Our first step in interpreting a statute is to determine whether the language at issue has a plain and unambiguous meaning with regard to the particular dispute in the case. . . .

The plainness or ambiguity of statutory language is determined by reference to the language itself, the specific context in which that language is used, and the broader context of the statute as a whole.

Robinson, 519 U.S. at 340-41.

The Court observed that "[a]t first blush" the term "employees" in § 704(a) seems to refer to current employees. However, the Court concluded that such a perception "does not withstand scrutiny" of the meaning of § 704(a). See id. at 341. The Court found that "there is no temporal qualifier in the statute such as would make plain that § 704(a) protects only persons still employed at the time of the retaliation." Id. Thus, the Robinson Court endorsed an inquiry into the meaning of statutory terms that extends beyond consideration of isolated words.

Finding § 704(a) ambiguous standing alone, the Court turned to an examination of the definition of the term "employee" included in § 701(f) of Title VII and determined that the statutory definition is itself inherently ambiguous as to the coverage of former employees:

Title VII's definition of "employee" likewise lacks any temporal qualifiers and is consistent with either current or post employment. Section 701(f) defines "employee" for purposes of Title VII as "an individual employed by an employer." 42 U.S.C. § 2000e(f). The argument that the term "employed," as used in § 701(f), is commonly used to mean "[p]erforming work under an employer-employee relationship," Black's Law Dictionary 525 (6th ed. 1990), begs the question by implicitly reading the word "employed" to mean "is employed." But the word "employed" is not so limited in its possible meanings, and could just as easily be read to mean "was employed."

Id. at 342.

The Court canvassed other provisions of Title VII in which the term "employees" was alternately inclusive and exclusive of former employees, commenting:

But those examples at most demonstrate that the term "employees" may have a plain meaning in the context of a particular section--not that the term has the same meaning in all other sections and in all other contexts. Once it is established that the term "employees" includes former employees in some sections, but not in others, the term standing alone is necessarily ambiguous and each section must be analyzed to determine whether the context gives the term a further meaning that would resolve the issue in dispute.

Id. at 343-44.

Satisfied that the term "employees" as used in § 704(a) is ambiguous, the Court resolved the ambiguity by turning to the broader context provided by other sections of Title VII. In particular, the Court observed that § 703(a), which protects against discriminatory discharge, provides an example of a section that "plainly contemplate[s] that former employees will make use of the remedial mechanisms of Title VII." Id. at 345. Finally, the Court agreed with the EEOC that an interpretation of § 704(a) to include former employees would be consistent with the purpose of the anti-retaliation provision to maintain access to statutory remedial mechanisms, and stated that "to hold otherwise would effectively vitiate much of the protection afforded by § 704(a)," "would provide a perverse incentive for employers to fire employees who might bring Title VII claims," and would make it possible for employers "to retaliate with impunity against an entire class of acts under Title VII--for example, complaints regarding discriminatory termination." Id. at 345-46.

In the case at bar we have been presented with markedly discrepant views as to the pertinence of Robinson to the ADA. Appellant and the EEOC as amicus argue that the approach taken by the Court in Robinson undermines the Gonzales conclusion that the definitions of "qualified individual with a disability" in § 12111(8), of "employee" in § 12111(4), and of "discriminate" in § 12112(b) unambiguously exclude former employees. We are urged by appellant and amicus to resolve the ambiguity attributed to those definitions and hence to § 12112(a) ("No covered entity shall discriminate against a qualified individual with a disability because of the disability of such individual in regard to . . . terms, conditions, and privileges of employment.") by looking to the broader context of the ADA and ultimately concluding that Congress intended to include former employees within the coverage of Title I. To this end, Johnson and the EEOC invoke opinions of the Second and Third Circuits--Castellano v. City of New York, 142 F.3d 58 (2d Cir. 1998) and Ford v. Schering-Plough Corp., 145 F.3d 601 (3d Cir.), cert. denied, 525 U.S. 1093 (1999)--which rely on Robinson in holding that former employees are covered by Title I of the ADA. Conversely, appellee K Mart asserts (1) that the similarity in the definitions of "employee" in Title VII and the ADA is irrelevant; (2) that Robinson does not supersede Gonzales because, in Gonzales, this court distinguished the ADA provision at issue from the Title VII anti-retaliation provision; and (3) that the definition of a "qualified individual with a disability" does contain a temporal qualifier. In support of this last proposition K Mart puts strong reliance on Weyer v. Twentieth Century Fox Film Corp., 198 F.3d 1104 (9th Cir. 2000), in which, subsequent to Robinson, the Ninth Circuit expressed its belief in the continued viability and correctness of Gonzales.6

Although, fundamentally, the task of statutory interpretation begins with the text, in this case, given our prior ruling, legal precedent provides our starting point. Thus, we assess the continuing viability of the components of the rationale in Gonzales, turning first to the aspects that appear to stand in the greatest tension with the Robinson analysis. With little difficulty we conclude that the part of Gonzales wherein we decided that the meaning of the term "employees" in § 12111(4) plainly excluded former employees has been undermined by Robinson. Having noted in Gonzales that the ADA definition of "employee" was imported from Title VII, now that the Supreme Court has held that the term is ambiguous in Title VII, we can no longer maintain that the term "employee" unambiguously excludes former employees in the ADA. Moreover, we recognize that the force of the reasoning in Robinson also undermines this court's previous finding that the term "discriminate," as defined in § 12112(b), plainly excludes former employees. Once it is accepted that the use of the words "applicant or employee" does not unambiguously exclude former employees, it is clear that the definition of the term "discriminate"--see, e.g., §§ 12112(b)(1)-(b)(3), quoted above--lacks language that expressly defines the temporal scope of cognizable discrimination. As defined, it is certainly within the linguistic possibilities of the term "discriminate" that an employer is prohibited from "limiting, segregating, or classifying" a former employee "in a way that adversely affects the opportunities or status of such . . . employee because of the disability of such . . . employee," see § 12112(b)(1); that an employer is prohibited from "participating in a contractual or other arrangement or relationship that has the effect of subjecting a covered entity's" former employee to discrimination, see § 12112(b)(2); and that an employer is prohibited from "utilizing standards, criteria, or methods of administration . . . that have the effect of" discriminating against former employees "on the basis of disability." See § 12112(b)(3). In light of the Court's emphasis in Robinson on the importance of temporal qualifiers, we must now acknowledge that it is hard to argue that §§ 12112(b)(1)-(b)(3) unambiguously contemplate discrimination encountered solely by job applicants and/or current employees. While we agree with K Mart that the modification of our understanding of these two terms is not necessarily dispositive, we do not agree that this new perspective is irrelevant to our analysis. Indeed, the interpretation of the terms "employee" and "discriminate" formed much of the basis for our holding in Gonzales.

Having determined that the above components of the reasoning in Gonzales have been undercut by Robinson, we are left to discern the viability of the distinctions that we drew in Gonzales between general Title I discrimination claims, on the one hand, and retaliation claims, on the other. Ultimately, we must determine whether the fact that the ADA provision under consideration refers to "discriminat[ion] against a qualified individual with a disability," § 12112(a), rather than to Title VII's "discriminat[ion] against any of [the employer's] employees," 42 U.S.C. § 2000e-3(a), renders the protections afforded to former employees in ADA cases significantly narrower than the protection embodied in Title VII.

In Gonzales, we distinguished the question of the coverage of former employees under § 12112(a) of the ADA from the question of the coverage of employees under § 704(a) of Title VII, 42 U.S.C. § 2000e-3(a), based on what we perceived as the peculiar necessity of allowing former employees to bring retaliation claims. However, this approach is out of step with Robinson, in which the Court reasoned that "[o]nce it is established that the term"--in the case at bar "qualified individual with a disability"--"includes former employees in some sections, but not in others, the term standing alone is necessarily ambiguous and each section must be analyzed to determine whether the context gives the term a further meaning that would resolve the issue in dispute." Robinson, 519 U.S. at 343-44. The Robinson Court determined that the word "employee" in § 2000e-3(a) is ambiguous by looking to sections of Title VII that are not related to retaliation in which "employee" clearly includes former employees. See Robinson, 519 U.S. at 345 ("[S]everal sections of the statute plainly contemplate that former employees will make use of the remedial mechanisms of Title VII."). Thus, it was not by reference to the purpose of § 2000e-3(a) that the Robinson Court established the ambiguity of the section. By contrast, in Gonzales we turned our attention to the underlying purposes of the statutory provisions prematurely. Following the reasoning of Robinson, we now recognize that language in Title I of the ADA leaves the meaning of "qualified individual with a disability" in 42 U.S.C. § 12112(a) open to more than one interpretation. As with provisions of Title VII other than 42 U.S.C. § 2000e-3(a), provisions of Title I other than 42 U.S.C. § 12112(a) "plainly contemplate that former employees will make use of the remedial mechanisms of" Title I of the ADA. For example, Title I of the ADA protects employees from discriminatory discharge and authorizes remedial action including reinstatement--the former a form of discrimination and the latter a remedy that the Robinson Court associated with the coverage of former employees. See Robinson, 519 U.S. at 342-43.7 The similarity between Title VII of the Civil Rights Act of 1964 and Title I of the ADA in this respect strongly suggests that the term "qualified individual with a disability," like the term "employee," is ambiguous as to the coverage of former employees. We note that narrowly construing the term "qualified individual with a disability" to exclude former employees would eviscerate Title I's prohibition of "discrimination against a qualified individual with a disability because of the disability of such individual in regard to . . . the . . . discharge of employees." 42 U.S.C. § 12112(a). Therefore, following Robinson, the distinction that we relied on in Gonzales between general anti-discrimination provisions and anti-retaliation provisions, see Gonzales, 89 F.3d at 1529, can no longer be understood as dispositive of the meaning of the term "qualified individual with a disability."

We now turn directly to an analysis of the comparative temporal significance of the terms "qualified individual with a disability" and "employee." In Gonzales, we relied, without elaboration, on the statutory definition of "qualified individual with a disability" in § 12111(8) ("an individual with a disability who, with or without reasonable accommodation, can perform the essential functions of the employment position that such individual holds or desires") for the holding that a plaintiff must either hold or desire a position with the defendant-employer "at or subsequent to the time the alleged discriminatory conduct was committed." 89 F.3d at 1526. Recently, the Ninth Circuit provided a more elaborate analysis in support of a similar conclusion. The Ninth Circuit opined that the use of the term "qualified individual" and the use of the present tense forms of the verbs "hold" and "desire" reveal congressional intent to "unambiguously exclude[] totally disabled persons" and "to unambiguously exclude former employees." Weyer, 198 F.3d at 1112. According to the Weyer court's reading, the word "holds" refers exclusively to current employees and the term "desires" refers exclusively to applicants. Thus, the Ninth Circuit found that the plain language of Title I is "well designed to help people get and keep jobs, not to help those no longer able to work get disability pay." Id. In contrast, the Second and Third Circuits have found that the term "qualified individual with a disability," taken in context, is not susceptible of such an unambiguous meaning. See Ford, 145 F.3d at 601; Castellano, 142 F.3d at 58. Rather, the Third Circuit, in Ford, recognized a "disjunction between the ADA's definition of 'qualified individual with a disability' and the rights that the ADA confers," noting that former employees must be eligible to bring Title I claims in order to effectuate Title I's prohibition against discrimination in terms of fringe benefits. See Ford, 145 F.3d at 605-06. Thus, the Third Circuit held that the temporal scope of § 12111(8) is ambiguous:

As with the term "employees" in Title VII, the ADA contains an ambiguity concerning the definition of "qualified individual with a disability" because there is no temporal qualifier for that definition. Congress could have restricted the eligibility for plaintiffs under the ADA to current employees or could have explicitly broadened the eligibility to include former employees. Since Congress did neither but still created rights regarding disability benefits, we are left with an ambiguity in the text of the statute regarding eligibility to sue under Title I.

Ford, 145 F.3d at 606-07; see also Castellano, 142 F.3d at 67.

Based on our reading of Robinson, we are convinced that this court in Gonzales and the Ninth Circuit in Weyer have found more clarity in the language of § 12111(8) than is justified. Although, "at first blush," the use of the words "qualified" and "holds or desires" may appear to refer to applicants and current employees, this initial impression does not withstand the level of scrutiny required by Robinson, particularly in light of the fact that at least some former employees were intended beneficiaries of Title I--that is, victims of discriminatory discharge and those who would benefit from reinstatement. See Robinson, 519 U.S. at 341. With regard to the Ninth Circuit's argument that all totally disabled persons are unambiguously excluded from Title I as a result of the use of the term "qualified," see Weyer, 198 F.3d at 1112, we note that the word "qualified" is as temporally indefinite as the word "employee." Similarly, we are unpersuaded by that court's interpretation of § 12111(8) in which "hold" correlates with current employees and "desire" correlates with applicants and the two words taken together unambiguously exclude former employees. In Robinson, the Court, construing the use of the words "employees or applicants" in § 704(a), disapproved of the proposition that the inclusion of the term "applicants" coupled with the term "employees" excludes former employees by negative inference. The Court concluded that because "applicants" is not synonymous with "future employees," there is "no basis" for an assumption that Congress intentionally elected to provide coverage for future and current employees but to deny coverage for former employees. See Robinson, at 519 U.S. at 344-45. Consistent with this logic, we find that the term "desire" does not unambiguously exclude former employees when paired with the term "hold," in part because the word "desire" is susceptible of application to employees as well as applicants. Set against Robinson and the evidence that Title I protects former employees from discriminatory discharge and provides them the remedy of reinstatement, we cannot maintain the holding that the definition of "qualified individual with a disability" contains an unambiguous temporal qualifier.

We further conclude that our reliance in Gonzales on the legislative history of the ADA does not withstand the analysis in Robinson. This court stated:

[T]he legislative history of the ADA specifically states that the purpose of including the phrase "essential functions" within the QID definition is to "ensure that employers can continue to require that all applicants and employees, including those with disabilities, are able to perform the essential, i.e., the non-marginal functions of the job in question [sic]."

Gonzales, 89 F.3d at 1527 (citation omitted) (emphasis supplied). We continued, "[t]hus, a review of both the ADA and its legislative history suggests that Congress intended to limit the protection of Title I" to either current employees or applicants. See id. However, the use of the words "applicants and employees" is no less ambiguous in § 12111(8) of the ADA, or in the legislative history, than it is in the provisions of Title VII that were analyzed by the Robinson Court. Seen in this light, the inclusion of the phrase "essential functions" simply clarifies the content of the word "qualified," not its temporal significance.

Finally, finding the term "qualified individual with a disability" ambiguous as used in § 12112(a), we now undertake to resolve the ambiguity in a manner consistent with the logic of Robinson. As the Robinson Court did, we look to the other sections of the statute that "plainly contemplate that former employees will make use of the remedial mechanisms" of Title I of the ADA. See Robinson, 519 U.S. at 345. As mentioned above, we find particularly instructive two sub-sections of Title I that plainly reflect congressional intent to include some former employees within the phrase "qualified individual with a disability"--the prohibition of discriminatory discharge based on disability in § 12112(a) and the provision of the remedy of reinstatement in § 12117(a). Reading § 12112(a) narrowly to exclude coverage of former employees would not only bar a whole class of plaintiffs from access to remedies regarding the administration of post-employment benefits, and as a result create a "perverse incentive" for employers to interfere with the post-employment benefits of former employees, see Robinson, 519 U.S. at 346, it would also nullify those portions of Title I that protect against discriminatory discharge and that authorize courts to order reinstatement. Against this background, we hold that Johnson, as a former employee, is entitled to bring a claim against his former employer under § 12112(a) of Title I of the ADA.

We turn, now, to the substantive coverage afforded by § 12112(a).

II Whether Differential Benefit Levels for Mental and Physical Disabilities in an Employer-Sponsored Long-Term Disability Plan Can Be Found to Violate Title I of the ADA

The substantive issue before this court is whether a former employee can state a claim under Title I of the ADA based on an employer-sponsored long-term disability insurance plan that provides more limited benefits to individuals with mental disabilities than to those with physical disabilities. In the case at bar, K Mart's LTD plan provides disability benefits (payments in lieu of salary) to employees who have been totally disabled due to mental illness for a maximum of 24 months. In contrast, the same disability benefits are available to employees who have been totally disabled due to physical illness until age 65. The District Court framed the issue as follows: "[T]he question for the Court is whether the Plan's differential treatment of mental and physical illnesses qualifies as discrimination under the ADA." The District Court concluded, in agreement with a number of our sister circuits, that such differential treatment does not fall within the meaning of the term "discrimination" under the ADA:

[S]everal federal circuit courts have addressed and resoundingly rejected the argument that the ADA is violated when an employer or a state provides varying benefits or coverage based on the type of disability. See, e.g., Lewis v. Kmart Corp., 180 F.3d 166, 170 (4th Cir. 1999) (Title I, § 102(a) of the ADA, does not require a long-term disability plan that is sponsored by a private employer to provide the same level of benefits for mental and physical disabilities); Rogers v. Dept. of Health and Environmental Control, 174 F.3d 431, 436 (4th Cir. 1999) (denying someone with a mental disability the same benefits as someone with a physical disability does not violate Title II of the ADA); Ford v. Schering-Plough Corp., 145 F.3d 601, 608 (3d Cir. 1998) ("ADA does not require equal coverage for every type of disability"), cert. denied, 119 S. Ct. 850 (1999); Parker v. Metropolitan Life Ins. Co., 121 F.3d 1006, 1019 (6th Cir. 1997) (en banc) (same), cert. denied, 522 U.S. 1084, 118 S. Ct. 871 (1998); EEOC v. CNA Ins. Cos., 96 F.3d 1039, 1045 (7th Cir. 1996) (same). The Court is persuaded by the reasoning of these courts.

Thus, the District Court dismissed appellant's claim.

We now address the question of the permissibility of such distinctions in LTD plans for the first time.8 The answer to this question is dependent upon the meaning that we ascribe to three sections of the ADA, two of which--42 U.S.C. § 12112(a) (the general rule against disability-based discrimination in employment) and 42 U.S.C. § 12112(b) (the definition of the term "discriminate" as used in Title I)--have been introduced in Part I of this opinion. The third statutory section important here is the so-called "safe harbor" provision, 42 U.S.C. § 12201(c), which provides both an exemption from liability for bona fide benefit plans and a limited exception to this exemption.9 In construing the meaning of these sections, we note that remedial statutes such as the ADA are generally to be accorded broad construction. See Steger v. Franco, Inc., 228 F.3d 889, 894 (8th Cir. 2000) ("[T]he ADA is a remedial statute . . . and should be broadly construed to effectuate its purpose."); Antenor v. Osnel, 88 F.3d 925, 933 (11th Cir. 1996) (interpreting FLSA and Migrant and Seasonal Agricultural Worker Protection Act); Jones v. Metropolitan Atlanta Rapid Transit Authority, 681 F.2d 1376, 1380 (11th Cir. 1982) (interpreting Rehabilitation Act).

In Part II(A)(1), we begin our analysis of whether the mental health cap in K Mart's LTD plan constitutes a legally cognizable form of discrimination with an examination of the scope of the concept of discrimination prohibited by Title I of the ADA. Thus, we look primarily to the construction of §§ 12112(a) and (b). In Part II(A)(2), we consider whether contemporaneous legislative history clarifies congressional intent regarding the scope of the anti-discrimination principle in Title I when applied to LTD plans. Finally, in Part II(B), we turn to the construction of the safe harbor provision in § 12201(c) to determine whether Congress has, as a matter of law, exempted LTD plans like K Mart's from liability under the Act.

(A) Determining the Scope of the Concept of Discrimination Embodied in Title I of the ADA

(1) The Language of the Statute

Whether the distinction between physical and mental disability benefits that marks K Mart's LTD plan is a violation of the ADA or a non-discriminatory differentiation turns largely on divergent views of the nature of the concept of discrimination embodied in Title I of the ADA. The District Court and a number of courts of appeals have concluded that claims based on differential treatment of people with mental and physical disabilities in LTD plans are not cognizable under Title I of the ADA because the Act only requires that the disabled receive access to the same benefits that are available to the non-disabled. In this vein, K Mart contends that the ADA merely prohibits discrimination between the disabled and the non-disabled. Appellant and the EEOC, as amicus, argue that the concept of discrimination in the ADA is not so limited. It is this core disagreement to which we address ourselves initially. We emphasize, however, that we clarify the concept of discrimination in Title I as a threshold matter. A finding that Title I prohibits a wider compass of discrimination than that between the disabled and the non-disabled merely allows us to proceed to a more specific consideration of disparate treatment in the insurance context, and, then, to an application of the insurance safe harbor provision, in Parts II(A)(2) and II(B) respectively.

We begin our analysis with the wording of § 12112(a), the general anti-discrimination provision of Title I: "No covered entity shall discriminate against a qualified individual with a disability because of the disability of such individual in regard to job application procedures, the hiring, advancement, or discharge of employees, employee compensation, job training, and other terms, conditions, and privileges of employment." As the Second Circuit has observed, the language is "capacious." EEOC v. Staten Island Savings Bank, 207 F.3d 144, 149 (2d Cir. 2000). Section 12112(a) appears to mirror § 703(a) of Title VII, 42 U.S.C. § 2000e-2(a): "It shall be an unlawful employment practice for an employer . . . to discriminate against any individual . . . because of such individual's race, color, religion, sex, or national origin." Moreover, § 12112(b)--which includes in the definition of discrimination a wide array of actions that "adversely affect[] the opportunities or status of [job applicants or employees] because of . . . disability"--not only reinforces a broad reading of the rule against disability-based discrimination but specifically prohibits discrimination in fringe benefits. See 42 U.S.C. § 12112(b)(2) (prohibiting employers from "participating in a contractual . . . relationship that has the effect of subjecting a covered entity's qualified . . . employee with a disability to the discrimination prohibited by this subchapter (such relationship includes a relationship with an employment or referral agency, labor union, an organization providing fringe benefits to an employee of the covered entity, or an organization providing training and apprenticeship programs)") (emphasis added).

The EEOC, as amicus curiae, contends that the District Court erred when it dismissed appellant's claim, arguing that K Mart's LTD plan is discriminatory because "it precludes disabled individuals with mental illnesses from obtaining benefits that are available to all other disabled individuals." In support of its reading of the statute, the EEOC argues that the Supreme Court in Olmstead v. L.C., 527 U.S. 581 (1999), on review of our decision, 138 F.3d 893, made clear that the concept of discrimination embodied in the ADA encompasses differential treatment of one disabled individual as compared to another disabled individual.10 Countering this argument, appellee K Mart asserts that its LTD plan is not discriminatory because the inequity created by the plan exists between people with physical disabilities and people with mental disabilities. According to K Mart, Congress intended only to bar unequal treatment of those who are disabled as compared with those who are not disabled. For such a reading of the ADA, K Mart contends that Olmstead--which arose under Title II of the ADA, rather than under Title I, the portion of the statute that frames the case at bar--is inapposite and that the pertinent precedents are Traynor v. Turnage, 485 U.S. 535 (1988),11 and Alexander v. Choate, 469 U.S. 287 (1985),12 both of which arose under the Rehabilitation Act. As a corollary to this argument, K Mart avers that its LTD plan is not discriminatory because all its employees are equally eligible for coverage under the plan.

Notwithstanding that Olmstead arose under Title II, we find that it controls our understanding of the concept of discrimination embodied in Title I of the ADA. Both our decision in Olmstead and the subsequent Supreme Court decision affirming our decision "in substantial part,"13 stand for the proposition that the ADA demands more than impartial treatment of the disabled as compared to the non-disabled. The gravamen of a disability-based discrimination claim is that an individual has been treated less favorably because of her disability. On this ground, the Supreme Court, speaking through Justice Ginsburg, declined to adopt the narrow concept of discrimination urged in defense of the state's action:

The State argues that L.C. and E.W. encountered no discrimination "by reason of" their disabilities because they were not denied community placement on account of those disabilities. Nor were they subjected to "discrimination," the State contends, because "'discrimination' necessarily requires uneven treatment of similarly situated individuals," and L.C. and E.W. had identified no comparison class, i.e., no similarly situated individuals given preferential treatment. We are satisfied that Congress had a more comprehensive view of the concept of discrimination advanced in the ADA.

Justice Thomas, joined by the Chief Justice and Justice Scalia, dissented in Olmstead. But the rationale of the dissent adds further support for the conclusion that Title I of the ADA--like Title II, as construed by the Olmstead majority--prohibits discrimination among the protected class of disabled persons. In challenging the majority's construction of Title II, the dissent argued that the "traditional" concept of discrimination does not encompass "disparate treatment among members of the same protected class." Olmstead, 527 U.S. at 616. The dissent invoked Alexander v. Choate and Traynor v. Turnage in support of its view that Title II should be read in the light of this "traditional" understanding. However, the dissent acknowledged that Title I may be violated by differential treatment of persons within the protected category. Justice Thomas reasoned that Congress advertently adopted a non-traditional concept of "discrimination" in Title I:

Elsewhere in the ADA, Congress chose to alter the traditional definition of discrimination. Title I of the ADA, § 12112(b)(1), defines discrimination to include "limiting, segregating, or classifying a job applicant or employee in a way that adversely affects the opportunities or status of such applicant or employee." . . . The majority's definition of discrimination--although not specifically delineated--substantially imports the definition of Title I into Title II by necessarily assuming that it is sufficient to focus exclusively on members of one particular group. Under this view, discrimination occurs when some members of a protected group are treated differently from other members of that same group.

In sum, the Supreme Court's construction of the ADA in Olmstead leads us to the conclusion that K Mart's LTD plan--which differentiates between individuals who are totally disabled due to a mental disability and individuals who are totally disabled due to a physical disability because of the given individual's type of disability--appears prima facie to distinguish among beneficiaries on a basis that constitutes a form of discrimination contravening Title I of the ADA.16

(2) The Legislative History of the ADA

The next step in our analysis is to address the question whether--as some circuits have concluded--the ADA's legislative history should be read as insulating from liability what otherwise would seem to be discrimination.17 The legislative history relevant to our discussion consists of three quite similar Reports issued in 1989 and 1990--one from the Senate Labor and Human Resources Committee, one from the House Judiciary Committee, and another from the House Education and Labor Committee. The pertinent portions of these reports articulated the position that it is permissible for employer-provided health insurance plans to limit coverage of certain "procedures or treatments" even if doing so would have a disproportionate impact on people with disabilities, as long as all employees have equal access to the health insurance plan in question. The Report of the Senate Labor and Human Resources Committee put the matter this way:

[E]mployers may not deny health insurance coverage completely to an individual based on the person's diagnosis or disability. For example, while it is permissible for an employer to offer insurance policies that limit coverage for certain procedures or treatments, e.g., only for a specified amount per year for mental health coverage, a person who has a mental health condition may not be denied coverage for other conditions such as for a broken leg or for heart surgery because of the existence of the mental health condition. A limitation may be placed on reimbursements for a procedure or the types of drugs or procedures covered[,] e.g., a limit on the number of x-rays or non-coverage of experimental drugs or procedures; but, that limitation must apply to persons with or without disabilities. All people with disabilities must have equal access to the health insurance coverage that is provided by the employer to all employees.

The Committee Reports suggest no intention to interfere with insurance arrangements which set caps on "procedures or treatments" that apply to persons with or without disabilities. But we are here dealing not with a limitation on procedures and treatments equally available to all but a limitation on compensation in lieu of salary which is expressly contingent on what kind of disability has caused a former employee to lose his job. Giving the legislative reports their full weight, the type of differentiation that they protect is not the type of differentiation at issue in this case. The statement "it is permissible for an employer to offer insurance policies that limit coverage for certain procedures or treatments," S. Rep. No. 101-116, at 29, is disability-neutral. It is reasonable to assume that the large majority of employees or former employees receiving employer-provided "procedures or treatments"--whether x-rays, or blood transfusions, or physical therapy, or mental health therapy--are neither physically nor mentally disabled. In contrast, there is no disability-neutral parallel that reflects the distinction at work in K Mart's LTD plan. Instead of limiting insurance coverage for a particular type of procedure or treatment, the K Mart plan, on its face, limits disability-benefits-in-lieu-of-salary for those with a specified type of disability. The parties in the case at bar are concerned with a single benefit--payments intended to partially replace the salary that disabled individuals can no longer earn--rather than coverage for one of a myriad potential medical procedures or treatments.19 Denial of that benefit on the express ground that the claimant is mentally disabled is discrimination of a sort prohibited by § 12112(a)--unless the ADA's safe harbor provision exempts such discrimination from liability. To the impact of the safe harbor provision we now turn.

(B) The Safe Harbor Provision

To this point in the analysis, our focus has been on ascertaining the meaning and scope of the general prohibition against disability-based discrimination contained in § 12112(a). We now turn to the construction of § 12201(c), which provides a safe harbor exempting an employer from liability for a bona fide benefit plan, provided the safe harbor is not used as a "subterfuge to evade the purposes of" the ADA. Section 12201(c) reads, in pertinent part:

Subchapters I through III of this chapter and title IV of this Act shall not be construed to prohibit or restrict . . .

(3) a person or organization covered by this chapter from establishing, sponsoring, observing or administering the terms of a bona fide benefit plan that is not subject to State laws that regulate insurance.

Paragraphs (1), (2), and (3) shall not be used as a subterfuge to evade the purposes of subchapters I and III of this chapter.

In the case at bar, the disputed issue regarding the safe harbor provision is whether K Mart's adoption of the mental health cap constitutes a use of the safe harbor provision as "a subterfuge to evade the purposes" of Title I. Because we are reviewing the District Court's grant of a motion to dismiss, we must determine whether we can decide this issue as a matter of law on the basis of the allegations contained in the complaint.

Appellant and the EEOC contend that the safe harbor provision, § 12201(c), is used as a "subterfuge" when a disability-based distinction in a benefit plan is not justified by increased costs demonstrably attributable to the disability.20 Therefore, the EEOC argues that this case must be remanded to the District Court for a hearing at which K Mart would have the burden of demonstrating that, taking into account the totality of the circumstances, the mental health cap is "justified by the risks or costs associated with" mental health disabilities. Countering this argument, K Mart relies on Public Employees Retirement System of Ohio v. Betts, 492 U.S. 158 (1989), as support for the assertion that the burden is on appellant to prove that K Mart used the safe harbor provision as a "subterfuge" when implementing the mental health cap by showing that K Mart specifically intended to discriminate on the basis of disability, see Betts, 492 U.S. at 171, and further, that the discrimination was designed to affect the non-fringe-benefit aspects of the employment relationship. See id. at 177. Along these lines, K Mart avers that it is exempt from liability as a matter of law with respect to its LTD plan because plaintiff failed to allege that K Mart used the LTD plan to discriminate in any non-fringe-benefit aspect of its employment relationship with Johnson. The common thread between these two arguments is that both sides discuss Betts, though to different effect.

In Betts, the Supreme Court construed the meaning of the word "subterfuge" as it appeared in an analogous provision of the ADEA. See also United Air Lines, Inc. v. McMann, 434 U.S. 192 (1977). At the time Betts was decided, § 4(f)(2) of the ADEA, 29 U.S.C. § 623(f)(2), provided that:

It shall not be unlawful for an employer, employment agency, or labor organization . . .

(2) to observe the terms of a bona fide seniority system or any bona fide employee benefit plan such as a retirement, pension, or insurance plan, which is not a subterfuge to evade the purposes of this chapter, except that no such employee benefit plan shall excuse the failure to hire any individual, and no such seniority system or employee benefit plan shall require or permit the involuntary retirement of any individual specified by section 631(a) of this title because of the age of such individual[.]

29 U.S.C. § 623(f)(2) (1988) (emphasis added).21 In Betts and McMann, the Court's construction of § 4(f)(2) was grounded, in part, on a determination of the ordinary meaning of the word "subterfuge": "[W]e stated in McMann that 'subterfuge' means 'a scheme, plan, stratagem, or artifice of evasion,' which in the context of § 4(f)(2), connotes a specific 'intent . . . to evade a statutory requirement.' The term thus includes a subjective element that the regulation's objective cost-justification requirement fails to acknowledge." Betts, 492 U.S. at 171 (quoting McMann, 434 U.S. at 203). As a result of the Court's interpretation that a subterfuge necessarily entails a specific intent to evade the purposes of the statute in question, the McMann Court held, and the Betts Court reaffirmed, that no benefit plan instituted prior to the passage of the ADEA could be considered a subterfuge. See id. at 168.

The EEOC and appellant argue that the reasoning in Betts is not transferable to the ADA because Congress wrote the subterfuge exception in § 12201(c) in such a way as to reject the Betts construction of the phrase "subterfuge to evade the purposes" of the Act. In this regard, appellant and amicus rely on both statutory language and legislative history. When drafting § 12201(c), Congress adopted a different phrasing of the "subterfuge" exception to the safe harbor than it had in the former § 4(f)(2) of the ADEA. The ADA provides that the safe harbor provision "shall not be used as a subterfuge to evade the purposes of" the Act, whereas the ADEA provided that the safe harbor provision only exempted a plan "which is not a subterfuge to evade the purposes of this chapter" of the ADEA. Appellant and the EEOC assert that the change in focus from whether a policy is a subterfuge, to whether the safe harbor is used as a subterfuge, demonstrates that an employer may be understood to use the safe harbor as a subterfuge when it continues to implement pre-Act terms of a plan. Appellant and the EEOC also point to contemporaneous legislative history that lends support to such a construction. See H.R. Rep. 101-485(II), 1990 U.S.C.C.A.N. 303, 419 ("[T]he decision to include this section may not be used to evade the protections of [the Act], regardless of the date an insurance plan or employer benefit plan was adopted."). Additionally, appellant and the EEOC rely on the legislative history for the proposition that the safe harbor provision is "used as a subterfuge" when a plan contains a disability-based distinction "except where the refusal, limitation, or rate differential is based on sound actuarial principles or is related to actual or reasonably anticipated experience." Id. at 420.

We find the difference in the wording of § 12201(c) of the ADA as compared to the former § 4(f)(2) of the ADEA to be insufficient to overcome the Supreme Court's clear statement in Betts that the phrase "subterfuge to evade the purposes of" the Act connotes that the employer must have had a specific intent to evade the purposes of the statute.22 Additionally, the reliance by the EEOC and appellant on the legislative history is contrary to the reasoning in Betts wherein the Court expressly rejected a reliance on legislative history with regard to whether or not the phrase "subterfuge to evade" mandated a finding of specific intent on the ground that the ordinary meaning of the phrase unambiguously required a showing of intent. See Betts, 492 U.S. at 172. We, therefore, agree with a number of our sister circuits that Betts requires us to read the exception to the ADA safe harbor provision to include a showing of specific intent to discriminate. See EEOC v. Aramark, 208 F.3d at 271; Fitts v. Federal National Mortgage Association, 236 F.3d 1 (D.C. Cir. 2001) (following Aramark); Ford, 145 F.3d at 611; Leonard F. v. Israel Discount Bank of New York, 199 F.3d 99, 105 (2d Cir. 1999) (Title III of the ADA).

We conclude that McMann and Betts make clear that if K Mart adopted the policy of providing differing long-term disability benefits for mental and physical disabilities prior to the enactment of the ADA, K Mart could not be found to be using § 12201(c) as a subterfuge to evade the purposes of the Act. If, however, K Mart adopted the mental health cap following the passage of the ADA, then appellant, in order to impose liability on K Mart, would have to establish that K Mart, in setting a 24-month cap on mental health disability benefits, specifically intended to use § 12201(c) as a subterfuge to evade the purposes of Title I.

Whereas K Mart correctly states that the Betts Court held that a benefit plan is not a subterfuge under the ADEA's safe harbor provision unless the plaintiff can show that the employer had a specific intent to interfere with non-fringe-benefit aspects of employment, that limitation on the categories of benefits addressed by the ADEA is not controlling here. While the specific intent requirement in Betts flows from the ordinary meaning of the phrase "subterfuge to evade the purposes [of the Act]," the requirement in Betts that discrimination affect a non-fringe benefit relates to elements of the ADEA's statutory context that are not shared with the ADA. Specifically, the Court held in Betts that, in order to be a subterfuge, a plan must discriminate in a non-fringe-benefit aspect of employment because the Court found that fringe benefit plans were not included in the ADEA's general prohibition against age discrimination, codified at 29 U.S.C. § 623(a)(1). See Betts, 492 U.S. at 177 ("[U]nder this construction of the statute, Congress left the employee benefit battle for another day, and legislated only as to hiring and firing, wages and salaries, and other non-fringe-benefit terms and conditions of employment."). Thus, the ADEA's focus on non-fringe-benefit aspects of employment does not square with the ADA's express prohibition against disability-based discrimination in fringe benefits plans contained in § 12112(b)(2) of Title I. See Gonzales, 89 F.3d at 1526 ("It is . . . undisputed that fringe benefits . . . are one set of the 'terms, conditions, and privileges of employment' protected from unlawful discrimination under the ADA."). Therefore, we conclude that in the context of the ADA the subterfuge exception to the safe harbor provision requires that a plaintiff show that the employer specifically intended to discriminate based on disability, whether the discrimination was aimed at fringe-benefit or non-fringe-benefit aspects of the employment relationship.23

For the reasons stated, we conclude that the District Court erred in granting K Mart's motion to dismiss. Accordingly, the judgment of the District Court is reversed and the cause remanded for further proceedings consistent with this opinion.

REVERSED.

I concur in Judge Pollak's comprehensive opinion and only write to explain why I believe Robinson v. Shell Oil Company, 519 U.S. 337 (1997) clearly abrogates Gonzales v. Garner Food Services, Incorporated, 89 F.3d 1523 (11th Cir. 1996).

A basic proposition in federal law is that an intervening decision of the Supreme Court subjects a prior decision of this Court to reconsideration. See Chambers v. Thompson, 150 F.3d 1324, 1326 (11th Cir. 1997) ("We are bound to follow a prior panel or en banc holding, except where that holding has been overruled or undermined to the point of abrogation by a subsequent en banc or Supreme Court decision.") (emphasis supplied); Davis v. Singletary, 119 F.3d 1471, 1482 (11th Cir. 1997) ("[t]o the extent of any inconsistency between our [prior opinions'] pronouncements and the Supreme Court's supervening ones, of course, we are required to heed those of the Supreme Court."); United States v. Hogan, 986 F.2d 1364, 1369 (11th Cir.1993) (recognizing same); Lufkin v. McCallum, 956 F.2d 1104, 1107 (11th Cir. 1992) (recognizing same).

On review of Gonzales and Robinson, it is clear that there is more than a minor "inconsistency" between the two cases. As Judge Pollak explains, Robinson explicitly invalidates several of the propositions that informed the Gonzales Court's conclusion that Title I of the Americans with Disabilities Act ("ADA"), 42 U.S.C. § 12101 et. seq., only covers former employees. In both Gonzales and Robinson, the basic question presented was whether the term "employee" in an employment discrimination statute only refers to current employees or includes former employees.1 In both cases, the employer argued that the plain definition of the term "employee" in an employment discrimination statute unambiguously refers to current employees. Robinson, 518 U.S. at 339; Gonzales, 89 F.3d at 1526. The Court in Gonzales agreed, holding that the term "employee" in Title I was plain and unambiguous, and only referred to current employees. Gonzales, 89 F.3d at 1526 (interpreting 42 U.S.C. § 12111(4)). However, the Supreme Court in Robinson rejected the notion that the term "employee," standing alone, is unambiguous, explaining that the term "employee" "lacks any temporal qualifier and is consistent with either current or past employment." Robinson, 519 U.S. at 342 (discussing 42 U.S.C. § 2000e -3(a)). The Supreme Court indicated that courts should not draw unnecessary meaning from the fact that the term "employee" is generally defined in an employment discrimination statute as a "person employed by an employer," explaining that the word "employed" in this context "is not . . . limited in its possible meanings, and could just as easily be read to mean 'was employed'" by an employer. Id. at 342 (emphasis in original).

The Supreme Court applied the same logic to interpret the phrase "applicants or employees" as it relates to the prohibition against discrimination in the employment context. Robinson, 519 U.S. at 344 (interpreting 42 U.S.C.§ 2000e(3)(a)). The Robinson Court explained that "[t]he use of the term 'applicants' [in an employment discrimination statute] does not serve to confine, by negative inference, the temporal scope of the term 'employees.' . . . [ as this] argument rests on the incorrect premise that the term 'applicants' is equivalent to the phrase 'future employees.'" Id. at 344-45. The Robinson Court further described the analytic flaw that gives rise to this view, explaining:

the term "applicants" would seem to cover many persons who will not become employees. Unsuccessful applicants or those who turn down a job offer, for example, would have been applicants, but not future employees. And the term fails to cover certain future employees who may be offered and will accept jobs without having to apply for those jobs. Because the term "applicants" in § 704(a) is not synonymous with the phrase "future employees," there is no basis for engaging in the further (and questionable) negative inference that inclusion of the term 'applicants' demonstrates intentional exclusion of former employees.

This ruling also severely undercuts the Gonzales Court's interpretation of Title I, as the Gonzales Court concluded that the plain meaning of the term "applicants and employees" established that Title I of the ADA only covered current and future employees. See Gonzales, 89 F.3d at 1527 (interpreting 42 U.S.C. § 12112(b)).

As shown above, the Robinson Court, analyzed the same terms we examined in Gonzales and found these terms to be ambiguous. The only remaining basis for the Gonzales Court's ruling was its view that the phrase "qualified individual with a disability" in 42 U.S.C. § 12111(8) could only be interpreted to mean current employees, because the provision refers to persons who "hold[] or desire[]" a position with an employer. Gonzales, 89 F.3d at 1526. However this proposition is also undercut by the Robinson case, as the Robinson Court established that courts may not interpret a provision in an employment discrimination statute to exclude former employees unless they can identify language in the provision that functions as a clear and explicit "temporal qualifier." Robinson, 518 U.S. at 343. The Supreme Court explained that statutory language serves as a "temporal qualifier" when it "specif[ies] the time frame in which the employment relationship [at issue] must exist." Id. at 342, n.2. The Robinson Court identified two kinds of language that meet this requirement. It explained that some statutory provisions will prohibit behavior or offer remedies which, by definition, only pertain to a certain class of employees. See id. at 343 (explaining that 42 U.S.C. § 2000e-16(b), which prohibits discrimination in advancement, and 42 U.S.C. § 2000e-2(h), which prohibits discrimination in salary payments or promotions, both, by definition, only cover current employees). Also, the Robinson Court noted that some provisions will specifically delineate the time frame to which they apply, and thereby limit the scope of their coverage. Id. at 343, n.2 (discussing 42 U.S.C. § 2000e(b), which provides that Title VII regulates any employer "who has fifteen or more employees for each working day in each of twenty or more calendar weeks in the current or preceding calendar year").

When viewed under this standard, it is clear that the phrase "holds or desires" in § 12111(8) is not a "temporal qualifier." First, the phrase does not prohibit any time-specific employer action or provide for time-specific remedies that only pertain to a certain class of employees. Second, the phrase does not contain any language that explicitly refers to a particular time frame. Certainly, the "holds and desires" language in § 12111(8) could support the inference that Congress intended for Title I to cover only applicants (those who "desire" positions) and current employees (those who "hold" positions). However, this language could also be interpreted to include former employees who still "desire" to work for their respective employers, but whose disabilities bar them from doing so unless they are provided with the appropriate accommodation. The basic fact of the matter is, the phrase "holds or desires" is time-neutral and ambiguous, and therefore § 12111(8) does not bar former employees from invoking Title I's protection.

Having concluded that the aforementioned Title I provisions do not clearly and unambiguously define the class of employees protected under the statute, our remaining task is to determine whether Title I provides a consistent definition for covered employees in its remaining provisions and, for this reason, has a clear and unambiguous meaning. See Robinson, 519 U.S. at 341 ("[T]he plainness or ambiguity of statutory language is determined by reference to the language itself, the specific context in which the language is used, and the broader context of the statute as a whole.")

Using this framework, it is clear that other provisions in Title I rebut the inference that the statute only protects current and future employees. Specifically, 42 U.S.C. § 12112(a), which prohibits discriminatory discharge, and 42 U.S.C. § 12117(a), which provides for the remedy of reinstatement, by definition pertain to former employees. Also, Title I's general definition of discrimination, 42 U.S.C. § 12112(b), suggests that Title I protects former employees because this provision prohibits employers from administering programs that discriminate against the disabled in providing "fringe benefits," and when interpreting other employment discrimination statutes courts interpret the term "fringe benefits" to include retirement and other post-employment medical benefits. See, e.g., Dobbs v. City of Atlanta Georgia, 606 F.2d 557 (5th Cir 1979) (Title VII suit alleging that an employer's racially discriminatory personnel practices prevented Black employees from accessing high quality pension plans). The aforementioned provisions make it clear that Congress designed Title I so that, in various sections, it offers protection to different groups of employees, in some cases offering protection and remedies to current employees, and in others extending its protection to former employees, and potential future employees ("applicants"). Therefore, because the statute does not provide a "coherent and consistent" definition of covered employees across its various sections, the Gonzales Court erred in deciding that Title I's "plain meaning" prevented former employees from invoking its protections.

In short, there is no doubt that the Robinson Court's decision invalidates Gonzales. The terms the Gonzales Court concluded had a "plain and ordinary meaning" instead were found to be ambiguous by the Supreme Court. Moreover, Robinson stands for the proposition that the Court's decision to deny an employment discrimination statute's protection to a class of employees should turn on language that serves as an explicit temporal qualifier. Therefore, Robinson's rationale and holding abrogate the rationale and holding of Gonzales. Under our precedent, such a conclusion requires that we follow the Supreme Court's intervening law.

Certainly, it is easier to conclude that a Supreme Court decision invalidates an Eleventh Circuit decision when the Supreme Court expressly overrules a particular case, or baldly criticizes the precise proposition that is the basis for an Eleventh Circuit panel's ruling. See, e.g., Lufkin, 956 F.2d at 1107 (recognizing that a Supreme Court case had invalidated the Eleventh Circuit's retroactivity rule and overruling contrary Eleventh Circuit authority "in order to give full effect to [an] intervening decision of the Supreme Court"); Cotrell v. Caldwell, 85 F.3d 1480, 1485 (11th Cir. 1996) ( recognizing that a Supreme Court decision had invalidated the line of Eleventh Circuit cases holding that officers could not appeal interlocutory orders denying them qualified immunity when the courts involved had ruled that material issues of fact were in dispute). However, we are also compelled to overrule our precedent when the rationale the Supreme Court uses in a intervening case directly contradicts the analysis that this Court has used in a related area, and establishes that this Court's current rule is wrong. See, e.g., Footman v. Singletary, 978 F.2d 1207, 1210-11 (11th Cir. 1992) (overruling Eleventh Circuit cases that allowed habeas petitioners to raise additional ineffective assistance of counsel claims not raised in state court because the policy rationale in Keeney v. Tamayo-Reyes, 504 U.S. 1 (1992), a decision on evidentiary hearings in habeas cases, indicated that the Supreme Court disfavors allowing petitioners to raise new federal habeas claims not raised in state court); Hogan, 986 F.2d at 1370 (11th Cir. 1993) (overruling Eleventh Circuit precedent applying a more liberal standard of the clear error test to review a trial court's determination that the defendant was competent to stand trial because, Demosthenes v. Baal, 495 U.S. 731 (1990), a Supreme Court case discussing a defendant's competence to waive habeas proceedings, suggested that a more stringent standard was required).

Judge Carnes cites a number of cases to support his view that, absent an express statement from the Supreme Court that Gonzales is overruled, the case is still good law. However no case cited, with the exception of Morris v. City of West Palm Beach 194 F.3d 1203 (11th Cir. 1999), involved an intervening Supreme Court case which undercut the basis for an earlier panel's ruling. For example, in United States v. Chubbuck, 252 F.3d 1300 (11th Cir. 2001), this Court declined to overrule an Eleventh Circuit panel's interpretation of the term "conviction" in light of an intervening Florida Supreme Court decision. In Lapides v. Board of Regents, 251 F.3d 1372 (11th Cir. 2001), this Court declined to overrule an Eleventh Circuit decision based on arguably conflicting Eleventh Circuit authority and a Supreme Court concurring opinion. Also, in United States v. Smith 122 F.3d 1355 (11th Cir. 1997), this Court refused to change its interpretation of the federal expert witness rules because its ruling was consistent with Supreme Court precedent, and because it did not find persuasive the conflicting authority the petitioner offered from other Courts of Appeal.

Saxton v. SCF Industries, 239 F.2d 1209 (11th Cir. 2001) is another case in which no conflicting Supreme Court decision was at issue; this Court merely recognized that a prior panel had failed to acknowledge the existence of the 1991 Amendments to Rule 15(c), and it applied the prior panel's rule and recommended en banc review. Similarly, in Turner v. Beneficial Corporation, 236 F.3d 1023 (11th Cir. 2001), a panel of this Court recognized that a prior panel had misinterpreted the requirements of the Truth in Lending Act, but the panel faithfully applied the questionable rule and recommended en banc review. See also United States v. Steele, 117 F.3d 1231 (11th Cir. 1997) (recognizing that a prior panel apparently had misunderstood the requirements for indicting a physician under 28 U.S.C. § 855(a)(1) but applying the prior panel's rule and recommending en banc review).

The only case Judge Carnes discusses that considered the effect of an intervening, related Supreme Court decision on an earlier Eleventh Circuit decision is Morris v. City of West Palm Beach 194 F.3d at 1207 n.6.2 In Morris, this Court recognized that Farrar v. Hobby, 506 U.S. 103 (1992), cast doubt on the viability of the "catalyst test," a court-created modification of the attorneys' fee provisions under 42 U.S.C. § 1988. Although § 1988 only authorizes attorneys' fees for prevailing parties, several courts had used the "catalyst test" to award plaintiffs attorneys' fees in civil right cases despite the fact that they did not prevail at trial, when it was clear that the plaintiffs' lawsuits had pressured the defendants into functionally providing the relief that was the subject of the plaintiffs' claims. The rationale behind the catalyst test is that § 1988 was designed to compensate any person who acts as a public attorney general and ensures the enforcement of the civil rights laws, and plaintiffs in catalyst test cases technically have fulfilled this role. After the development of this test, the Supreme Court issued its decision in Farrar, which re-established that in order "to qualify as a prevailing party [under § 1988], a civil rights plaintiff must obtain at least some relief on the merits of his claim." Farrar, 506 U.S. at 108. Stated simply, the Supreme Court explained that the recipient of an attorneys fee award under § 1988 "must obtain an enforceable judgment against the defendant from whom fees are sought or comparable relief through a consent decree or settlement." Id. After Farrar, some courts concluded that the catalyst test was no longer viable. In Morris, however, we rejected this view. This Court held that the catalyst test survived Farrar because the Farrar Court did not directly challenge the catalyst test and, given the strong policy considerations supporting a broad interpretation of § 1988-namely ensuring the enforcement of the civil rights laws-the test should survive.

While Morris bears comparison to the instant case, I believe it is distinguishable. The Robinson decision directly challenges the statutory analysis we performed in Gonzales, as it overrules our decision regarding the threshold determination of whether the terms in the ADA were ambiguous. Additionally, the policy concerns that inform the ADA counsel that we must give the statute a broad cast, and therefore suggest that we should revisit the Gonzales Court's more restrictive interpretation of the statute. See 42 U.S.C. § 12101(a)(8) (explaining that the ADA's goal is to ensure that the disabled enjoy the same "equality of opportunity, full participation, independent living, and economic self sufficiency" as the non-disabled).

In summary, because Robinson clearly invalidates our analysis in Gonzales I believe that it requires that we revisit our interpretation of Title I of the ADA. As for the substantive question of whether the ADA only protects current employees, I join in Judge Pollak's cogent discussion explaining the statutory basis for interpreting the ADA to cover current as well as former employees, as well as his analysis of the statute's safe harbor provision.

The threshold question in this case is whether a former employee is eligible to file suit against his erstwhile employer under the Americans with Disabilities Act of 1990, 42 U.S.C. § 12112(a), or is that right and remedy confined to current employees or applicants. As the majority concedes, a prior decision of this Court, Gonzales v. Garner Food Services, Inc., 89 F.3d 1523 (11th Cir. 1996), cert. denied, 520 U.S. 1229, 117 S. Ct. 1822 (1997), squarely answers that question with the holding that a former employee may not sue under the ADA, any remedy under the Act being confined to current employees or applicants. The majority does not pretend that the Gonzales decision, which is directly on point, has ever been overruled by the Supreme Court, by this Court sitting en banc, or by Congress. Yet the majority refuses to follow Gonzales.

The only reason the majority gives for not following the law of the circuit as established in Gonzales, is the Supreme Court's later decision in Robinson v. Shell Oil Co., 519 U.S. 337, 117 S. Ct. 843 (1997). The Robinson case, however, arose and was decided under Title VII, not the ADA, and the two statutes have different language and different legislative histories. The Supreme Court in Robinson did not purport to overrule Gonzales or to establish any law about the ADA. And the majority does not contend that Robinson addressed any ADA issues or that it purported to overrule Gonzales. Because I think that the majority's action violates the prior panel precedent rule, which requires that we follow the Gonzales decision unless and until it is overruled by the Supreme Court or this Court sitting en banc, I dissent.

The prior panel precedent rule is an essential principle, the number one ground rule, by which all of us on this Court must abide. It helps keep the peace, stabilizes our circuit law, and promotes efficiency by enabling us to move on once issues have been decided by a panel. See generally Bonner v. City of Prichard, 661 F.2d 1206, 1209-10 (11th Cir. 1981) (en banc) (deciding to have Eleventh Circuit panels bound by all pre-split Fifth Circuit decisions, because "[s]tability and predictability are essential factors in the proper operation of the rule of law"); Jaffree v. Wallace, 705 F.2d 1526, 1533 (11th Cir. 1983), aff'd, 472 U.S. 38, 105 S. Ct. 2479 (1985) ("Judicial precedence serves as the foundation of our federal judicial system. Adherence to it results in stability and predictability."). Those important goals are achieved because, and only to the extent that, the prior precedent rule requires a later panel to follow a prior one's decision even though convinced it is wrong. See generally Cohen v. Office Depot, Inc., 204 F.3d 1069, 1076 (11th Cir. 2000), cert. denied, 531 U.S. 957, 121 S. Ct. 381 (2000) ("[T]he prior panel precedent rule is not dependent upon a subsequent panel's appraisal of the initial decision's correctness."); United States v. Steele, 147 F.3d 1316, 1317 - 18 (11th Cir. 1998) (en banc) ("Under our prior precedent rule, a panel cannot overrule a prior one's holding even though convinced it is wrong.") (quoting Cargill v. Turpin, 120 F.3d 1366, 1386 (11th Cir. 1997)).

Our allegiance to the prior precedent rule, as well as the proper corrective mechanism for purging our circuit law of earlier panel errors, is illustrated in two cases that resulted in en banc decisions earlier this year. One is Saxton v. ACF Industries, Inc., 239 F.3d 1209 (11th Cir. 2001), rev'd en banc, 254 F.3d 959 (11th Cir. 2001). The panel in that case dutifully followed an earlier one's holding even though convinced it was wrong. See 239 F.3d at 1215. Then rehearing en banc was promptly granted and, in a decision written by the author of the second panel's decision, 254 F.3d at 959, the en banc court overruled the first panel's erroneous, id. at 963 - 66. The same thing happened in Turner v. Beneficial Corp., 236 F.3d 643 (11th Cir. 2000), rev'd en banc, 242 F.3d 1023 (11th Cir. 2001), cert. denied, 70 USLW 3234 (Oct. 1, 2001) (No. 00-1799). The panel in that case also followed a prior panel's decision even though convinced the earlier panel had overlooked controlling statutory language that dictated a different result. 236 F.3d at 649 - 50. Rehearing en banc was granted, 236 F.3d 651, and the en banc court overruled the erroneous prior precedent that had bound the later panel, 242 F.3d 1023, all within the space of two months. As in Saxton, the en banc opinion in Turner was written for the Court by the author of the second panel opinion. See also United States v. Steele, 117 F.3d 1231, 1234 - 35 (11th Cir. 1997) (reluctantly following a prior panel decision even though it had overlooked a controlling statute), rev'd en banc, 147 F.3d 1316 (11th Cir. 1998) (overruling the erroneous prior decision that had bound the panel), cert. denied, 528 U.S. 933, 120 S. Ct. 335 (1999); Smith v. GTE Corp., 236 F.3d 1292, 1303 n.10 (11th Cir. 2001) ("The erroneous result reached by the prior panel whose decision bound the Steele panel was corrected en banc, which is how erroneous panel decisions should be corrected.") (internal citation omitted). If the Gonzales panel erred in holding that the ADA does not apply to former employees - and it is not at all clear to me that its holding is error - the proper way to change circuit law on the issue is to use this case as a vehicle for overruling Gonzales en banc, which is how we corrected erroneous circuit law in Saxton, Turner, and Steele.

Of course, prior panel decisions can be overruled by intervening Supreme Court decisions, as well as by the en banc court, and where there is a conflict between the holding of an earlier panel decision and that of a later Supreme Court decision, subsequent panels must follow the Supreme Court decision. See In re Provenzano, 215 F.3d 1233, 1235 (11th Cir. 2000), cert. denied, 530 U.S. 1256, 120 S. Ct. 2710 (2000) ("We would, of course, not only be authorized but also required to depart from [the prior decision] if an intervening Supreme Court decision actually overruled or conflicted with it."); Cottrell v. Caldwell, 85 F.3d 1480, 1485 (11th Cir. 1996) ("Where prior panel precedent conflicts with a subsequent Supreme Court decision, we follow the Supreme Court decision."). But a panel is justified in disregarding a prior panel decision because of the Supreme Court's holding in a later case only when that intervening holding is squarely on point.

As we have held on numerous occasions, a prior panel decision that has been weakened but not overruled by a later Supreme Court decision must be followed by later panels. See United States v. Chubbuck, 252 F.3d 1300, 1305 n.7 (11th Cir. 2001) ("We are not at liberty to disregard binding case law that is so closely on point and has been only weakened, rather than directly overruled, by the Supreme Court.") (internal quotes and citation omitted); Lapides v. Bd. of Regents, 251 F.3d 1372, 1377 n. 3 (11th Cir. 2001) (same); Morris v. City of West Palm Beach, 194 F.3d 1203, 1207 n.6 (11th Cir. 1999) (same); United States v. Smith, 122 F.3d 1355, 1359 (11th Cir. 1997) (same), cert. denied, 522 U.S. 1021, 118 S. Ct. 614 (1997); see also N.L.R.B. v. Datapoint Corp., 642 F.2d 123, 129 (5th Cir. Unit A April 1981) ("Without a clearly contrary opinion of the Supreme Court or of this Court sitting en banc, we cannot overrule a decision of a prior panel of this Court...."). So, we have said in at least four decisions that a later panel must follow an earlier one's holding that has been weakened but not overruled by an intervening Supreme Court decision. Seven judges of this Court either wrote or joined the opinions in one or more of those four decisions: Judges Edmondson, Birch, Black, Wilson, Fay, Cox, and me. I think that we meant what we said.

The temptation can be great for a judge to read into an intervening Supreme Court decision more than is there in order to justify not following a particularly pesky prior panel precedent. To be tempted in that respect is just human, but we should resist the temptation. It is critical that when sitting as panel judges we be diligent in policing the line between prior precedent that has been contradicted and thereby overruled by intervening Supreme Court precedent, and that which has been only weakened by it. The strength and integrity of our prior precedent rule, upon which so much rests, is dependent upon that distinction.

The most that can be justifiably said about the effect of Robinson on the holding in Gonzales is that some of what is said, but not anything that is held, in the Robinson opinion arguably weakens some of the reasoning in the Gonzales opinion. That, however, is not enough for this panel to toss the Gonzales decision aside. The majority says that "legal precedent provides our starting point," Majority Op. at 1044, but when prior panel precedent is directly on point, as the Gonzales decision is, it also ought to be the ending point for subsequent panels until that controlling precedent has been overruled. Instead of ending with the controlling precedent of Gonzales, the majority, in its own words, proceeds to "assess the continuing viability of the components of the rationale in Gonzales." Majority Op. at 1044. The exception to the prior precedent rule for intervening Supreme Court decisions does not authorize this panel to assess the continuing viability of the components of the rationale that led a prior panel to reach the holding it did. See Cohen, 204 F.3d at 1076 ("[T]he prior panel precedent rule is not dependent upon a subsequent panel's appraisal of the initial decision's correctness.").

Supreme Court decisions may provide a springboard for the en banc court to reconsider settled circuit law, but they should not be viewed as an opportunity for a later panel to cast aside circuit law with which that panel disagrees. A fair reading of the majority opinion indicates that is what this panel has done. The question is not, as the majority terms it, whether the reasoning of an intervening Supreme Court decision has "undermined" the reasoning of a prior panel precedent, but whether the Supreme Court decision has overruled that prior precedent. When it comes to their effect on prior panel precedents, Supreme Court decisions either overrule those precedents, or they do not. So far as a later panel is concerned, the effect of a Supreme Court precedent is all or nothing. And here it is nothing, so far as this panel (to be distinguished from the en banc court) is concerned. The Supreme Court's Robinson decision does not explicitly overrule Gonzales or its holding. Nor does Robinson clearly contradict Gonzales to the point of implicitly overruling it.

The extent to which the majority's decision defies our prior panel precedent rule is evident from the fact that the Supreme Court in Robinson held nothing more than what this Court had already held before the Gonzales decision was released. This Court had held in Bailey v. USX Corp., 850 F.2d 1506 (11th Cir. 1988), that former employees may sue under the anti-retaliation provision of Title VII. The plaintiff in Gonzales, relying on Bailey, argued by analogy that the same should be true of former employees and the ADA. 89 F.3d at 1527. The Gonzales panel acknowledged that Title VII's anti-retaliation clause applied to former employees, as Bailey had held, but it distinguished that provision of Title VII from the ADA on the basis of the different statutory language and legislative history. Id. at 1527 - 29. What the Supreme Court did in Robinson was agree with our circuit law, as set out in Bailey, that Title VII's anti-retaliation provision does apply to former employees, something the Gonzales panel itself had acknowledged and considered in deciding that the ADA does not. It is a bit audacious for the majority to say that a Supreme Court decision whose holding was anticipated, acknowledged, and considered by a prior panel when deciding a different issue has undermined that prior panel's decision on the different issue to such an extent that it may be disregarded.

The fact that the Supreme Court's Robinson decision is not inconsistent with our Gonzales precedent is also evident from the post-Robinson decisions of two other federal court of appeals which reached exactly the same conclusion as Gonzales even in the wake of Robinson. The Court in Weyer v. Twentieth Century Fox Film Corp., 198 F.3d 1104, 1112 (9th Cir. 2000), squarely faced the issue of whether the Supreme Court's Robinson decision was inconsistent with the holding of this Court in Gonzales and of other circuits that had reached the same conclusion this circuit had in Gonzales. After contrasting the language of the ADA with that of Title VII, and fully considering the Supreme Court's Robinson decision, the Ninth Circuit, which was not bound to accord our Gonzales decision any presumption or deference, nonetheless followed Gonzales. It did so after concluding that Robinson had not undermined Gonzales. Weyer, 198 F.3d at 1108 - 13.

Likewise, in Morgan v. Joint Admin. Bd., 268 F.3d 456, (7th Cir.2001), the Seventh Circuit re-affirmed its pre-Robinson decision holding that the former employees are not protected by the employment provisions of the ADA. After pointing out that the ADA requires plaintiffs to be able to perform the essential functions of their job before they are entitled to the protection of the Act, the Court pointedly asked about whether former employees are covered: "How could they be? They cannot perform the essential functions of their job, and therefore they have no rights under the statutory provisions [of the ADA]." Id. at 458.

The plaintiffs in Morgan conceded that EEOC v. CNA Ins. Cos., 96 F.3d 1039 (7th Cir. 1996),a pre-Robinson decision of the Seventh Circuit panel, had held that the ADA does not apply to former employees, but they asked the Court to re-examine that position in light of Robinson's holding that Title VII's anti-retaliation provision does apply to former employees. Id. at 458. The Court concluded that Robinson had changed nothing about Seventh Circuit law on the ADA, reasoning that there is a "stark" difference between the anti-retaliation provision of Title VII and the employment provisions of the ADA, and that extending the scope of the ADA to former employees would "create perverse incentives" resulting in fewer employers offering disability benefits to their workers. Id.

Further, the Morgan Court observed that in CNA the panel had considered the very reasoning that would later be adopted by Robinson in a Title VII retaliation case, and had declined to apply it to the ADA. Morgan, 268 F.3d at 458. ("We anticipated the difference between [the Title VII] situation and the one here in the case the plaintiffs ask us to overrule."). Because CNA had already distinguished Robinson's Title VII-retaliation situation from that of an ADA case not involving retaliation, the Morgan Court recognized that Robinson did not govern the case before it, and followed the prior panel decision, CNA. Similarly, as noted above, our Gonzales opinion distinguished Bailey, a Title VII retaliation case, from the case before it, an ADA case. Thus Gonzales, just like CNA, anticipated Robinson's Title VII arguments and rejected their applicability to an ADA case. In Morgan, the Seventh Circuit respected its CNA decision, specifically its distinction between Title VII and the ADA. In this case, our prior panel precedent rule requires that this panel afford the same respect to Gonzales.

A couple of other circuits in post -Robinson decisions have reached conclusions on the coverage issue that are different from the Ninth Circuit's in Weyer and the Seventh Circuit's in Morgan (and that of our prior panel in Gonzales). See Ford v. Schering-Plough Corp., 145 F.3d 601 (3rd Cir. 1998), cert. denied, 525 U.S. 1093, 119 S. Ct. 850 (1999); Castellano v. City of New York, 142 F.3d 58 (2d Cir. 1998). The point, however, is not whether Weyer and Morgan are right or wrong on the ultimate issue involving the scope of the ADA as it concerns former employees. The point is that Weyer and Morgan irrefutably demonstrate that reasonable jurists can conclude Robinson does not undermine or contradict Gonzales to the point of overruling it, because two panels composed of six federal court of appeals judges have reached the same conclusion as Gonzales after they fully examined and considered Robinson.

In view of that, Judge Barkett's statement that "there is no doubt that the Robinson Court's decision invalidates Gonzales" is one with which I cannot agree. Judge Barkett obviously disagrees with the conclusions of Judges Flaum, Posner, and Williams of the Seventh Circuit in Morgan and of Judges Beezer, Wiggins, and Kleinfeld of the Ninth Circuit in Weyer, but the fact that there is disagreement, and a two-to-two split of the other circuits that have addressed the issue since Robinson was released proves that there is substantial doubt about whether Robinson overruled or undermined to the point of abrogation the Gonzales decision.

That ought to be all this panel needs to know in order to decide this appeal, because if reasonable jurists can disagree about whether a Supreme Court decision contradicts the holding of a prior panel, as they undeniably have, that prior panel holding must be followed. The prior panel precedent rule - the fundamental ground rule under which we operate as members of a court that sits in panels, the rule that keeps the peace among us, and the rule that is essential to the stability and predictability of our circuit law - demands no less.

I express no view on how I would vote to decide the ADA coverage issue if we were reconsidering it en banc, see generally United States v. Steele, 147 F.3d at 1318 (en banc court not bound by prior panel decisions), nor do I express any view on the other issue the panel addresses (the one involving differential treatment of the mentally and physically disabled), because that issue would be moot if this panel followed Gonzales, as it is bound to and should do.

Before ANDERSON, Chief Judge, and TJOFLAT, EDMONDSON, BIRCH, DUBINA, BLACK, CARNES, BARKETT, HULL, MARCUS and WILSON, Circuit Judges.

A member of this court in active service having requested a poll on the suggestion of rehearing en banc and a majority of the judges in this court in active service having voted in favor of granting a rehearing en banc,

IT IS ORDERED that the above cause shall be reheard by this court en banc. The previous panel's opinion is hereby vacated.