Tr. at 43-44. The language on which CMS relies for this statement is as follows:
58 Fed. Reg. at 43,166-167. CMS's reliance on the preamble to the final rule is also misplaced. First, CMS's position that this clarifies that all granny grant and tax credit programs render the related tax programs impermissible is inconsistent with the acknowledgment that "grant programs could be structured to avoid hold harmless problems." Second, while the preamble indicates that using non-Medicaid funds to indirectly compensate providers for the cost of the tax imposed on private charges violates section 1903(w)(4)(A) of the Act, reading this statement to preclude such indirect compensation even if it was not positively correlated to one of the specified variables is not reasonable. The statute and regulation clearly require both a non-Medicaid payment and a positive correlation. This requirement is reflected in the following statement from the preamble to the final rule:
58 Fed Reg. at 43,167 (emphasis added). The preamble statement that section 1903(w)(4)(A) is violated by a grant program indirectly compensating providers for the cost of the tax imposed on private charges must be read in such a way that it does not conflict with the plain language of section 1903(w)(4)(A) and its implementing regulation. The preamble statement makes sense if it is read as referring to the situation described in the original preamble example - where part of the tax revenue is returned to the provider through Medicaid rate increases and the "remaining portion" is used for grants to cover the cost of the tax on private charges. The implication is that, in that situation, each provider is getting part of its total tax cost back directly from Medicaid payments and the rest indirectly from the grants, so that the indirect grant payment to the provider equals the difference between the Medicaid payment and the total tax cost to the provider and thus is automatically positively correlated to that difference. The further preamble statement that the non-Medicaid payment "is positively correlated" to the amount of the tax is again contingent on whether the tax "is considered to be" levied on a third party. Again, no clarification is provided about how this determination will be made or by whom. As explained above, there is no factual basis for considering the taxes here to be taxes levied on the private pay patients. In sum, rather than clarifying the circumstances in which a grant program will violate the positive correlation test, the preamble to the final rule does little more than reiterate a concern about grant programs and reflect the view that such grants may be considered a form of indirect payment to provider taxpayers or a form of direct payment to taxpayers if the tax is considered to be imposed on the private pay patients. In context, the preamble statements cannot reasonably be read as a basis for finding a hold harmless situation based solely on the existence of a grant payment or tax credit. Under the statute and regulations, in the absence of a guarantee, the tax is rendered impermissible by a non-Medicaid payment only if that payment is positively correlated (in the statistical sense) either to the total tax cost to the taxpayer or to the difference between the total tax cost and the Medicaid payment. (13) Finally, we note that, while CMS suggests that the preamble example was specifically "directed at" the programs in Tennessee, Illinois, and Louisiana (CMS Brief at 48), each of these States' programs was different from the example, since each imposed a tax on licensed or occupied beds (rather than charges, as in the example). Also, CMS did not ultimately find that any of these States returned part of the funds through increased Medicaid reimbursement and the remainder through grants given to private pay patients to compensate them for taxes added to their nursing home bills, as in the example. Moreover, CMS did not explain why, if CMS had these States' programs in mind when developing the example and addressing the comments on it, none of CMS's December 1994 letters questioning these programs cited either the positive correlation test or the preamble example as a basis for questioning whether a hold harmless provision was in effect.
In questioning the States' programs over the years, CMS failed to refer to the statistical meaning of the term "positively correlated." In its briefs, CMS treats the statistical meaning as irrelevant. At the oral argument, CMS tried for the first time to illustrate what it saw as demonstrating a positive correlation through charts for each State. CMS Demonstrative Exs. 1-5. These charts do not demonstrate that the positive correlation test was met, however. First, none of the charts correlates the alleged indirect non-Medicaid payment to the taxpayer provider (in the form of grant payments or tax credits to private pay patients in the facility) to the total tax cost to the provider. CMS's charts are not graphs showing one of these variables along the x axis and the other variable along the y axis, like the type of graph (referred to as a scatter diagram) usually used to illustrate a positive correlation analysis. Compare CMS Demonstrative Exs. 1-5 with the figures in JAF Exs. 1 and 2. Second, CMS's charts for Hawaii and Maine compare rates of payment of provider taxes and tax credits for private pay residents. CMS Demonstrative Exs. 1 and 2. As the States point out, however, the statute and regulation refer to the "amount" of the payment and the "amount" of the tax (total tax cost). Two rates could be the same, but if they are applied to different base amounts, the ultimate amounts will be different. If the bases do not automatically increase together, the fact that the rate is the same (or proportional) is irrelevant to the issue of whether the two variables are positively correlated. For example, if the tax is a percentage of charges and the facility provides additional services to Medicaid patients (increasing its total charges), the total tax cost would increase, but the amount of the non-Medicaid payment would not increase because the charges to the private pay patients have remained the same. Even the charts that use amounts rather than rates are useless to show a positive correlation between an indirect non-Medicaid payment and the total tax cost to a taxpayer provider. These charts compare the per bed tax amount for a particular period (per month, per year, or per quarter) to the amount of a grant to an individual private pay patient for the period. The per bed tax amount, however, is not the "total tax cost" to the taxpayer provider referred to in the regulation. To determine the total tax cost to any provider in a period, the per bed amount generally has to be multiplied by the total number of beds (either occupied or licensed, depending on the State) in the taxpayer facility in that period. (14) CMS's charts, therefore, do not take into account how the total tax cost for any period would vary according to the total number of beds. Similarly, CMS's chart does not take into account the fact that the amount of any indirect non-Medicaid payment to any taxpayer for any period would depend on the number of private pay patients in the facility eligible for the grants, not just the amount of any individual grant authorized for that period. In sum, the charts compare variables that are not the variables described in the statute and regulations.
While we agree with CMS that a statistical analysis is not always required in order to show a positive correlation between two variables, a statistical analysis is required unless one of the relevant variables would automatically increase as the other did. The States' programs at issue here, however, were structured so that other variables would affect the total tax cost to a provider, or any indirect payment to a provider through grants or credits to private pay patients, or both. For example, the total tax cost to a provider under programs where the tax was imposed per licensed or occupied bed (as in Louisiana, Tennessee, and Illinois) would increase if the number of licensed or occupied beds increased, but if the number of private pay patients in the facility went down, the imputed indirect Medicaid payment to the provider from the grants would go down. Also, in all of the States but Hawaii, not all private pay patients were eligible for the grants or tax credits at issue, so even if the total tax cost to a provider increased because of increased income from (or a bed occupied by) a private pay resident, the non-Medicaid payment would go up only if the private pay resident was eligible for the grant or credit. In Tennessee, since the tax was on licensed beds, the tax could remain the same even if the number of licensed beds occupied by private pay patients eligible for grants increased or decreased. Moreover, Louisiana made the amount of the grant discretionary, so that the amount of any indirect compensation to the provider from passing the tax through to the private pay patients receiving grants varied according to the discretionary grant amount, as well as to the number of private pay patients the provider had during any period. Also, while Louisiana imposed the tax only on an occupied bed day, the amount of any monthly grant was the same, regardless of how many days in the month the private pay patient actually occupied a bed. Given variables such as these, the States cannot reasonably be said to have provided for a non-Medicaid payment to a provider taxpayer that would automatically increase as its total tax cost increased. It may be that in fact the alleged indirect non-Medicaid payment to a provider under these programs (in the form of the grants or credits to private pay patients in the facility) increased as the total tax cost increased. But, determining this would require a factual analysis using valid statistical methods, which CMS did not perform here in spite of having had ample opportunity.
CMS's application of the hold harmless provision in section 1903(w)(4)(C) of the Act is also inconsistent with the regulation and its preambles. The provision at section 1903(w)(4)(C) of the Act is repeated in the lead-in language in the implementing regulation at section 433.68(f)(3), indicating that a hold harmless is in effect if "the State . . . imposing the tax provides, directly or indirectly, for any payment, offset, or waiver that guarantees to hold taxpayers harmless for all or a portion of the tax." In its brief, CMS refers to this language as a "catchall" provision preluding "any indirect payments to providers, that [pay] for any portion of the tax costs" that "effectively dooms each of the tax programs at issue in this case." CMS JBr. at 3 (emphasis in original). As we discuss in this section of our decision, this description of the provision fails to consider how the regulations implemented the requirement and why the regulatory choices were made and also fails to give effect to the plain meaning and preamble interpretation of the statutory term "guarantees."
While the lead-in language in section 433.68(f)(3) simply repeats the statutory language, the interim final rule went on to state that "[i]f an explicit guarantee does not exist, then a two-prong 'guarantee test' will be applied" and to describe the two-prong test. 42 C.F.R. � 433.68(f)(3)(i). The preamble explained: "[I]f an explicit guarantee exists, the tax would be impermissible and the two-prong test will not apply. If an explicit guarantee does not exist, the two-prong test will apply." 57 Fed. Reg. at 55,129. The preamble also said: "If an explicit guarantee does not exist and if the tax is applied at a rate that is in excess of 6 percent of the revenue received by the taxpayer, we will apply the second prong of the test to determine if an inexplicit guarantee exists in violation of the hold harmless provision." Id. CMS made no finding applying the two-prong indirect guarantee test in the regulations to any of the States. Yet, as we discuss below, some of the audits on which CMS relied referred to "indirect guarantees" in the States' laws. Each of the disallowance determinations referred to the State "indirectly" holding taxpayers harmless, and none relied on finding an explicit guarantee. The States argue that, since CMS did not find any explicit guarantee in their State laws, the two-prong test applies and, therefore, no hold harmless was in effect. CMS argues that the lack of an explicit guarantee is irrelevant, pointing out that the reference to an "explicit" guarantee was dropped in the final rule. CMS does not explain how it can reasonably rely on this change to support the disallowances for periods prior to the change. Moreover, the history of this change shows it was not intended to have the effect CMS now advocates regarding when the two-prong test applies. First, no change was made to the lead-in language. Rather, the key change was to substitute the statement that "[a]n indirect guarantee will be determined to exist under a two prong 'guarantee' test" for the statement that "[i]f an explicit guarantee does not exist, then a two-prong 'guarantee' test will be applied." The preamble to the final rule explained this change as follows:
58 Fed. Reg. at 43,167. In other words, the change was described as merely a clarification, not a substantive change. At most, one could imply from this clarification that the relevant distinction is between a direct guarantee and an indirect guarantee, rather than an explicit guarantee and an inexplicit guarantee. Any implication that the two were not equated, however, is undercut by the fact that the preamble went on to say:
In other words, the two-prong test was still viewed as ensuring that the hold harmless provision was not violated in a situation where there was no explicit guarantee of repayment. This is not substantially different from the wording of the interim final rule. The preamble language clearly suggests that if a State did not explicitly guarantee a payment, offset, or credit to a taxpayer and if it timely met the two-prong test, its tax would be considered permissible. Indeed, CMS's November 16, 1993 letter to State Medicaid Directors describing the final rule stated: "The regulation at 433.68(f)(3)(i) specifies a two-prong test if an explicit guarantee does not exist." JAF Ex. 54, at 10; see, also, JAF Ex. 70, at 1-2 (2002 e-mail). In light of these statements, we find unreasonable CMS's reliance on the general statement in the preamble to the interim final rule that the guarantee provision was being interpreted "to mean that use of any State payment . . . in a way that is guaranteed to repay the taxpayer for all or part of the cost of health care-related taxes is a hold harmless situation." CMS JBr. at 15, citing 57 Fed. Reg. at 55,129. CMS concludes from this statement that "if the state created a payment scheme, in the form of a grant or credit, which would make monies available to pay for any portion of the provider tax, CMS would consider payment to be guaranteed and a hold harmless situation to exist." CMS JBr. at 15. Reading this statement as authorizing CMS to examine the use of a payment without regard to the two-prong test where there is no explicit guarantee is unreasonable. First, this reading is inconsistent with the language of the interim final rule itself, which specified that the two-prong test would apply if there was no explicit guarantee. Second, the preamble statement went on to say that the "third statutory criterion would also consider as a hold harmless any sort of explicit guarantee . . . that assures repayment of tax costs." 57 Fed. Reg. at 55,129 (emphasis added). In other words, the statement about "use of any State payment" was referring to a situation in which there is no explicit guarantee - the situation in which the two-prong test applies. CMS also erroneously suggests, based on preamble language, that the two-prong test is implicated only when a State imposes a "disproportionate" tax, that is, a tax in excess of six percent of revenues. See, e.g., CMS IL Br. at 19, n. 8. The six percent test, however, is the first prong of the two-prong test and thus is part of the intended evaluation of whether a State is in fact using a payment in a way that guarantees to hold a taxpayer harmless, even though no explicit guarantee is given. CMS further argues that the States here each provided for an indirect payment that was proscribed by the general lead-in provision in section 433.68(f)(3). (15) This lead-in language essentially repeats the statutory language and at first blush seems to apply to any payment that guarantees to return to a provider even a small part of the tax. The regulatory history as a whole, however, makes it clear that the regulatory choice to distinguish explicit and indirect guarantees and to adopt the two-prong test was made in order to reconcile this broad language with the statement in section 1903(w)(4) of the Act that the hold harmless provisions "shall not prevent use of the tax to reimburse health care providers in a class for expenditures under this title" and in order to permit the states some flexibility in designing their tax and payment programs. 57 Fed. Reg. at 55,129. In sum, finding indirect guarantees without applying the two-prong test is inconsistent with the regulations. CMS's findings (even if accurate) were not legally sufficient as a basis for concluding that the taxes were impermissible under the indirect guarantee provision.
Although none of the audit reports issued to the States found any explicit or direct guarantee in the States' laws, CMS now says that the State law provisions should be considered to be direct guarantees, even though there is no explicit promise or assurance of any payment to the provider. CMS says that all it needs to do to show a direct guarantee is to see if, on the face of the State statute, there was a direct or indirect payment made, and whether the effect of that payment was to make State money available to reimburse nursing homes for a portion of the tax costs. In support, CMS points to the example of an explicit guarantee in the preamble - a State law that provides for a property tax credit for providers. The States do not dispute that a provision of State law other than the law enacting the tax program could constitute a direct or explicit guarantee if it establishes a legal entitlement on the part of a provider to a payment. The States argue, however, that their programs are distinguishable from the property tax credit program in the preamble example. They point to aspects of their grant or tax credit programs which they say mean that the providers had no assurance of receiving any payment as a result of the grant or tax credit programs because of the way they were structured. The States also argue that the history and context of the guarantee provision show it was intended to address the situation where taxpayers were assured of repayment by operation of law. Neither the interim final or the final rule defined the term "guarantee" or provided any guidance (other than the property tax credit example) for determining if an explicit (or direct) guarantee existed. The States cited the following dictionary definition of the term "guarantee" in their initial brief:
Webster's II New College Dictionary (1995), definition of "guarantee" cited in JBr. at 25. Applying this definition is consistent with the preamble to the interim final rule, which said that the "third statutory criterion would also consider as a hold harmless any sort of explicit guarantee, for example, in a State law authorizing a health care-related tax, that assures repayment of tax costs." 57 Fed. Reg. at 55,129 (emphasis added). Nothing in the preambles supports the CMS view that merely making indirect payments "available" through a private pay patient constitutes a guarantee to hold a provider harmless. Indeed, the term "hold harmless" is usually used in conjunction with some sort of indemnification that is legally enforceable. If making a payment available to some taxpayers were sufficient by itself to constitute a hold harmless guarantee, there would be no need for the two-prong test. Contrary to what CMS now argues, moreover, there is no direct assurance of any payment to a provider taxpayer under the State programs at issue here. As the States point out, even assuming that the grant payment or tax credit could be considered an indirect payment to the provider (which the States do not concede), the provider had no assurance that such a payment would be made to it since even imputed receipt of such a payment was contingent on factors beyond the provider's control, such as the provider having private pay patients who qualified for the grant or credit and those patients actually applying for and obtaining the grant or credit. (16) Contrary to what CMS argues, the mere "possibility" that some funds might ultimately flow through to the provider taxpayer is not tantamount to a direct guarantee of payment. See Tr. at 127-28. CMS also says that there was a direct guarantee here because the grants or tax credits were "targeted" to private pay patients who would bear the burden of the tax. Tr. at 56-57. This is simply not enough, by itself, to constitute any direct or explicit assurance to any provider that it would in fact receive any payment at all, however. We also note that, although the States pointed out in response to the draft audit reports that the drafts neither applied the two-prong indirect guarantee test nor found an explicit or direct guarantee, three of the final audit reports continued to refer to "indirect guarantees." Moreover, all of the disallowance letters referred to the providers being "indirectly" held harmless. One has to question how, if over the period of at least seven years during which it was aware of these States' programs, CMS never alleged that there was a direct or explicit guarantee to the taxpayer provider in these States' laws, CMS can reasonably argue now that such a guarantee existed, based on the same information it has always had.
CMS's current position on the guarantee test, as summed up during the oral argument, is as follows:
Tr. at 23-24; see also Tr. at 36. In other words, CMS is now saying that the problem is that the States provided a guarantee to the private pay patients. The plain language of the statute and regulations, however, addresses a payment that guarantees "to hold taxpayers harmless for all or a portion of the tax." We have discussed above why we conclude that the taxpayers here are the providers, not the private pay patients. As noted above, CMS itself mostly characterizes the private pay patients as the individuals that were shouldering (or bearing) the burden of the tax, rather than as the taxpayers, and did not base the disallowances here on any finding that the private pay patients were the ones "paying the tax." (17) Finally, even if the State laws here could be viewed as providing for an indirect guarantee to the provider taxpayers of repayment of part of their taxes by means of a direct guarantee to private pay patients, the States reasonably thought that timely meeting the two-prong test would protect them from any reduction in their Medicaid funding, given the plain wording of the regulation.
Under the regulation, a provider tax is considered permissible under the indirect guarantee provision so long as it timely meets the 6 percent test, even if the tax is imposed only on providers primarily providing Medicaid services (such as ICF/MRs) and the state returns more than 75 percent of the payments to more than 75 percent of the providers. This at least raises a question about whether there is a legitimate cause for concern about any tax that meets the 6 percent test, and is applied more broadly to nursing facilities (including facilities with fewer Medicaid patients and in some instances excluding ICF/MRs), merely because some private pay patients are receiving grants or credits that might enable the provider to pass the tax through more easily. The correspondence between CMS and the States over their programs indicates that one concern CMS had was that the States were both returning part of the tax revenues through increased Medicaid rates and paying the remainder through the grants or credits, as in the preamble example of a program that might violate the positive correlation test. Before us, however, CMS has been silent on the effect of the combined Medicaid increases and non-Medicaid payments, and some of the States presented evidence that their Medicaid rates did not in fact go up substantially over the years or that any increases were to cover other nursing facility costs, such as a mandatory increase in pay for certified nurse assistants. CMS suggests that there is a reason to be concerned about these grants or tax credits, even if the 6 percent test was met, because, if the private pay patients would not feel the burden of the tax, they would be less likely to oppose the tax program and Congress intended to put brakes on such health care-related taxes. (18) While some CMS officials and others have in the past opposed any health care-related tax, Congress decided to permit such taxes so long as they met certain restrictions. As the States argue, those restrictions were carefully negotiated and are much more narrow than the restrictions that CMS had originally promulgated (and that Congress rejected) that would have proscribed any "linkage" between a tax program and any payment to a taxpayer. See 56 Fed. Reg. 46,380; 46,387 (Sept. 12, 1991); 56 Fed. Reg. 56,132, 56,139 (Oct. 31, 1991). CMS cannot now reasonably fail to apply the restrictions Congress enacted, as interpreted in the regulations and preambles, simply because grant or credit programs might counteract opposition to the tax programs. As late as 2000, CMS was also relying for its position that the States' tax programs were impermissible on its assertion that Medicaid and Medicare were bearing the "full burden" of the tax costs under the States' programs. When the States provided information that the providers (or private pay residents or their insurers when the tax was passed on) were in fact bearing a substantial part of the tax costs related to private pay patients and that Medicaid reimbursement rates were subject to limits that affected whether the taxes were in fact being reimbursed by Medicaid, CMS then decided that this factor was irrelevant. We agree that a hold harmless could be found even if Medicaid and/or Medicare do not bear the full burden of the tax, but note the following points. First, in questioning these States' programs over the years, CMS was focusing on what it thought was the effect of the programs rather than on the regulatory language and the facts about how the programs were structured. Second, CMS's initial view of the effect did not withstand scrutiny. CMS here characterizes what the States did as "scams" or "schemes" intended to increase federal match with no real cost to the taxpayer or state, but does not adequately explain how the States' grant or credit programs have that result. CMS also argues that the States knowingly flouted the law. Yet, the States submitted evidence that they in good faith were attempting to comply with the 1991 law when they enacted their programs. That evidence has not been effectively rebutted by CMS. CMS presented some evidence that some persons in the States raised questions about whether the programs met the terms of the 1991 law, but we do not find that evidence persuasive. Those raising the questions either were not those responsible for evaluating the programs' legality (and in some instances were providers who opposed the programs) or were basing their questions on their understanding of the example referring to grants to private pay patients in the draft preamble to the interim final rule, before the word "might" was added. While good faith alone is certainly not enough to show compliance with the law, it undercuts CMS's position that the States knew they were violating the law, but nonetheless proceeded. Nor does CMS's evidence support its assertion that the States intended to "supplant" State Medicaid dollars with federal dollars. The State legislative histories on which CMS relies for this argument show merely that some States intended to use some or all of their increased revenues from the taxes to increase Medicaid reimbursement rates to the nursing facilities, as is specifically permitted by the 1991 law, so long as the terms of the law are met. References to increased federal funding in connection with the taxes does not necessarily suggest anything nefarious. Under the Medicaid law, the States are generally entitled to federal funding in the higher reimbursement rates. The real concern about artificially inflating the federal share arose from some States setting rates (generally for public facilities) in excess of the costs of providing Medicaid services. The histories also show that the States here were struggling to cover Medicaid expenditures caused by what they viewed as unfunded federal mandates, that their Medicaid reimbursement rates were low and were being challenged in court, and that facilities had been shifting Medicaid costs to private pay residents. In this context, it is not surprising that the States were conscious of what effect the increased taxes might have on private pay residents if they were passed through, and acted to mitigate the burden on some residents. In any event, as noted above, the Secretary opted in the regulations to reject an approach based on subjective factors such as legislative intent, in favor of objective tests. CMS may regret this choice, but is bound by it in the absence of a regulatory amendment, as are we. See 45 C.F.R. � 16.14. Finally, CMS relies on the fact that at one time some States had indicated a willingness to settle the dispute with CMS over their programs (and had drafted legislation to accomplish this) as evidence that the States knew their programs were bad. We decline to draw this inference. In fact, the record indicates that CMS also was willing to settle these disputes, but we do not infer from that willingness that CMS knew its position was unwarranted. Instead, our decision is based on the language of the statute, the regulations, and the preambles.
In this section, we set out a more detailed analysis of each of the States' programs, explaining why we conclude that some of CMS's State-specific factual findings or assumptions are unsupported and how our general analysis above applies to the specific facts as we find them. For each State, we describe the tax and grant or tax credit programs and the history of correspondence on the hold harmless issues between the State and CMS, before addressing issues about application of the positive correlation and guarantee tests specific to the State. In describing the correspondence, we focus primarily on issues specific to the particular State. We note at the outset, however, that the key letters to the States, even when issued by a regional official, contained boilerplate language that was mostly very general, concluding that providers were indirectly held harmless with little or no analysis of how a particular State's programs were structured. Some of the boilerplate language was imprecise in describing the statutory hold harmless provisions. For example, the boilerplate in the letters issued on September 1, 1999 describes the 1991 law as providing that State taxes must "Avoid 'hold harmless' Arrangements - The taxes collected must not be returned to the taxpayers directly or indirectly." See, e.g., IL Ex. 318, at 2. This statement is overbroad, however. The 1991 law explicitly permitted States to increase Medicaid reimbursement to provider taxpayers and implicitly permitted some non-Medicaid payments, so long as the State did not have in effect a hold harmless provision under one of the three carefully worded statutory tests. Several other things are striking when one reviews the CMS letters. First, they contain no references to the interpretations in the regulations and preambles, except the preamble example referring to grant programs. Even that example was not relied on by CMS until the draft audit reports issued in June 2000, moreover. Those drafts indicate that CMS thought at the time that the grant or tax credit payments represented the difference between Medicaid payment increases and the tax cost to the provider - a position CMS has since abandoned. Second, the parts of the letters that address the specific State's program mainly cite subjective factors such as CMS's view of the "intent" of the State provisions, despite the specific rejection of such subjective factors in the rulemaking proceedings. Finally, as discussed above, some of the letters also took the position that, as a result of the States' programs, the Medicaid and/or Medicare programs bore the full burden of the taxes - a conclusion that CMS later determined was irrelevant. Having made these general observations, we now turn to the individual States.
Effective July 1, 1993, Hawaii imposed a 6 percent tax on nursing facility income. Income was defined to include the total compensation received for furnishing nursing facility services, including all receipts from ancillary services and receipts from items supplied in connection with these services, but to exclude compensation for Medicare-covered services and some other types of revenue. Nursing facilities were not precluded from passing on the amount of the tax to patients, but could not change the amount contributed by a Medicaid recipient for his or her care. Each operator of a nursing facility was required to "identify separately the tax imposed by this section in all invoices or statements to persons whose payments result in nursing facility income." Hawaii Rev. Stat. � 346E-2; HA Ex. 203 and 204. Some facilities could not legally pass the tax amount through to private pay residents, because they had pre-existing financing arrangements. Odo Decl. at � 4. Also effective July 1, 1993, Hawaii amended an existing 4 percent medical services excise tax credit for individual taxpayers by enacting provisions permitting an income tax credit equal to 6 percent of nursing facility expenses for services provided to a resident individual taxpayer (or to the taxpayer's dependent) during the taxable year for which the income tax return was filed. Nursing facility expenses included any amounts that constituted "income" to the facility for purpose of the nursing facility tax, regardless of whether the expenses were covered by insurance. Hawaii Rev. Stat. � 235-55.9; HA Ex. 202. The income tax credit was available to a resident (or to a person claiming the resident as a dependent) regardless of whether the facility in fact passed on the provider taxes imposed on it. Id. The credit was available only if claimed within one year after the end of the tax year in which the expenses were incurred. Id. Nursing facility residents were not required to use the credit to pay for their care.
At the time it enacted its programs, Hawaii was attempting to conform to federal Medicaid law, and had reduced the original amount proposed for the tax from 10 percent to 6 percent of income to meet the hold harmless indirect guarantee test (which it referred to as a "safe harbor"). HA Ex. 201; Odo Decl. at �� 8, 10. In a December 19, 1994 letter, the CMS Acting Regional Administrator informed Hawaii that the "income tax credit associated with the nursing facility tax violates section 1903(w)(4)(A) of the Social Security Act" (the positive correlation test). HA Ex. 212, at 1. (19) This letter contained no analysis of the language of the positive correlation test or of the factual basis for determining that it applied to Hawaii. No mention was made of the guarantee test. Hawaii was provided an opportunity to submit information "before [CMS] takes a disallowance." Id. Hawaii responded in a letter dated January 18, 1995, setting out its analysis of the hold harmless provisions and why it thought they did not apply. HA Ex. 213. The Acting Regional Administrator responded in a brief letter dated July 21, 1995, stating:
HA Ex. 214. No supporting rationale was provided, nor did the letter address Hawaii's arguments about why there was no hold harmless. No disallowance was issued at that time. Hawaii terminated both its nursing facility tax program and its income tax credit effective June 30, 1997. The next correspondence on this issue from CMS to Hawaii (other than the State Medicaid Directors' letters referred to in the Legal Background section above) was a letter from the Director of the Center for Medicaid and State Operations, dated September 1, 1999 (over four years after the July 1995 correspondence and over two years after Hawaii had ended its programs). HA Ex. 217. The letter explained that State questions about HHS policy and regulations had led to a thorough review of the policies and to some legislative proposals (which were not enacted). It further explained that, based on the information CMS had to date, it was CMS's belief that "the tax credit program associated with [Hawaii's] nursing facility tax program violated" both the positive correlation and the guarantee tests. (20) The letter informed Hawaii that CMS would contact it to gather information to "confirm the possible permissibility of the tax in question" and, if the "provider tax" were found to be impermissible, would do an audit to determine federal matching funds paid as a result of the impermissible tax. A letter from the Director of the Center for Medicaid and State Operations, dated June 15, 2000 notified Hawaii that CMS had "preliminarily concluded" that the tax was impermissible and provided Hawaii an opportunity to comment on a draft audit report. HA Ex. 221. The draft audit report contained almost verbatim the brief rationale previously given for finding the program impermissible. Id. Hawaii responded in a letter dated July 14, 2000, arguing that the draft report mixed the two hold harmless provisions, and failed to acknowledge the "specific limitations of each that are applicable to the Hawaii case." HA Ex. 221, at 2. The letter then went on to analyze the two provisions, stating why Hawaii did not think they applied. Among other things, Hawaii pointed out that the auditors had not cited any direct guarantee and had made no contention that the Hawaii tax violated the indirect guarantee aspect of the regulations. Id. at 4. The final audit report attached to the CMS response, dated December 19, 2000, for the first time gave a more detailed rationale in response to Hawaii's argument. Among other things, the final audit report stated that-
HA Ex. 222, Att. at 13. The report went on to mention the simultaneous enactment of the nursing facility tax and the amendment to the tax credit provision, the requirement that facilities separately identify the tax on invoices or statements, and the fact that the credit could be claimed only for expenses subject to the tax on nursing facilities. Id. The report also cited the same factors as "evidence that the nursing facility expense income tax credit was intended to express an indirect guarantee for taxpayers to be held harmless for a portion of the nursing facility tax cost." Id. at 14 (emphasis added). The report did not find that the two-prong indirect guarantee test in the regulations was violated, however. The formal disallowance letter, dated January 19, 2001, stated that CMS was disallowing $17,750,950 in federal financial participation (FFP) claimed for federal fiscal years (FFYs) 1993 through 1997. The letter summarized the final audit report as determining that the "nursing facilities were indirectly held harmless for some or all of the tax payment." HA Ex. 223, at 1.
As indicated, the factors found in the audit report on which CMS relied for its determination were primarily factors from which CMS inferred that the intent of the programs was to increase federal Medicaid funding or based on which it determined that the tax credit program was "associated with" or had a "link" to the tax program. The report did not state how these factors show a positive correlation in the statistical sense, nor did it provide any statistical analysis of the relationship between alleged "indirect payments" to a nursing facility from tax credits to its residents and the total tax cost to the provider. The only numerical reference in the report was to the fact that the tax rate (as amended) was 6 percent of the facility's income and that the tax credit was 6 percent of the private pay resident's expenses that constituted income for purposes of the tax. As indicated above, however, the fact that the rates are the same is irrelevant. The statute and regulations look at the "amount of the tax," and interpret that as the "total tax cost" to the "provider or others paying the tax." A rate is not the same as the amount. Nor does the fact that the rate is the same mean that the non-Medicaid payment will automatically increase as the total tax cost does. The rates have to be applied to different bases in order to determine these different amounts for each taxpayer. Assuming that one would consider the amount of the tax credits available to private pay patients in a facility to be an indirect non-Medicaid payment to the facility, the amount of the indirect payment would be 6 percent of the private pay patients' expenses. (21) This amount would not automatically increase, however, every time the provider's total tax cost increased. If the tax increase was related solely to increased Medicaid payments for expenses incurred by a Medicaid patient for whom no tax credit could be claimed, the amount of the non-Medicaid payment would stay the same (or could even decrease if the Medicaid patient replaced a private pay patient in the facility). CMS's attempt at the oral argument to demonstrate a positive correlation in the statistical sense is likewise flawed, since it simply compares the rates of the nursing facility tax and of the credits to the private pay patients. CMS Demonstrative Ex. 1. Thus, we conclude that the record here does not provide a basis for concluding that Hawaii's programs violated the positive correlation test.
We first note that the finding in the final audit report was that certain aspects of Hawaii's programs evidenced an "indirect guarantee" of payment to the provider. Finding an indirect guarantee without applying the two-prong test conflicts with the regulation, however. CMS now says that the indirect guarantee test is irrelevant and that there is a direct guarantee, but points to nothing in the State law provisions that either explicitly or directly assures any payment to the taxpayer provider. We find no such assurance in those provisions. Nothing required the private pay patients to apply for the tax credit, nor was a provider even assured by the law of having some private pay patients who could claim the tax credit. The law did not preclude the provider from passing the health care-related tax on (except to Medicaid patients who paid part of their costs), but neither did it assure the provider that it could pass the tax through, and some providers had pre-existing agreements that precluded any increase in payment rates. (22) While Hawaii's law did assure the private pay patients that they could claim the tax credit, if they timely filed for it, nothing in the law makes the private pay patient the one "paying" the health care-related tax. (23) Thus, we conclude that there is no basis in this record for determining that Hawaii's programs violated the guarantee test. Accordingly,
we reverse the disallowance for Hawaii since there is no basis in the
record for determining that there was a hold harmless provision in effect
in Hawaii during the disallowance period, under either the positive correlation
test or the guarantee test.
Effective July 1, 1993, Maine imposed a tax on "all persons engaged in the business of . . . nursing home operation in the State" at a rate of 7 percent of the "gross receipts of the business from the provision of nursing home patient care including all charges made by the business upon or with respect to patients receiving nursing home care." 36 Maine Rev. Stat. Ann. �� 2821, 2822; ME Exs. 501, 502. (24) Nursing homes were not precluded from passing the cost of the tax on to private pay patients, but the State statute provided that the tax was imposed "upon the person selling the item or service rather than the consumer." 36 Maine Rev. Stat. Ann. � 2823; ME Ex. 502. Nursing homes were required to provide an annual statement to each person paying for nursing home care during the previous calendar year, showing the total gross receipts tax arising from payments by the payor. 36 Maine Rev. Stat. Ann. � 2825; ME Ex. 502. Also effective July 1, 1993, Maine enacted a refundable tax credit, payable to nursing home residents (or a resident trust), equal to 80 percent of the gross receipts tax on the nursing home arising from the payments made by the resident or resident trust for nursing home care in the preceding calendar year. 36 Maine Rev. Stat. Ann. � 5219-I; ME Ex. 503. The credit had to be claimed in the tax return filed for the preceding year. Id.
Before enacting these programs, Maine reviewed the statute and regulations to assess whether the tax would be permissible and whether the tax credit would violate the hold harmless provision and believed it complied with the requirements. MacLean Decl. at � 4; ME Ex. 509. In December 1993, CMS raised concerns about Maine's programs. Maine reevaluated the income tax credit, repealing the initial tax credit effective January 1, 1994 (although Maine nonetheless believed its program was permissible), and replacing it with a 2.7% catastrophic medical expenses credit. MacLean Decl. at � 8. (25) On April 6, 1994, a letter on behalf of the CMS Administrator informed Maine that its tax program was being reviewed, but that no formal decision had been made regarding its permissibility. ME Ex. 517; see ME Ex. 519. Meanwhile, Maine developed data that it considered adequate to assure CMS that the 75/75 percent test (the second prong of the indirect guarantee test) was met, and provided that data to CMS on May 25, 1994. ME Exs. 518; 522. CMS's Associate Regional Administrator accepted that data as showing permissibility of the program effective January 1, 1994, but asked for guidance from the CMS central office regarding the period between July 1, 1993 and December 31, 1993. ME Ex. 523. On December 19, 1994, the CMS Regional Administrator sent Maine a letter similar to the one sent to Hawaii on the same date, referring to the positive correlation test and seeking more information from Maine. ME Ex. 524. By letter dated January 30, 1995, Maine responded, providing its analysis of why the language of the positive correlation test did not apply to its programs. ME Ex. 526. The CMS response of July 19, 1995, like the response to Hawaii, stated without further analysis that the information provided "does not adequately support the permissibility of the nursing facility tax program." ME Ex. 527. Maine repealed the gross receipts tax effective January 1, 1997. Like Hawaii, Maine next received specific correspondence on its tax program by letter dated September 1, 1999 from the Director, Center for Medicaid and State Operations, citing for the first time both the guarantee test and the positive correlation test. ME Ex. 529. A draft audit report was issued to Maine by letter dated June 15, 2000. ME Ex. 531. This report discussed Maine's earlier response. Among other things, the report stated that the nursing homes were "indirectly held harmless" by the tax credit and that the existence of a positive correlation was evidenced by the ability of the nursing facility to pass on the tax and by the fact that the tax credit was "based on a percentage (i.e., 80%) of the nursing facility gross receipts tax absorbed by the private pay residents from increases in their payment rates for the nursing facility services they received in the preceding calendar year." Id. at 4. Maine responded on July 14, 2000, analyzing both hold harmless provisions and pointing out that the draft report neither contended that the Maine tax violated the indirect guarantee test nor cited any direct guarantee. ME Ex. 532. The final audit report was transmitted to Maine by letter dated December 19, 2000. ME Ex. 533. The final audit report contained a more complete analysis of Maine's arguments, concluding among other things that "the simultaneous enactment of the nursing facility tax and the nursing home resident tax credit, along with explicit statutory language regarding the calculation of the nursing home resident tax credit is strong evidence that the nursing facility tax credit was intended to express an indirect guarantee for taxpayers to be held harmless for a portion of the nursing facility tax cost." Id. at 12 (emphasis added). The report did not state how the relationship between the tax credits and taxes constituted a positive correlation in the statistical sense, nor did it provide any statistical analysis of the relationship. Nor did the report find that the two-prong indirect guarantee test was violated. By letter dated January 19, 2001, the CMS Regional Administrator notified Maine that CMS was disallowing $7,687,661 in FFP for FFYs 1993 and 1994. The disallowance was based on a determination that Maine had an impermissible tax program during the period July 1, 1993 through December 31, 1993, "because the nursing facilities were indirectly held harmless for some or all of the tax payment." ME Ex. 534.
At first blush, there seems to be a positive correlation between Maine's gross receipts tax and its income tax credit, given that the tax credit was calculated by applying 80 percent to an amount defined in terms of the gross receipts tax. Upon further analysis, however, it is clear that the 80 percent is not applied to the total tax cost to the provider. Thus, the application of the 80 percent does not mean that the imputed indirect non-Medicaid payment to the provider taxpayer will automatically increase as the total tax cost increases. Instead, the total gross receipts subject to the nursing home tax (and therefore the total tax cost to the nursing home) could increase, without any corresponding increase in the imputed non-Medicaid payment from the tax credits to private pay residents of a nursing home, if the increased receipts were from payments made by the third party payors such as Medicaid or Medicare, rather than from payments made by a resident (or resident trust) for which the credit could be claimed. Given the lack of an automatic increase in one relevant variable as the other relevant variable increased, there is no legally sufficient basis in this record for concluding that the positive correlation test was violated. CMS did not base the disallowance on any statistical analysis, nor does the chart which it presented at the oral argument show a positive correlation under the regulatory requirements. The chart merely reflects that the credit for an individual resident was at 80 percent of payments made by that resident resulting in a gross receipts tax to the nursing home. CMS Demonstrative Ex. 2. It does not use as one variable the total tax cost to the nursing home from all charges subject to the gross receipts tax, nor demonstrate that the total tax cost in fact increased as income tax credits to private pay residents increased. The subjective factors on which CMS relies to show a relationship or association between the gross receipts tax and the income tax credit simply are not sufficient under the regulations to show a positive correlation between the relevant variables. Thus, we conclude that the record here does not provide a basis for concluding that Maine's programs violated the positive correlation test.
Like the final audit report for Hawaii, the final audit report for Maine found that certain aspects of Maine's programs evidenced an "indirect guarantee" of payment to the provider. CMS made no finding that the two-prong indirect guarantee test was not met. Finding an indirect guarantee without applying the two-prong test conflicts with the regulation, however. Moreover, while Maine's tax was at a 7 percent rate, Maine provided evidence that 75/75 percent test was met. CMS now says that the indirect guarantee test is irrelevant and that there is a direct guarantee, but points to nothing in the state law provisions that either explicitly or directly assures any payment to the taxpayer provider. We find no such assurance in those provisions. Nothing required the private pay patients to apply for the tax credit, nor was a provider even assured by the law of having some patients who could claim the tax credit and would in fact timely do so. (26) Moreover, the Maine statutes specifically provided that the gross receipts tax was imposed on the nursing homes and not on the consumers. The final audit report treated the gross receipts tax as a tax on the nursing homes. The mere fact that part of the tax cost might be passed on to private pay residents is not a sufficient basis for determining that the private pay residents were the ones "paying the tax" instead of the nursing homes. Thus, we conclude that there is no basis in the record for finding that the guarantee test was violated by Maine's programs. Accordingly, we reverse the disallowance for Maine since there is no basis in the record for determining that there was a hold harmless provision in effect in Maine during the disallowance period, under either the positive correlation test or the guarantee test.
Effective
July 1, 1992, Louisiana statutes authorized fees for certain health care
services, imposing the fee on nursing facilities and on some other Medicaid
providers such as pharmacies. The fee on nursing facilities could not
exceed $10 per occupied bed per day. Louisiana Rev. Stat. �� 46:2605,
46:2625;
LA Exs. 403, 404. The actual amount of the fee has varied from year to
year, as follows:
Castile Decl. at � 9. The statute did not preclude nursing facilities from increasing their charges to private pay patients because of the fee. (27) Effective July 1, 1992, Louisiana also enacted a long-term care assistance program administered by the office of elderly affairs. The assistance program was "subject to the availability, appropriation, and allocation of funds for the program" and could also include private grants, federal funds, and any other revenue source for the operation of the program." Louisiana Rev. Stat. � 46:2802; LA Ex. 405. Participation in the program was limited to Louisiana residents who received long term care; did not qualify for any other federal or state health care assistance program providing payment of long term care expenses; had an annual net income of less than $60,000; and were not fully reimbursed for long-term care services by health care insurance. Id. Although the assistance amount originally was $350 per month, the relevant provision was amended in 1993 to provide an amount "up to" $350 per month and to indicate that the amount would be "established by the commissioner of administration" with oversight by Louisiana legislative committees. Louisiana Rev. Stat. � 46:2802; LA Ex. 409. The amount of the assistance provided varied during the disallowance period as follows:
LA Br. at 5, citing LA Exs. 412-415. An individual eligible for the assistance grant for any part of a month received the entire cash payment for that month. LA Ex. 419, at 2. Funding for the assistance program ended as of April 31, 2000, and the enabling statute was repealed in 2001.
Before enacting its programs, Louisiana reviewed the 1991 law and concluded that its fee was permissible and that a grant to private pay nursing home residents would not violate the hold harmless provisions of the law. Castille Decl. at � 6; see LA Ex. 407. Correspondence between the CMS Regional Administrator and Louisiana prior to the effective date of the indirect guarantee tests indicates that questions were raised about Louisiana's programs (including whether the fees were broad-based and uniformly applied and whether there was a hold harmless situation because of increases in reimbursement rates and the grant program), and that Louisiana had sought clarification of how the 75/75 percent indirect guarantee test would be applied. LA Exs. 406, 407, 409, 410. Effective July 1, 1993, Louisiana reduced the amount of the fee from $10 per occupied bed day to $3.68 per bed day "in order to bring the tax under 6% and thus within the 6% safe harbor found in the indirect guarantee test in the interim final regulations." Castille Decl. at � 10. On December 19, 1994, the Regional Administrator sent Louisiana a letter seeking more information regarding its grant program. This letter was similar to the ones sent to Hawaii and Maine on the same date, except that it said that Louisiana's grant program violated the guarantee test, rather than the positive correlation test. LA Ex. 418. Louisiana responded by letter dated January 18, 1995, providing Louisiana's analysis of why its grants did not give rise to a hold harmless situation. LA Ex. 419. CMS's July 14, 1995 response stated without further analysis that the information provided did "not adequately support the permissibility of the nursing facility tax program." LA Ex. 420. CMS obtained information from Louisiana about the amount of nursing facility taxes collected, but took no further formal action until September 1, 1999 (over four years later), when the Director of CMS's Center for Medicaid and State Operations sent a letter to Louisiana similar to that sent to Hawaii and Maine, for the first time citing both the positive correlation and guarantee tests. LA Ex. 426. On June 15, 2000, CMS provided Louisiana an opportunity to comment on a draft audit report. LA Ex. 432. By letter dated July 14, 2000, Louisiana responded, arguing among other things that the draft audit report's analysis of the positive correlation test was based on a faulty premise (that the grant rate increased as the tax rate increased), and that the draft report neither applied the indirect guarantee test from the regulations nor cited any direct guarantee. LA Ex. 433. The final audit report cited to the preamble example of a grant program to reject Louisiana's position that the grant was not a payment to a taxpayer. The report based its conclusion that the positive correlation test was met on findings that the two programs were enacted simultaneously, that the "amount of the grant repayment ranges from 88% to 117% of the monthly fee" (and therefore the "amount of the tax passed on to private pay residents and the amount of grant payments made are all closely related"); that 81% of private pay residents are eligible for the grants; and that "each time the fee has increased/decreased, the grant has simultaneously increased/decreased." CMS Ex. 434, at 8-9. The final audit report did not discuss Louisiana's point about the guarantee test, but states that "through the grant assistance program, the State of Louisiana makes non-Medicaid payments that indirectly hold harmless the taxpaying nursing facilities for a portion of the tax cost." Id. at 7 (emphasis added). The report did not find that the two-prong indirect guarantee test in the regulations was violated or that there was an explicit or direct guarantee. Nor did the report state how the relationship between the grants and taxes constituted a positive correlation in the statistical sense or provide any statistical analysis of the relationship. By letter dated January 19, 2001, the Regional Administrator informed Louisiana that CMS was disallowing $293,919,322 in FFP for the period October 1, 1992 through September 30, 1999. The reason given was that as "expressed in the final [audit] report, the Louisiana nursing facility tax, in conjunction with the Louisiana grant program for private pay patients of such facilities, is impermissible . . . because the nursing facilities are indirectly held harmless for some or all of the tax payment." LA Ex. 436, at 1. By letter dated August 29, 2001, the Regional Administrator informed Louisiana of an additional disallowance of $20,239,931 in FFP, on the same basis, for the period October 1, 1999 through April 30, 2000.
We first note that some of the findings in the final audit report for Louisiana are inconsistent with the evidence before us. The statement in that report that "each time the fee has increased/decreased, the grant has simultaneously increased/decreased" is inconsistent with the evidence presented by Louisiana about the timing of changes in the amount of the grants compared to the timing of the increases in the fees. LA Exs. 412-415; Castile Decl. at � 9. As shown by the schedules set out above (which are based on this evidence), the assistance amount increased in October 1993, but the next increase in the per bed fee was not until July 1994. Between July 1994 and March 2000, the fee increased five times, but the assistance amount changed only once (in October 1997) and that increase was not simultaneous with any of the fee increases. CMS presented no evidence to support its audit finding. Also, the audit report's statement that the "amount of the [monthly] grant repayment ranges from 88% to 117% of the monthly fee" does not take into account the undisputed fact that a bed could be occupied by a private pay resident for as little as one day, but the resident might still receive the full monthly grant amount. Thus, the so-called "monthly fee" for any bed would depend on the number of days it was occupied during the month, while the monthly "grant repayment" would depend on the number of residents who occupied that bed for one day or more and were eligible for a grant. These variations were not taken into account by the final audit report in determining the "88% to 117% range." In any event, the conclusion based on the audit findings that the grant and tax programs were "closely related" is irrelevant to the issue of whether the positive correlation test was met, for the reasons explained in our general analysis above. As indicated, the other factors found in the audit report on which CMS relied in making its disallowance determination were primarily factors from which it inferred that the intent of the grant program was to offset the nursing facility tax. The report did not state how these factors show a positive correlation in the statistical sense, nor did it provide any statistical analysis of the relationship between alleged "indirect payments" to a provider from grants to its residents and the total tax cost to the provider. Under Louisiana's programs, the amount of the imputed indirect grant payments to any facility would not automatically increase with any increase in the total tax cost to the facility. The occupied bed days on which the tax was calculated could go up because of a bed being occupied by a patient who either was not a private pay patient or was not eligible for the grant, in which case the total tax cost for the facility would increase, but the grant amount would not. Moreover, the per occupied bed amount of the tax was set separately from the monthly grant amount, so the total tax cost could increase as a result of an increase in the amount of the tax per occupied bed, without any corresponding effect on the grant amount. Also, if a private pay patient eligible for a grant in one month stayed for only one day in the following month and the bed remained unoccupied for the rest of that month, the grant amount would be the same for each month but the tax would go down in the second month. The chart CMS presented during the oral argument does not show a positive correlation under the regulation because it does not compare the non-Medicaid payment to a nursing facility (in the form of grant payments to residents) to the total tax cost to the facility for any period. CMS Demonstrative Ex. 3. Instead, it compares the dollar amount per month for any one occupied bed during each year of the disallowance period with the monthly amount of the grant available to an eligible individual. (Even then, it shows that the monthly grant amount did not always increase as the monthly tax amount increased.) Moreover, the charting of the monthly tax amounts seems to show a tax decrease from $300 per month to a little over $100 per month in mid-1994, with the next increase being in mid-1995, whereas the evidence shows that the tax amount decreased from $10 per occupied bed day (a maximum of about $300 per month per bed) to $3.68 per occupied bed day (a maximum of about $110 per month per bed) effective July 1, 1993 and then increased to $3.94 per occupied bed day (a maximum of about $118 per month per bed) effective July 1, 1994. Also, the chart does not appear to show the increases in the grant amounts in October 1993 and October 1997. In other words, the chart not only does not make the correct comparison, but it is not factually accurate. (28) Thus, we conclude that the record here does not provide a legally sufficient basis for determining that the positive correlation test was violated.
The final audit report neither found an explicit or direct guarantee in Louisiana's statutes nor applied the indirect guarantee test in concluding that the nursing facilities were indirectly held harmless for part of the tax cost. No indirect guarantee can properly be found without applying the two-prong test, and Louisiana timely amended its tax program to meet the 6 percent test. Before us, CMS argues that there was a direct guarantee, but does not explain how the nursing facilities were assured of any indirect non-Medicaid payment. The nursing facilities were not assured that any of their private pay residents would qualify for or apply for a grant award, nor even that they would have private pay residents. Moreover, while the Louisiana statute authorized payment of the grants up to a maximum amount, payment of any grant amount was subject to the availability, appropriation, and allocation of funds for the program. The commissioner of administration had the authority to set the grant amount at zero, which he in fact did prior to the termination of the program. Thus, there was not even a guarantee to the private pay residents of any grant payment, much less any assurance to the nursing facilities. CMS argues before us, however, that there was a guarantee because a newspaper article said that the commissioner of administration had assured members of the House Appropriations Committee that "the rebate for private pay patients will remain as long as the provider fee is in effect." CMS LA Br. at 6. Louisiana pointed out that the article was issued in 1992, prior to changes Louisiana made to its statutes to conform them to the federal regulations. In any event, no provider would have any legally enforceable right based on this newspaper report, nor did the statement that the "rebate for private pay patients will remain" provide any assurance to a particular nursing facility that it would in fact receive an indirect payment by way of a grant, given all of the factors on which that would depend. Thus, we conclude that Louisiana's programs did not violate the guarantee test. Accordingly, we reverse the disallowances for Louisiana since there is no basis in the record for determining that there was a hold harmless provision in effect in Lousisiana during the disallowance periods, under either the positive correlation test or the guarantee test.
Effective July 1, 1992, Tennessee provided for an annual nursing home tax "based on the number of nursing home beds licensed by the state of Tennessee as of July 1, 1992, excluding beds in nursing homes specifically certified as intermediate care beds for the mentally retarded." The tax was to be "uniformly applied to all licensed beds at the rate of two thousand six hundred dollars ($2,600) per licensed bed." Licensed facilities which are owned or operated by an agency of the state were not excluded from paying the tax, nor were there any "exclusions, deductions or adjustments applied to the tax of any licensed facility different from any other such facility." The authorizing statute provided that the tax "may not be billed by the nursing home as a separately stated charge, but this shall not prevent the nursing home from adjusting its rates to defray the cost associated with the tax." The tax was considered a licensing fee for the privilege of doing business in the State. Tennessee Code Ann. � 68-11-216; TN Ex. 602. (29) Since the tax was on licensed beds, the tax was payable even if the bed was not occupied. Effective July 1, 1992, Tennessee created a "grant assistance program to provide financial support for eligible individuals residing in nursing homes licensed by the state of Tennessee, which individuals do not have their nursing home care paid for, in whole or in part, by a federal, state or combined federal-state medical care program." Tennessee Code Ann. � 71-5-1301; TN Ex. 603. To be eligible to apply for a grant, a person had to be residing in a state-licensed nursing home (other than an intermediate care facility for the mentally retarded) after June 30, 1992; have an annual net income (less the cost of nursing care) at or below 350 percent of the federal poverty standard; and not be receiving (or have recently applied for) federally or state subsidized assistance for the days for which grant assistance was sought. Tennessee Code Ann. � 71-5-1302. The resident had to apply quarterly on a prescribed form in order to receive a grant, and the nursing home was required to certify the number of days the individual resided in the nursing home during the quarter, and to provide quarterly information about the per day expense for care at the facility. Tennessee Code Ann. � 71-5-1303. Applications were to be "processed by the department on a first come, first served basis" and were "subject to the amount annually appropriated for this program in the general appropriations act." Individual payment "for adjusted annual incomes ranging from zero dollars ($0.00) to an amount calculated to be one hundred eighty-five percent (185%) of the individual poverty guidelines [was] six dollars and fifty cents ($6.50) per day." Maximum individual payments "for adjusted annual incomes ranging from one hundred eighty-six percent (186%) to three hundred fifty percent (350%) of the individual poverty guidelines [were] not to exceed six dollars ($6.00) per day, as determined by rules and regulations promulgated by the commissioner of health in order to meet the sums annually appropriated for this program." Tennessee Code Ann. � 71-5-1304. Grant payments were "non-assignable and payable only to the individual or such individual's legally authorized representative which shall not be the nursing home." Tennessee Code Ann. � 71-5-1305. The grant program was terminated effective July 31, 2001.
Before enacting its programs, Tennessee had performed a review that concluded "that the amended licensed bed tax on nursing homes and the grant assistance program to private pay nursing home residents would be permissible under the 1991 Act." Manning Decl. at 3. Tennessee provided information on its programs to the federal Office of Management and Budget (OMB) on July 2, 1992, noting that it had some flexibility in promulgating regulations to respond to any concerns if it received a quick response. TN Ex. 601. In a letter to Tennessee dated November 3, 1992, CMS's Director of the Medicaid Bureau raised questions about whether the grant program created a hold harmless situation, and asked for Tennessee's response so CMS could fully evaluate Tennessee's tax program. TN Ex. 605. Tennessee replied in a letter dated March 22, 1993, explaining all of the factors that made the grants vary and why there was nothing guaranteed, either to the private pay patient or to any nursing home in particular. TN. Ex. 608, at 1-2. Tennessee noted its understanding that the subject of private grants to nursing home residents was being discussed in connection with the interim final regulations and reserved further comments pending completion of the discussion. Id. at 2. Tennessee provided further information to CMS on August 13, 1993. TN Ex. 609. On December 19, 1994, CMS's Regional Administrator sent a letter to Tennessee stating that the grant program associated with the nursing facility tax violated the guarantee test and providing Tennessee an opportunity to submit information to support permissibility of the tax program. TN Ex. 613. Tennessee responded by letter dated March 13, 1995, providing an analysis of why its grant program did not guarantee any payments, either to residents or to facilities, and why there was no positive correlation between the grants and tax assessments. TN Ex. 615. CMS's response of July 19, 1995 stated without further analysis that the information provided did "not adequately support the permissibility of the nursing facility tax program." CMS Ex. 616. The next correspondence to Tennessee on the issue was a letter of September 1, 1999 (over four years later) from the CMS Director of the Center for Medicaid and State Operations. TN Ex. 618. This letter is similar to the ones to the other states, concluding that the State violated both the positive correlation test and the guarantee test. (30) On June 15, 2000, CMS sent a letter to Tennessee with a draft audit report, based on which CMS "preliminarily concluded" that the bed tax was impermissible. TN Ex. 621. The draft audit report based this finding on the facts that the tax and grant programs were enacted simultaneously, that nursing homes were allowed to adjust their rates to defray the costs of the tax, and that, in 1993-94, the total grants were equal to 38 percent of the total non-Medicaid portion of the taxes. Id. Tennessee responded by letter dated July 14, 2000, pointing out, among other things, that the draft report contained no statistical analysis to support the conclusion of a positive correlation, that there were a number of factors that made the grant payments vary differently from the taxes, and that the draft report neither applied the indirect guarantee test nor cited any direct guarantee. TN Ex. 622. On December 19, 2000, CMS issued its final audit report. TN Ex. 623. Among other things, this report cited the preamble example of a situation that might constitute a hold harmless, rejected Tennessee's arguments on the positive correlation test as not supporting "the position that a relationship between the nursing facility tax and the Nursing Home Resident's Grant Assistance Program does not exist," and asserting that the statutory wording of the guarantee test did not require a dollar for dollar repayment arrangement (emphasizing the words "any payment," "indirectly" and "any portion" in the statutory language). The report did not find that the two-prong indirect guarantee test was violated nor state that there was an explicit or direct guarantee. Nor did the report state how the relationship between the grants and taxes constituted a positive correlation in the statistical sense or provide any statistical analysis of the relationship. By letter dated January 19, 2001, the CMS Regional Administrator notified Tennessee that CMS was disallowing $519,864,853 in FFP for the period October 1, 1992 through September 30, 2000, on the basis that Tennessee's tax program was impermissible because "the nursing facilities are indirectly held harmless for some or all of the tax payment." TN Ex. 624. By letter dated June 11, 2001, the Regional Administrator notified Tennessee of an additional disallowance of $32,774,000 in FFP for the period October 1, 2000 through March 31, 2001, on the same basis.
Like with the other States discussed above, the factors on which CMS relied for its disallowance determination with respect to Tennessee do not show a positive correlation in the statistical sense. A mere "relationship" between the tax and grant programs is not sufficient to establish a positive correlation within the meaning of the regulations. The factors CMS cited do not show a positive correlation in the statistical sense, nor did CMS provide any statistical analysis of the relationship between alleged "indirect payments" to a provider from grants to its residents and the total tax cost to the provider. Under Tennessee's programs, the amount of the imputed indirect grant payments to any facility would not automatically increase with any increase in the total tax cost to the facility. The total amount of the tax could go up due to an increase in the number of licensed beds, but the indirect grant payment would not go up if the added beds were occupied by Medicaid residents or by private pay patients who were not eligible for grants. Moreover, the grant amount for any private pay resident could go up because of a decrease in the resident's income that would make the resident eligible for a higher grant award, but there would be no corresponding increase in the total tax cost. Further, if the decrease in the resident's income made the resident eligible for Medicaid, the resident would no longer be eligible for any grant award at all, so the amount of the indirect non-Medicaid payment would go down, but the total tax cost would remain the same. CMS's chart presented at the oral argument does not show a positive correlation under the regulation. CMS Demonstrative Ex. 4. It does not chart the total tax cost to any provider, but only the dollar amount per year on a licensed bed, compared to the dollar amounts of the grants to the two groups of eligible individuals, over the disallowance period. Thus, the record provides no legally sufficient basis on which to conclude that the positive correlation test was violated.
CMS did not base its finding that the guarantee test was violated on any finding of an explicit or direct guarantee in Tennessee law. Finding an indirect guarantee without applying the two-prong test is inconsistent with the regulation. (31) CMS now argues that there was a direct guarantee, but points to no language constituting such an assurance, and we find none. Under Tennessee's grant program, no nursing facility had any assurance that its private pay residents would qualify for and apply for grants, nor even that it would have any private pay patients. While the program may have provided an assurance to some private pay patients that they would receive grant awards if they applied, the private pay patients were not the ones paying the tax. The final audit report described the tax as a tax on the nursing facilities. TN Ex. 623, at 6. Moreover, the tax was imposed on licensed beds, regardless of whether the beds were occupied. The mere fact that the nursing facilities were permitted to defray the tax cost by raising rates does not transform the tax into a tax on the private pay patients, as CMS now suggests. Thus, we conclude that Tennessee's programs did not violate the guarantee test. Accordingly,
we reverse the disallowances for Tennessee since there is no basis in
the record for determining that there was a hold harmless provision in
effect in Tennessee during the disallowance periods, under either the
positive correlation test or the guarantee test.
Effective July 1, 1992, Illinois imposed an assessment "upon each long-term care provider . . . in an amount equal to $6.30 times the number of occupied bed days for the most recent calendar year ending before the beginning of" the State fiscal year (July 1, 1992). 305 Illinois Comp. Stat. 5/5B-2; IL Ex. 301, at 19. This assessment was considered a tax, but could "not be added to the charges of an individual's nursing home care that is paid for in whole, or in part, by a federal, State, or combined federal-state medical care program, except those individuals receiving Medicare Part B benefits solely." Id. "Long-term care provider" was defined to mean "(i) a person licensed by the Department of Public Health to operate and maintain a skilled nursing or intermediate long-term care facility or (ii) a hospital provider that provides skilled or intermediate long-term care services . . . ." 305 Illinois Comp. Stat. 5/5B-1; IL Ex. 301, at 18. "Occupied bed days" was to "be computed separately for each long-term care facility operated or maintained by a long-term care provider," and was defined to mean "the sum for all beds of the number of days during the year on which each bed is occupied by a resident (other than a resident receiving care at an intermediate care facility for the mentally retarded within the meaning of Title XIX of the Social Security Act)." Id. The assessment for the State fiscal year beginning July 1 was to be based on occupied bed days from the prior calendar year. 305 Illinois Comp. Stat. 5/5B-5; IL Ex. 301, at 20-21. On July 1, 1993, Illinois replaced this tax with a $1.50 fee per licensed bed per day. Effective October 1, 1992, Illinois established a Nursing Home Grant Assistance Program "to provide financial support for eligible individuals who reside in nursing homes and who do not have their nursing home care paid for, in whole or in part, by a federal, State, or combined federal-State medical care program . . . ." 305 Illinois Comp. Stat. 40/10; IL Ex. 302, at 2. The program applied only to State fiscal year 1993. Id. Any nursing home that was a residence to one or more eligible individuals and received an application for assistance from one or more applicants was designated as a "distribution agent" authorized to receive and distribute assistance payments to eligible individuals. 305 Illinois Comp. Stat. 40/5; IL Ex. 302, at 1. The nursing home was required to file a certification each quarter, providing relevant information, and to pay a "fee in the amount of $1 for each occupied bed day for the quarter immediately preceding the quarter in which the certification [was] made." 305 Illinois Comp. Stat. 40/15; IL Ex. 302, at 2-3. This fee could not be "billed or passed on to any resident or third-party payor." Id. The fees went into a fund for payment of grants and for other purposes, such as payment to the Illinois Department of Revenue (which administered the program) of administrative expenses of up to 2.5 percent of the moneys received. For each eligible individual, the amount of the quarterly grant could not exceed "$500 multiplied by a fraction equal to the number of days that the eligible individual's nursing home care was not paid for, in whole or in part, by a federal, State, or combined federal-State medical care program, divided by the number of calendar days in the quarter." 305 Illinois Comp. Stat. 40/20; IL Ex. 302, at 4-5. If the amount appropriated or available in the assistance fund was insufficient to make payments for all eligible individuals certified by the nursing homes, the payments were to be uniformly reduced. Id. To be eligible for a grant, an individual had to reside in a nursing home, and have an "annual adjusted gross income, after payment of any expenses for nursing home care" that did "not exceed 250% of the federal poverty guidelines" published by HHS. 305 Illinois Comp. Stat. 40/5; IL Ex. 302, at 1-2. Nursing homes were subject to penalties if they did not distribute the quarterly grant payments to the eligible individuals or their legally authorized representatives within 48 hours after receipt. 305 Illinois Comp. Stat. 40/35; IL Ex. 302, at 6-7. Illinois terminated its grant program effective June 30, 1993.
CMS was aware of the Illinois tax and grant programs shortly after they were enacted. IL Ex. 303, 304. In August of 1992, Illinois' Washington office wrote OMB, explaining that the Governor had attempted to comply with the 1991 law, but that Illinois could not have anticipated the mathematical hold harmless test in the draft interim final regulation (of which Illinois had seen a "leaked" copy) and that payments to intermediate care facilities for the mentally retarded might not meet the 75/75 percent test. Illinois requested that it be given time to amend the program, without calling a special legislative session (which it would have been required to do to meet the October 1, 1992 deadline in the draft rule). Il Ex. 306. On October 5, 1992, Congressman J. Dennis Hastert wrote the Deputy Secretary of HHS, expressing his concerns about the draft regulations, noting that the draft preamble cited a program that holds taxpayers harmless "that is very similar to Illinois' nursing home grant assistance program." Il Ex. 306, at 1. The letter described the intent of the grant program and requested that CMS change the preamble "in a manner that will give you the flexibility to determine that Illinois' program is not indirectly holding taxpayers harmless." Id. at 2 (emphasis removed). The letter also urged CMS to change the second prong of the indirect guarantee test to "the original 90/90 test devised" by CMS, and noted that OMB had agreed to allow states until January 1, 1993 to comply with the test, but that this did not do much for Illinois, whose legislature would not meet until around April 1. Id. The Acting Administrator of CMS responded in a letter dated December 4, 1992, urging the Congressman to comment on the interim final rule published on November 24, and noting that the preamble "provides examples of situations which might violate the hold harmless criteria." IL Ex. 397 (emphasis added). The interim final rule extended the deadline for compliance until April 1, 1993. Guidance on the 1991 law issued by the Chicago regional office in December 1992 did not mention the hold harmless provisions or grant programs. IL Ex. 308. Two years later, by letter dated December 19, 1994, the CMS Regional Administrator informed Illinois that its grant program violated the guarantee provision because it was "established to offset the nursing facility taxes." IL Ex. 313, at 1. Illinois responded by letter dated January 19, 1995, providing an analysis of why Illinois thought its programs did not violate any of the hold harmless provisions. IL Ex. 314. Illinois pointed out, among other things, that section 433.68(f)(3)(ii) of the final regulations provided that CMS would not disallow funds received from a tax in excess of 6 percent that did not meet the requirements of section 433.68(f)(3)(ii) if the state modified the tax to meet the requirements by September 13, 1993, and that Illinois had modified its tax program as of July 14, 1993, so that it would produce revenues less than 6 percent of the revenues received by the nursing homes. Id. at 3. The July 17, 1995 response from CMS stated without further analysis that the information provided did "not adequately support the permissibility of the nursing facility tax programs." IL Ex. 315. Illinois subsequently provided information on its tax receipts, but did not formally hear from CMS again regarding its grant program until a letter of September 1, 1999 (over four years later) from the Director of the Center for Medicaid and State Operations. This letter was similar to those sent on the same date to the other States, and for the first time asserted that both the positive correlation and the guarantee tests were violated. IL Ex. 318. Like letters to other States this one cited the preamble example and was premised on the view that the State "returned a portion of the revenue raised from the taxes to only private pay (non-Medicaid) patients and used the remainder as the State's share of Medicaid payments/increases." Id. at 4 (emphasis added). In a letter to Illinois dated June 15, 2000, the Director stated that CMS had "preliminarily concluded" that the $6.30 per occupied bed day tax and the $1.00 per occupied bed day fee imposed on nursing facilities in Illinois were impermissible. IL Ex. 320, at 1. CMS sent Illinois a draft audit report for comment. The draft audit report conclusion was based on the "simultaneous enactment" of the taxes and the grant program, the "implicit statutory language permitting nursing facility taxpayers to adjust their rates to private pay residents to defray the costs associated with the $6.30 occupied bed day tax," and the fact that "the amount of the grant payment available to a facility was based on the number of private pay patients at the facility, so it also varied directly based on the amount of the tax paid for those patients." Id., at 6 (emphasis added). The draft also said that the transition period in the regulations "does not apply when there is an explicit hold harmless arrangement such as in Illinois." Id., at 5. The reason given was that "our review indicates that the grant program that indirectly repays the provider was directly correlated to the amount of the tax paid by the provider, because it was enacted simultaneously and was based on the number of patient days for private pay patients (and the amount of the tax paid for those patients)." Id., at 5-6. By letter dated July 14, 2000, Illinois commented on the draft audit report. Illinois pointed out, among other things, that its grant program was adopted prior to the publication of the interim final regulation, that the preamble example was at best ambiguous, that no statistical analysis had been done of the correlation between the grants and taxes, and that the Illinois programs at issue had expired before publication of the final rule. Illinois also noted other factors, such as that the $1 fee could not be passed on to private pay patients and that the relationship of private pay bed days to all bed days varied from facility to facility. Illinois also challenged the draft conclusion on the guarantee test, saying that failure to explicitly prohibit payment by the grant recipients to the nursing homes is not equivalent to a guarantee of payment to those homes and that the auditors had not cited any direct guarantee. IL Ex. 321. The final audit report was transmitted to Illinois by letter dated December 19, 2000, from the Director of the Center for Medicaid and State Operations. Among other things, the final audit report states that "HHS expressly considered in the rulemaking process the situation at issue in Illinois . . . and stated that this situation would be considered as a hold harmless provision." TN Ex. 322, at 8 (citing the example in the preamble to the interim final rule). The final report also stated that the transition period at section 433.68(f)(3)(ii) "does not apply when there is an explicit hold harmless arrangement such as Illinois." Id. The reason given was again that the CMS review "indicates that the grant program that indirectly repays the provider was directly correlated to the amount of tax paid by the provider, because it was enacted simultaneously and was based on the number of patient days for private pay patients (and the amount of the tax paid for those patients)." Id. at 9. Thus, the final report, like the draft report, found an "explicit arrangement" (rather than an explicit or direct guarantee), despite Illinois' comment on this issue. The report did not address Illinois' points about why it thought the taxpayer provider was not guaranteed any payment through the grant program. Nor did the report state how any indirect payment to the provider from the grants to the private pay patients was positively correlated to the total tax cost to the provider or provide any statistical analysis of the relationship. By letter dated January 19, 2001, the Regional Administrator notified Illinois that CMS was disallowing $89,566,749 in FFP for the period October 1, 1992 through June 30, 1993 on the basis that the nursing facility taxes were impermissible "because the nursing facilities were indirectly held harmless for some or all of the tax payments." IL Ex. 323. (32) The disallowance amount was calculated using Illinois' revenues from both the $6.30 assessment and the $1.00 fee.
As indicated above, CMS's initial questioning of Illinois' programs under the positive correlation test was premised on the view that the preamble example described the programs and that Illinois "returned a portion of the revenue raised from the taxes to only private pay (non-Medicaid) patients and used the remainder as the State's share of Medicaid payments/increases." In 1994, after the draft preamble example had been revised to indicate that it was an example of a situation that "might" constitute a hold harmless and the preamble to the final rule had stated that a grant program could be structured to comply with the positive correlation test, CMS relied only on the guarantee test to question Illinois' programs. Then, starting in 1999, CMS again cited the positive correlation test, based on the premise that the preamble example fit (but without acknowledging the crucial "might" added to the preamble example). Now, before us, CMS asserts only that Illinois returned through the grant payments a portion of the tax revenue raised from taxes to private pay patients. CMS does not claim that the "remainder of the taxes were used as the State's share of Medicaid payments/increases." As discussed in our general analysis, however, asserting that the non-Medicaid payment is positively correlated to the amount of the tax is not the same as asserting that the non-Medicaid payment is positively correlated to the difference between the Medicaid payment and the amount of the tax. CMS acknowledges the difference, but continues to rely on the preamble example, even though the example uses a different variable for comparison than what CMS now uses. The draft audit report stated that "the amount of the grant payment available to a facility was based on the number of private pay patients at the facility, so it also varied directly based on the amount of the tax paid for those patients." This statement reflects factual errors. First, not all private pay patients were eligible for the grants. The amount of the grant payment to any facility would be based on the number of private pay patients at the facility only if all of them were eligible. Second, the assessment was based on total occupied bed days for the prior calendar year. The number of occupied bed days for private pay patients in that calendar year was not necessarily the same as the number in the following State fiscal year, in which the assessment was imposed. Third, although the $1.00 fee was based on the number of occupied beds in a facility in any quarter for which grant payments were made, Illinois law specifically prohibited the facilities from passing the $1.00 fee on to private pay patients. The ability of the facility to pass the tax cost on is a fundamental part of the CMS rationale for stating that a grant should be treated as an indirect payment to the provider to cover its tax cost, but this rationale simply does not apply to the $1.00 fee. (33) Even if the draft audit finding were accurate, however, it is meaningless for purposes of the positive correlation test. As explained above, the regulation interpreted the term "the amount of the tax" to mean "total tax cost," not the tax amount related to only some of the facility taxpayer's beds. CMS's more general argument before us that the grant and tax programs were closely related because the quarterly grant amount of $500 "closely approached the $567 per quarter that could be passed on" to residents eligible for a grant takes into account only the $6.30 assessment, not the total tax cost per quarter. CMS IL Br. at 5. If the $1.00 fee per occupied bed day is included, the quarterly tax amount would be about $666. Also, the quarterly amount of $500 for the grant payment was a maximum amount, subject to reduction. (34) In any event, a close relationship in per quarter amounts per bed would not establish a positive correlation of the relevant variables under the regulations. Moreover, some of the audit findings on which CMS based the disallowance effectively treat the taxes as though they were imposed on the private pay patients, rather than on the nursing facilities. CMS does not cite to any basis in the State laws for this view, however. The specific statutory prohibition against passing on the $1.00 fee, moreover, makes it even more clear that the fee could not reasonably be considered a tax on the private pay patients. Like with other States, CMS relies primarily on subjective factors for its conclusion, such as its view of the intent of the grant program. This reliance is misplaced since use of such factors was specifically rejected in the rulemaking as a basis for finding that a hold harmless exists. CMS's chart presented at the oral argument also is insufficient to show a positive correlation, within the meaning of the regulations. That chart merely illustrates that the presumed grant amount per occupied bed per quarter and the amount of the assessment (or the assessment plus the fee) per occupied bed per quarter each remained the same over the four quarters in which they were in effect. CMS Demonstrative Ex. 5. It does not show that the amount of the grant payment imputed to any facility increased as the total tax cost to the facility increased. (35) Under the Illinois law, an increase in the amount of the imputed indirect grant payment to a facility (based on presuming that the $6.30 assessment would be passed through) would not automatically mean an increase in the amount of the tax, or vice versa. The amount of the grant payment for any facility would go up from one quarter to the next if the number of beds in the facility occupied by private pay patients eligible for the grants went up. The total quarterly assessment to the facility would stay the same, however, since it was based on occupied bed days for the prior calendar year. Moreover, from one fiscal year to the next (if the programs had remained in effect for more than one year), the per quarter assessment amount for a particular facility would go up if the total number of occupied bed days from the relevant calendar year was more than in the prior calendar year. This increased amount would then stay the same for the entire fiscal year. The amount of the imputed indirect grant payment to the facility would not automatically go up, however. Instead, whether it went up or down for any quarter would depend on the number of occupied bed days in that quarter attributable to private pay patients eligible for the grants. Thus, we conclude that, based on this record, there is no legally sufficient basis for determining that the positive correlation test was violated by Illinois' programs.
CMS found that there was an "explicit hold harmless arrangement" in Illinois, but did not find an explicit or direct guarantee, nor explain how an "arrangement" constituted a guarantee. CMS now says there was a direct guarantee, based primarily on the fact that the private taxpayers were the ones bearing the burden of the tax. The private pay patients were not, however, the ones "paying the tax" under Illinois law (and the $1.00 fee could not even be passed on to them). We find no direct or explicit hold harmless guarantee to the taxpayers in the Illinois laws. CMS argued that, by designating the nursing homes as the distribution agents for the grant money, the legislature created a means by which the nursing home could assure that the grant money was used to pay them. CMS Il Br. at 21. This ignores the provision in the law that required that the nursing homes remit the grant payments to the patients within 48 hours of receipt (imposing penalties if this deadline was not met). In any event, nothing in the State law assured that any facility would have private pay patients who were eligible for grant payments. Nor was it within the control of the facility whether any resident who qualified for a grant would, in fact, apply for it. While certain aspects of the Illinois grant program made it more likely that eligible residents would apply, and it may have been likely that many facilities had at least one eligible private pay patient, a likelihood is not the same as a guarantee. Finally, CMS's basis for rejecting the Illinois argument that no funding reduction applies because Illinois modified its programs within the grace period in section 433.68(f)(3)(ii) of the regulations is not consistent with the language of that section. As codified, that section provides:
Illinois had enacted a tax in 1992 in excess of 6 percent that did not meet the requirements in paragraph (f)(3)(i). Illinois had modified that tax by July 1993 to come into compliance. On its face, the grace period would appear to apply. (36) Yet, CMS still disallowed funds for the period October 1, 1992 to July 1, 1993. The description of the grace period in the preamble to the final rule referred to the grace period very broadly as extending the period for compliance with the hold harmless provisions. 58 Fed. Reg. at 43,158; see also JAF Ex. 54 (State Medicaid Directors letter stating that the "compliance deadline for the hold harmless requirements has been extended to September 13, 1993"). CMS argues this did not mean that the grace period was to apply to all three of the provisions referred to as hold harmless provisions in the statute. CMS says that the grace period was meant to apply only to the guarantee test since that is the only one of the three provisions that specifically mentions the term "hold harmless" within the provision. At the very least, however, Illinois could have reasonably thought, based on the language of the final rule and the preamble, that the grace period applied to its tax program, and that, because Illinois modified the tax to comply within the grace period, it would not be subject to a disallowance. Indeed, the record shows that Illinois was considering whether to hold a special legislative session to modify its programs, but had been provided assurances that the grace period would be extended. Not giving Illinois the benefit of that period would be patently unfair, as well as inconsistent with the language of the provision. Thus, while the primary basis of our conclusion here is that there was no explicit or direct guarantee in Illinois law, an alternative basis is that the grace period applied to preclude any disallowance for the period at issue here. Accordingly, we reverse the disallowance for Illinois since there is no basis in the record for determining that there was a hold harmless provision in effect in Illinois during the disallowance period, under either the positive correlation test or the guarantee test. Conclusion For the reasons stated above, we reverse the disallowances in their entirety. |
|
| JUDGE | |
|
Cecilia Sparks Ford Donald F. Garrett Judith
A. Ballard |
|
| FOOTNOTES | |
|
1. At the time, the HHS operating division that administers Medicaid was called the Health Care Financing Administration, or HCFA. In this decision, we use CMS to refer to either HCFA or its successor, CMS. 2. We have placed a "sic" in this quote because the relevant comparison under the statute and regulations is between the non-Medicaid payment (rather than the "tax paid") and the difference between the Medicaid payment and the total tax cost. 3. The States (other than Illinois, which ended its program June 30, 1993) assert that they also met the 75/75 percent test. They either provide evidence to show this or assert that, since CMS had not claimed they did not meet the guarantee test until the draft audit report issued in 2000 (in some cases years after the tax and/or grant program had ended), they do not now have sufficient information to establish that they met the test. In any event, CMS made no finding for any State that the 75/75 percent test was not met. 4. Another statistics text describes the relationship as follows:
JAF Ex. 1, at 3-7. A common measure of association is the correlation coefficient r. Id. at 4-5. 5. Nor do we find merit to CMS's argument that it would be impossible to do such an analysis. First, CMS wrongly assumes that the correlation needs to track the relevant variables over time. While this might be one method of showing a positive correlation, there may be other ways, such as comparing the relevant amounts across facilities for a particular time period. Second, CMS does not explain how doing such an analysis would be any more complicated than applying the 75/75 percent test. 6. Contrary to what CMS argues, this Department's rejection of the use of subjective factors was not merely a rejection of an "egregiousness" standard. CMS Br. at 49. CMS's argument is inconsistent with the specific comment being addressed by the preamble statement (which appears in its entirety at JAF Exhibit 46, at 33-34). More importantly, it fails to take into account the fact that the 1991 law was drafted after Congress had rejected CMS's earlier regulations (which would have reduced federal funding based on factors such as statements of legislative purpose, use of dedicated tax funds for Medicaid reimbursement, or other legislative history showing a "linkage" between a tax and Medicaid payments). See 56 Fed. Reg. 56,132; 56,139 (Oct. 31, 1991). The history of the interactions between CMS and the States after the 1991 law was passed, moreover, indicates that CMS had determined to adopt numerical and statistical tests in implementing the 1991 law. See, e.g., JAF Ex. 33, at 9. 7. Although Exhibit JAF 25 references "total tax paid," the regulation adopted the term "total tax cost" - presumably to reflect that the total tax cost might not in fact be paid (if collected by offset) and that payments might be partially rebated. For our purposes, the relevant fact is that the regulation contemplates looking at the total amount of the tax cost to each taxpayer, not merely part of it, to determine whether a positive correlation exists. We also note that the "directly correlated" test used to determine whether a tax is uniformly imposed was also described as looking at the total tax cost. JAF Ex. 29, at 6; see 42 C.F.R. � 433.68(d)(2)(i)(B). Although the interim final rule provided that, to obtain a waiver, a state must meet both the "directly correlated" test and the hold harmless tests, CMS had originally treated the "directly correlated" and "positively correlated" tests as essentially the same, and the preamble to the final rule appears to use the terms interchangeably. 8. The 75/75 percent test looks at whether receipt of enhanced Medicaid payments and other state payments constitutes an indirect guarantee, but is not irrelevant to the positive correlation test. In response to a comment suggesting that correlation matters only if the degree of correlation is "significant," the preamble to the final rule stated that "the regulations do allow for a correlation to exist by a certain degree according to the statistical thresholds provided for in the hold harmless tests." 58 Fed. Reg. at 43,167. CMS concedes that the only statistical thresholds in the regulation are in the 75/75 percent test. Tr. at 83. CMS has never provided any specific guidance, however, on how to apply the threshold in the 75/75 percent test to determine the permitted degree of correlation, nor did it find here that the degree of correlation between the non-Medicaid payments to providers and their total tax costs exceeded that threshold. 9. It is undisputed that the "might" was added when Congressman J. Dennis Hastert of Illinois learned of the example in the draft rule and questioned whether such grants would be a hold harmless arrangement. IL Exs. 306, 307. Clearly, some in CMS were concerned about what they called "granny grants" and the example reflected their concern. However, the addition of the word "might" prior to publication (whatever its genesis) undercuts CMS's position here that the mere existence of a "granny grant" program constitutes a hold harmless arrangement. The "might" indicates that further analysis is needed to determine whether the positive correlation test applies. 10. The States provided evidence that the intent of this provision was to cover the situation of a tax that is technically imposed on a private pay patient, but merely collected by the nursing facility in which the patient resided. Smith Decl. at � 14. CMS provided no evidence to the contrary. 11. Indeed, as explained below, the Illinois disallowance includes an FFP reduction for revenues from a $1 per bed fee, even though the state statute specifically precluded the nursing facility from passing this fee on to the private pay patient. 12. The States point out that the example addresses only grant programs and is silent on tax credits. They argue that the fact that the positive correlation test uses only the word "payment," whereas the guarantee test refers to "payment, offset, or waiver," means that the positive correlation test was not intended to apply to credits. They also say that the timing of any payment from a private pay patient to a provider to cover the cost of a passed-through tax compared to the timing of any tax credit to the private pay patient shows that the credit would not be used to pay the provider. In view of how we resolve this case, we do not need to reach these issues here. 13. The States point out that, shortly after the final rule was published, CMS officials stated that there was "no official word yet" on whether "granny grants" were permissible although they "appear to be bad." JAF Ex. 53, at 1. This statement undercuts CMS's current position that the preambles definitively resolved the issue and that no further guidance was needed. 14. For Illinois, the total assessment amount depended on the number of occupied bed days for the prior calendar year, which could differ from the number of occupied bed days in the tax period. 15. CMS several times states that the statute uses the term "indirect payments." The statutory language, however, refers to a situation where a state "provides (directly or indirectly)" for any payment that guarantees to hold a taxpayer harmless. While the preambles refer to direct or indirect payments (or compensation) with reference to the "positive correlation" test, the regulations and the preamble discussion regarding the "guarantee" test focus on whether the guarantee of a payment to the provider is direct (explicit) or indirect (inexplicit). 16. CMS argues that the State need only "provide" for the payment, and that we should reject the States' reliance on the fact that not all providers in fact passed the tax through since any guarantee might require some action on the part of the recipient of the guarantee to invoke the guarantee. Since there is some merit to this argument, we do not rely for our result on the fact that not all providers passed the tax through. We nonetheless conclude that factors outside of the providers' control do undercut the notion that the providers were "explicitly" or "directly" guaranteed repayment of part of the tax. 17. Moreover, nothing in the statute, regulations, or preambles equates "targeting" a payment with guaranteeing a payment. Although some of the States' laws did assure that the private pay patients could receive a grant or credit if they met the eligibility requirements and applied on time, one State law on which CMS relied for the disallowances only authorized the grant payments, making actual payment and the payment amount discretionary (Louisiana), and another made payments subject to a cap and distributable on a first come, first served basis (Tennessee). We question how these provisions could be considered a guarantee, even to the private pay patient. 18. CMS points to a statement by Senator Durenberger expressing a hope that the legislation would mean that no health care-related taxes would be enacted. See 137 Cong. Rec. at S18190. 19. The States point out that, the next day, the CMS Administrator issued a statement indicating that States were being notified that "they appear to be in violation" of the law - a statement that is less definitive. JAF Ex. 57. We note that, to simplify our discussion, we refer to the regional officials here by their titles only, without specifying the names or the relevant HHS region. 20. This letter briefly describes the information as demonstrating that the Federal Government was paying more than its share of Medicaid because Hawaii "returned a portion of the revenue raised from the tax through tax credits to only the private pay (non-Medicaid) patients and used the remainder as the State's share of Medicaid payments/increases." HA Ex. 217, at 4. Before the Board, however, CMS relies only on the tax credits and does not assert that the remainder of the tax revenue was used for Medicaid payments/increases. 21. Hawaii argues that the positive correlation test refers only to a "payment" to the provider or others paying the health care-related tax and does not refer to a credit or offset like the guarantee test. Hawaii also points out that the preamble example on which CMS relies refers only to grant programs, not to tax credits for private pay patients. Therefore, Hawaii argues, it could have reasonably thought that the positive correlation test would not apply to its tax credits. We do not need to resolve this issue here, so for purposes of our analysis, we are assuming that, by making the tax credits available to the private pay patients in a facility, Hawaii was providing indirectly for non-Medicaid payments to the facility. 22. Hawaii points out that nothing in the State law required that the resident would pay the facility the amount of the tax credit, once obtained. We assume, for purposes of our decision, that Hawaii should nonetheless be considered as having provided indirectly for a payment to the provider. We note, however, that the connection between the payment to the facility to cover any passed through tax and any tax credit the resident obtains in the following year is somewhat tenuous, particularly if an insurance company, rather than the resident, is actually paying the facility. 23. Moreover, even assuming that the tax credits effectively offset the non-Medicaid part of the total provider tax burden, the effect would be essentially the same as a 6 percent tax imposed only on ICFs/MR (whose residents are almost exclusively Medicaid) that is nonetheless considered permissible under the indirect guarantee test. 24. Charges that were included in gross receipts for purposes of the tax included charges for room and board, medical supplies, ancillary services, telephone, cable TV, and personal services such as beautician, barber, and laundry services, but did not include bad debts and other uncollectible charges to patients, which could be deducted from taxable charges for the month in which they were written off as uncollectible. Moreover, charges were to be adjusted downward for purposes of the tax if there were contractual arrangements (including with Medicaid, Medicare, or other third party payors such as private insurers) under which full charges were not reimbursed. See, e.g., ME Ex. 505, at 3. 25. The catastrophic medical expenses credit was equal to 2.7 percent of an individual's federal tax deduction for itemized medical and dental expenses. ME Exs. 521, 525. Maine described this credit to CMS, which seemed satisfied with it. MacLean Decl. at � 8. CMS did not disallow any funds based on the replacement credit. 26. Maine presented evidence that only 1,512 of the 4,169 Maine taxpayers eligible for the credit claimed the credit on their 1993 Maine income tax returns. Gerard Decl. � 5. 27. The statute did provide that the fee "shall be considered an allowable cost for purposes of insurance or other third party reimbursements and shall be included in the establishment of reimbursement rates." Louisiana Rev. Stat. � 46:2605, � 46:2625; LA Exs. 403, 404. Including a fee as an allowable cost does not necessarily mean that the rate will increase by the amount of the fee. Prospective rate-setting systems may have ceilings on allowable costs or use other techniques that mean that full allowable costs are not reimbursed. It appears that Louisiana used a Medicaid reimbursement methodology that, during the disallowance period, set rates at either the 60th or 80th percentile of provider costs. LA Ex. 428. 28. CMS argues that Louisiana "fluctuated the rates of the tax and grant in a way they hoped might cover the correlation between the two and insulate them from the law." CMS LA Br. at 22. Louisiana presented evidence that it reduced the tax rate initially and later adjusted it in order to meet the 6 percent test. While the evidence on which CMS relies shows that Louisiana was conscious that, if it raised the grants at the same time as it raised taxes, CMS might assert that this showed a correlation, and therefore Louisiana sought to avoid this situation, this can also be viewed as trying to comply with the law, rather than to circumvent it. Since evaluating intent often depends on one's perspective, the preamble wisely rejected reliance on such subjective factors. 29. As of July 1, 2001, the amount of the tax was reduced to $2,225 per licensed bed per year. 30. This letter stated that Tennessee "returns a portion of the revenue raised from the tax using grants that go only to private pay (non-Medicaid) patients and used the remainder as the State's share of Medicaid payments/increases." This description appears to track the example in the preamble of a situation that might violate the positive correlation test. Before us, however, CMS did not claim that "the remainder" of the tax revenues were used for Medicaid payment increases, and specifically said it was not relying on the part of the positive correlation test that uses the difference between the Medicaid payment and the total tax cost as one of the variables. 31. Tennessee presented evidence that, even considering both increased Medicaid reimbursement and grant payments for the years for which Tennessee still had data, it still met the 75/75 percent standard. Morgan Decl., Att. A. Tennessee also points out, without contradiction, that some Medicaid rate increases were to cover costs other than tax costs, such as an increase mandated by the Omnibus Budget Reconciliation Act of 1987 for nurse aide costs. 32. Although Illinois' programs were in effect as of July 1, 1992, no reduction was taken for the period prior to October 1, 1992. Since Illinois had previously had a different impermissible tax program, it was given through September 30, 1992 to come into compliance with the 1991 law, under transition provisions not at issue here. CMS IL Br. at 3, n.2. 33. CMS also seems to assume with respect to the Illinois grant program that, because the facilities were distribution agents for the grants, the facilities would simply retain the grant amounts to offset what the private pay patients owed them. This ignores the provision in Illinois law that the facility was required to remit the grant payments to the residents or their legal representatives within 48 hours of receipt. 34. The quarterly amounts for the first three quarters that grant payments were made were in fact reduced (to $430.58, $355.05, and $357.06) because the collected fees were insufficient to pay the $500 maximum. IL Ex. 310. Subsequently, the Illinois legislature passed a supplemental appropriation to retroactively bring the amounts up to $500 for each quarter. IL Exs. 311, 312. 35. Illinois says it did a statistical analysis showing that any correlation was inconsequential ("less than 12 percent of the variation in the level of total grants issued to residents of nursing facilities can be explained by the taxes paid by nursing facilities) or trivial ("less than one percent of the variation in the level of total grants issued to residents of nursing facilities can be explained by the facilities' Medicaid payments"). Il Ex. 324. CMS devoted only a footnote to this analysis, calling it "irrelevant" and arguing alternatively that it does show a positive correlation and that the correlation need not be consequential. CMS IL Br. at 31, n.13. We do not consider the reference to the Illinois analysis to provide a basis for disallowance, for several reasons. Since CMS treated the analysis as irrelevant, we simply do not know whether it was valid and used the correct variables. Second, the preamble to the final rule said that the regulation permits some degree of correlation, which CMS concedes must have been referring to the 75/75 percent indirect guarantee test. Tr. at 83. CMS provided no guidance on how to apply that test in this context, however, so we cannot determine that the threshold was exceeded (and it would appear that it was not). Third, even if the analysis reliably showed a positive correlation exceeding the threshold, applying it to programs that were enacted before any regulatory guidance was given and that were then timely modified to comply would raise notice questions under the Administrative Procedure Act, which we would resolve in favor of Illinois. 36. The interim final rule did appear to set a different effective date for finding that a hold harmless was in effect based on an explicit guarantee, by stating that "any tax in effect before April 1, 1993, containing an explicit guarantee will also be considered to violate the statutory hold harmless provision." This statement does not appear in the final rule, however, and the preamble to the final rule does not limit the later effective date to state laws that violate only the indirect guarantee test. |
|