Meacham v. Knolls Atomic Power Laboratory

delivered the opinion of the Court.

A provision of the Age Discrimination in Employment Act of 1967 (ADEA), 81 Stat. 602, as amended, 29 U. S. C. § 621 et seq., creates an exemption for employer actions “otherwise prohibited” by the ADEA but “based on reasonable factors other than age” (RFOA). § 623(f)(1). The question is whether an employer facing a disparate-impact claim and planning to defend on the basis of RFOA must not only produce evidence raising the defense, but also persuade the factfinder of its merit. We hold that the employer must do both.

I

The National Government pays private companies to do some of the work maintaining the Nation’s fleet of nuclear-powered warships. One such contractor is respondent KAPL, Inc. (Knolls), the operator of the Government’s Knolls Atomic Power Laboratory, which has a history dating back to the first nuclear-powered submarines in the 1950s. The United States Navy and the Department of Energy jointly fund Knolls’s operations, decide what projects it should pursue, and set its annual staffing limits. In recent years, Knolls has been charged with designing prototype naval nuclear reactors and with training Navy personnel to run them.

The demands for naval nuclear reactors changed with the end of the Cold War, and for fiscal year 1996 Knolls was ordered to reduce its work force. Even after 100 or so employees chose to take the company’s ensuing buyout offer, *88Knolls was left with 30-some jobs to cut.1 Petitioners (Meacham, for short) are among those laid off in the resulting “involuntary reduction in force.” Brief for Petitioners 6. In order to select those for layoff, Knolls told its managers to score their subordinates on three scales, “performance,” “flexibility,” and “critical skills.”2 The scores were summed, along with points for years of service, and the totals determined who should be let go.

Of the 31 salaried employees laid off, 30 were at least 40 years old.3 Twenty-eight of them sued, raising both disparate-treatment (discriminatory intent) and disparate-impact (discriminatory result) claims under the ADEA and state law, alleging that Knolls “designed and implemented its workforce reduction process to eliminate older employees and that, regardless of intent, the process had a discriminatory impact on ADEA-protected employees.” Meacham v. Knolls Atomic Power Laboratory, 381 F. 3d 56, 61 (CA2 2004) (Meacham I). To show a disparate impact, the workers relied on a statistical expert's testimony to the effect that results so skewed according to age could rarely occur *89by chance;4 and that the scores for “flexibility” and “criticality,” over which managers had the most discretionary judgment, had the firmest statistical ties to the outcomes. Id., at 65.

The jury found for Meacham on the disparate-impact claim (but not on the disparate-treatment claim). The Court of Appeals affirmed, after examining the verdict through the lens of the so-called “burden shifting” scheme of inference spelled out in Wards Cove Packing Co. v. Atonio, 490 U. S. 642 (1989). See Meacham I, supra, at 74-76.5 After Knolls sought certiorari, we vacated the judgment and remanded for further proceedings in light of Smith v. City of Jackson, 544 U. S. 228 (2005), decided while Knolls’s petition was pending, see 544 U. S. 957 (2005).

On remand, the same Court of Appeals panel ruled in favor of Knolls, over a dissent. 461 F. 3d 134 (CA2 2006) (Meacham II) (case below). The majority found its prior ruling “untenable” because it had applied the Wards Cove “business necessity” standard rather than a “reasonableness” test, contrary to City of Jackson; and on the latter standard, Meacham, the employee, had not carried the burden of persuasion. 461 F. 3d, at 140-141, 144 (internal quotation marks *90omitted).6 In dissent, Judge Pooler took issue with the majority for confusing business justifications under Wards Cove with the statutory RFOA exemption, which she read to be an affirmative defense with the burden of persuasion falling on defendants. 461 F. 3d, at 147, 149-152.7

Meacham sought certiorari, noting conflicting decisions assigning the burden of persuasion on the reasonableness of the factor other than age; the Court of Appeals in this case placed it on the employee (to show the non-age factor unreasonable), but the Ninth Circuit in Criswell v. Western Airlines, Inc., 709 F. 2d 544, 552 (1983), had assigned it to the employer (to show the factor was a reasonable one). In fact it was in Criswell that we first took up this question, only to find it not well posed in that case. Western Air Lines, Inc. v. Criswell, 472 U. S. 400, 408, n. 10 (1985). We granted certiorari, 552 U. S. 1162 (2008), and now vacate the judgment of the Second Circuit and remand.8

*91II

A

The ADEA’s general prohibitions against age discrimination, 29 U. S. C. §§623(a)-(c), (e), are subject to a separate provision, § 623(f), creating exemptions for employer practices “otherwise prohibited under subsections (a), (b), (c), or (e).” The RFOA exemption is listed in § 623(f) alongside one for bona fide occupational qualifications (BFOQ): “It shall not be unlawful for an employer ... to take any action otherwise prohibited under subsections (a), (b), (c), or (e)... where age is a bona fide occupational qualification reasonably necessary to the normal operation of the particular business, or where the differentiation is based on reasonable factors other than age ....” § 623(f)(1).

Given how the statute reads, with exemptions laid out apart from the prohibitions (and expressly referring to the prohibited conduct as such), it is no surprise that we have already spoken of the BFOQ and RFOA provisions as being among the ADEA’s “five affirmative defenses,” Trans World Airlines, Inc. v. Thurston, 469 U. S. 111, 122 (1985). After looking at the statutory text, most lawyers would accept that characterization as a matter of course, thanks to the familiar principle that “[w]hen a proviso . .. carves an exception out of the body of a statute or contract those who set up such exception must prove it.” Javierre v. Central Altagracia, 217 U. S. 502, 508 (1910) (opinion for the Court by Holmes, J.); see also FTC v. Morton Salt Co., 334 U. S. 37, 44-45 (1948) (“[T]he burden of proving justification or exemption under a special exception to the prohibitions of a statute generally rests on one who claims its benefits . . . ”); United States v. First City Nat. Bank of Houston, 386 U. S. 361, 366 (1967) (citing Morton Salt Co., supra, at 44-45). That longstanding convention is part of the backdrop against which the Congress writes laws, and we respect it unless we have compelling reasons to think that Congress meant to put *92the burden of persuasion on the other side. See Schaffer v. Weast, 546 U. S. 49, 57-58 (2005) (“Absent some reason to believe that Congress intended otherwise, therefore, we will conclude that the burden of persuasion lies where it usually falls, upon the party seeking relief”).

We have never been given any reason for a heterodox take on the RFOA clause’s nearest neighbor, and our prior cases recognize that the BFOQ clause establishes an affirmative defense against claims of disparate treatment. See, e.g., City of Jackson, 544 U. S., at 233, n. 3; Western Air Lines, Inc., supra, at 414-419, and nn. 24, 29. We have likewise given the affirmative defense construction to the exemption in the Equal Pay Act of 1963 for pay differentials based on “any other factor other than sex,” Corning Glass Works v. Brennan, 417 U. S. 188, 196 (1974) (internal quotation marks omitted); and there, we took account of the particular weight given to the interpretive convention already noted, when enforcing the Fair Labor Standards Act of 1938 (FLSA), id., at 196-197 (“[T]he general rule [is] that the application of an exemption under the [FLSA] is a matter of affirmative defense on which the employer has the burden of proof”). This focus makes the principle of construction the more instructive in ADEA cases: “in enacting the ADEA, Congress exhibited both a detailed knowledge of the FLSA provisions and their judicial interpretation and a willingness to depart from those provisions regarded as undesirable or inappropriate for incorporation,” Lorillard v. Pons, 434 U. S. 575, 581 (1978). And we have remarked and relied on the “significant indication of Congress’ intent in its directive that the ADEA be enforced in accordance with the ‘powers, remedies, and procedures’ of the FLSA.” Id., at 580 (quoting 29 U. S. C. § 626(b); emphasis deleted); see also Fogerty v. Fantasy, Inc., 510 U. S. 517, 528 (1994) (applying reasoning of Lorillard); Thurston, supra, at 126 (same). As against this interpretive background, there is no hint in the text that Congress meant § 623(f)(1) to march out of step with either *93the general or specifically FLSA default rules placing the burden of proving an exemption on the party claiming it.

With these principles and prior cases in mind, we find it impossible to look at the text and structure of the ADEA and imagine that the RFOA clause works differently from the BFOQ clause next to it. Both exempt otherwise illegal conduct by reference to a further item of proof, thereby creating a defense for which the burden of persuasion falls on the “one who claims its benefits,” Morton Salt Co., supra, at 44-45, the “party seeking relief,” Schaffer, supra, at 57-58, and here, “the employer,” Corning Glass Works, supra, at 196.

If there were any doubt, the stress of the idiom “otherwise prohibited,” prefacing the BFOQ and RFOA conditions, would dispel it.9 The implication of affirmative defense is *94underscored by contrasting § 623(f)(1) with the section of the ADEA at issue in Public Employees Retirement System of Ohio v. Betts, 492 U. S. 158 (1989), and by the way Congress responded to our decision there. In Betts, we said the issue was whether a provision in a former version of § 623(f)(2), one about employee benefit plans, merely “redefine[d] the elements of a plaintiff’s prima facie case,” or instead “established] a defense” to what “otherwise would be a violation of the Act.” Id., at 181.10 Although the provision contained no “otherwise prohibited” kind of language, we said that it “appears on first reading to describe an affirmative defense.” Ibid. We nonetheless thought that this more natural view (which we had taken in Thurston) was overridden by evidence of legislative history, by the peculiarity of a pretext-revealing condition in the phrasing of the provision (that a benefit plan “not [be] a subterfuge to evade the purposes” of the ADEA), and by the parallel with a prior case construing an “analogous provision of Title VII” of the Civil Rights Act of 1964 (analogous because it also contained a pretext-revealing condition). 492 U. S., at 181. A year later, however, Congress responded to Betts by enacting the Older Workers Benefit Protection Act, 104 Stat. 978, avowedly to “restore the original congressional intent” that the ADEA’s benefits provision be. read as an affirmative defense, id., § 101. What is instructive on the question at hand is that, in clarifying that § 623(f)(2) specifies affirmative defenses, Congress not only set the burden in so many words but also added the phrase “otherwise prohibited” as a part of the preface (just as in the text of § 623(f)(1)). 11 Congress thus *95confirmed the natural implication that we find in the “otherwise prohibited” language in § 623(f)(1): it refers to an excuse or justification for behavior that, standing alone, violates the statute’s prohibition. The amendment in the aftermath of Betts shows that Congress understands the phrase the same way we naturally read it, as a clear signal that a defense to what is “otherwise prohibited” is an affirmative defense, entirely the responsibility of the party raising it.

B

Knolls ventures that, regardless, the RFOA provision should be read as mere elaboration on an element of liability. Because it bars liability where action is taken for reasons “other than age,” the argument goes, the provision must be directed not at justifying age discrimination by proof of some extenuating fact but at negating the premise of liability under § 623(a)(2), “because of. . . age.”

The answer to this argument, however, is City of Jackson, where we confirmed that the prohibition in § 623(a)(2) extends to practices with a disparate impact, inferring this result in part from the presence of the RFOA provision at issue here.12 We drew on the recognized distinction between disparate-treatment and disparate-impact forms of liability, and explained that “the very definition of disparate impact” was that “an employer who classifies his employees without respect to age may still be liable under the terms *96of this paragraph if such classification adversely affects the employee because of that employee’s age.” 544 U. S., at 236, n. 6 (plurality opinion); id., at 243 (Scalia, J., concurring in part and concurring in judgment) (expressing agreement with “all of the Court’s reasoning” in the plurality opinion, but finding it a basis for deference to the EEOC rather than for independent judicial decision). We emphasized that these were the kinds of employer activities, “otherwise prohibited” by § 623(a)(2), that were mainly what the statute meant to test against the RFOA condition: because “[i]n disparate-impact cases . . . the allegedly ‘otherwise prohibited’ activity is not based on age,” it is “in cases involving disparate-impact claims that the RFOA provision plays its principal role by precluding liability if the adverse impact was attributable to a nonage factor that was ‘reasonable.’ ” Id., at 239 (plurality opinion).

Thus, in City of Jackson, we made it clear that in the typical disparate-impact case, the employer’s practice is “without respect to age” and its adverse impact (though “because of age”) is “attributable to a nonage factor”; so action based on a “factor other than age” is the very premise for disparate-impact liability in the first place, not a negation of it or a defense to it. The RFOA defense in a disparate-impact case, then, is not focused on the asserted fact that a non-age factor was at work; we assume it was. The focus of the defense is that the factor relied upon was a “reasonable” one for the employer to be using. Reasonableness is a justification categorically distinct from the factual condition “because of age” and not necessarily correlated with it in any particular way: a reasonable factor may lean more heavily on older workers, as against younger ones, and an unreasonable factor might do just the opposite.13

*97III

The Court of Appeals majority rejected the affirmative defense reading and arrived at its position on the burden of proof question by a different route: because it read our decision in City of Jackson as ruling out the so-called “business necessity” enquiry in ADEA cases, the court concluded that the RFOA defense “replaces” it and therefore must conform to its burden of persuasion resting on the complaining party. But the court’s premise (that City of Jackson modified the “business necessity” enquiry) is mistaken; this alone would be reason enough to reject its approach. And although we are now satisfied that the business necessity test should have no place in ADEA disparate-impact cases, we agree with the Government that this conclusion does not stand in the way of our holding that the RFOA exemption is an affirmative defense. See Brief for United States as Amicus Curiae 25-27.

To begin with, when the Court of Appeals further inferred from the City of Jackson reference to Wards Cove that the Wards Cove burden of persuasion (on the employee, for the business necessity enquiry) also applied to the RFOA defense, it gave short shrift to the reasons set out in Part II-A, *98supra, for reading RFOA as an affirmative defense (with the burden on the employer). But we think that even on its own terms, City of Jackson falls short of supporting the Court of Appeals’s conclusion.

Although City of Jackson contains the statement that “ Wards Cove’s pre-1991 interpretation of Title VII’s identical language remains applicable to the ADEA,” 544 U. S., at 240, City of Jackson made only two specific references to aspects of the Wards Cove interpretation of Title VII that might have “remain[ed] applicable” in ADEA cases. One was to the existence of disparate-impact liability, which City of Jackson explained was narrower in ADEA cases than under Title VII. The other was to a plaintiff-employee’s burden of identifying which particular practices allegedly cause an observed disparate impact, which is the employee’s burden under both the ADEA and the pre-1991 Title VII. See 544 U. S., at 241. Neither of these references, of course, is at odds with the view of RFOA as an affirmative defense.

If, indeed, City of Jackson’s reference to Wards Cove could be read literally to include other aspects of the latter case, beyond what mattered in City of Jackson itself, the untoward consequences of the broader reading would rule it out. One such consequence is embraced by Meacham, who argues both that the Court of Appeals was wrong to place the burden of persuasion for the RFOA defense on the employee, and that the court was right in thinking that City of Jackson adopted the Wards Cove burden of persuasion on what Meacham views as one element of an ADEA impact claim. For Meacham takes the position that an impact plaintiff like himself has to negate business necessity in order to show that the employer’s actions were “otherwise prohibited”; only then does the RFOA (with the burden of persuasion on the employer) have a role to play. To apply both tests, however, would force the parties to develop (and the court or jury to follow) two overlapping enquiries: first, whether the employment practice at issue (based on a factor other than age) is *99supported by a business justification; and second, whether that factor is a reasonable one. Depending on how the first enquiry proceeds, a plaintiff might directly contest the force of the employer’s rationale, or else try to show that the employer invoked it as a pretext by pointing (for example) to alternative practices with less of a disparate impact. See Wards Cove, 490 U. S., at 658 (“first, a consideration of the justifications an employer offers for his use of these practices; and second, the availability of alternative practices to achieve the same business ends, with less racial impact”); see also id., at 658-661. But even if the plaintiff succeeded at one or the other, in Meacham’s scheme the employer could still avoid liability by proving reasonableness.

Here is what is so strange: as the Government says, “[i]f disparate-impact plaintiffs have already established that a challenged practice is a pretext for intentional age discrimination, it makes little sense then to ask whether the discriminatory practice is based on reasonable factors other than age.” Brief for United States as Amicus Curiae 26 (emphasis in original). Conversely, proving the reasonableness defense would eliminate much of the point a plaintiff would have had for showing alternatives in the first place: why make the effort to show alternative practices with a less discriminatory effect (and besides, how would that prove pretext?), when everyone knows that the choice of a practice relying on a “reasonable” non-age factor is good enough to avoid liability?14 At the very least, developing the reasonableness defense would be substantially redundant with the direct contest over the force of the business justification, especially when both enquiries deal with the same, narrowly *100specified practice. It is not very fair to take the remark about Wards Cove in City of Jackson as requiring such a wasteful and confusing structure of proof.

Nor is there any good way to read the same line from City of Jackson as implying that the burden of proving any business-related defense falls on the plaintiff; most obviously, this would entail no longer taking the BFOQ clause to be an affirmative defense, which City of Jackson confirmed that it is, see 544 U. S., at 233, n. 3. What is more, City of Jackson could not have had the RFOA clause in mind as “identical” to anything in Title VII (for which a Wards Cove’s reading might be adopted), for that statute has no like-worded defense. And as Wards Cove did not purport to construe any statutory defenses under Title VII, only an over-reading of City of Jackson would find lurking in it an assumption that Wards Cove has anything to say about statutory defenses in the ADEA (never mind one that Title VII does not have).

IV

As mentioned, where City of Jackson did get help from our prior reading of Title VII was in relying on Wards Cove to repeat that a plaintiff falls short by merely alleging a disparate impact, or “pointing] to a generalized policy that leads to such an impact.” City of Jackson, 544 U. S., at 241. The plaintiff is obliged to do more: to “isolat[e] and identif[y] the specific employment practices that are allegedly responsible for any observed statistical disparities.” Ibid, (quoting Wards Cove, supra, at 656; emphasis in original; internal quotation marks omitted). The aim of this requirement, as City of Jackson said, is to avoid the “result [of] employers being potentially liable for ‘the myriad of innocent causes that may lead to statistical imbalances.’” 544 U. S., at 241 (quoting Wards Cove, supra, at 657; some internal quotation marks omitted). And as the outcome in that case shows, the requirement has bite: one sufficient reason for rejecting the *101employees’ challenge was that they “ha[d] done little more than point out that the pay plan at issue [was] relatively less generous to older workers than to younger workers,” and “ha[d] not identified any specific test, requirement, or practice within the pay plan that ha[d] an adverse impact on older workers.” City of Jackson, supra, at 241.

Identifying a specific practice is not a trivial burden, and it ought to allay some of the concern raised by Knolls’s amici, who fear that recognizing an employer’s burden of persuasion on an RFOA defense to impact claims will encourage strike suits or nudge plaintiffs with marginal cases into court, in turn inducing employers to alter business practices in order to avoid being sued. See, e. g., Brief for General Electric Co. as Amicus Curiae 18-31. It is also to the point that the only thing at stake in this case is the gap between production and persuasion; nobody is saying that even the burden of production should be placed on the plaintiff. Cf. Schaffer, 546 U. S., at 56 (burden of persuasion answers “which party loses if the evidence is closely balanced”); id., at 58 (“In truth, however, very few cases will be in evidentiary equipoise”). And the more plainly reasonable the employer’s “factor other than age” is, the shorter the step for that employer from producing evidence raising the defense, to persuading the factfinder that the defense is meritorious. It will be mainly in cases where the reasonableness of the nonage factor is obscure for some reason, that the employer will have more evidence to reveal and more convincing to do in going from production to persuasion.

That said, there is no denying that putting employers to the work of persuading factfinders that their choices are reasonable makes it harder and costlier to defend than if employers merely bore the burden of production; nor do we doubt that this will sometimes affect the way employers do business with their employees. But at the end of the day, amici’s concerns have to be directed at Congress, which set the balance where it is, by both creating the RFOA exemp*102tion and writing it in the orthodox format of an affirmative defense. We have to read it the way Congress wrote it.

* * *

As we have said before, Congress took account of the distinctive nature of age discrimination, and the need to preserve a fair degree of leeway for employment decisions with effects that correlate with age, when it put the RFOA clause into the ADEA, “significantly narrow[ing] its coverage.” City of Jackson, 544 U. S., at 233. And as the outcome for the employer in City of Jackson shows, “it is not surprising that certain employment criteria that are routinely used may be reasonable despite their adverse impact on older workers as a group.” Id., at 241. In this case, we realize that the Court of Appeals showed no hesitation in finding that Knolls prevailed on the RFOA defense, though the court expressed its conclusion in terms of Meacham’s failure to meet the burden of persuasion. Whether the outcome should be any different when the burden is properly placed on the employer is best left to that court in the first instance. The judgment of the Court of Appeals is vacated, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Justice Breyer took no part in the consideration or decision of this case.