Part I. Rulings — Internal Revenue Bulletin 2001-44

lationship for a conventional corporate taxpayer. Comparable treatment of PLL for the group and the conventional taxpayer can be achieved only if PLEs are compared with the loss amount at the consolidated level after CNOL has been determined, for CNOL is the only NOL measure for the group. An approach based on comparable treatment is also (relatively) easy to understand and to apply. Pp. 5–7.

SOUTER, J., delivered the opinion of the Court, in which REHNQUIST, C.J., and O’CONNOR, SCALIA, KENNEDY, THOMAS, GINSBURG, and BREYER, JJ., joined. THOMAS, J., filed a concurring opinion. STEVENS, J., filed a dissenting opinion.

Section 172.—Net Operating Loss Deduction

Ct. D. 2072

SUPREME COURT OF THE

UNITED STATES

No. 00–157

UNITED DOMINION INDUSTRIES,

INC. v . UNITED STATES

CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR

THE FOURTH CIRCUIT

June 4, 2001

Syllabus Under the Internal Revenue Code of 1954, a “net operating loss” (NOL) results from deductions in excess of gross income for a given year. 26 U.S.C. Sec. 172(c). A taxpayer may carry its NOL either backward or forward to other tax years in order to set off its lean years against its lush years. Sec. 172(b)(1)(A). The carryback period for “product liability loss[es]” is 10 years. Sec. 172(b)(1)(I). Because a product liability loss (PLL) is the total of a taxpayer’s product liability expenses (PLEs) up to the amount of its NOL, Sec. 172(j)(1), a taxpayer with a positive annual income, and thus no NOL, may have PLEs but can have no PLL. An affiliated group of corporations may file a single consolidated return. Sec. 1501. Treasury Regulations provide that such a group’s “consolidated taxable income” (CTI), or, alternatively, its “consolidated net operating loss” (CNOL), is determined by taking into account several items, the first of which is the “separate taxable income” (STI) of each group member. In calculating STI, the member must disregard items such as capital gains and losses, which are considered, and factored into CTI or CNOL, on a consolidated basis. Petitioner’s predecessor in interest, AMCA International Corporation, was the parent of an affiliated group filing consolidated returns for the years 1983 through 1986. In each year, AMCA reported CNOL exceeding the aggregate of its 26 individual members’ PLEs. Five group members with PLEs reported positive STIs. Nonetheless,

AMCA included those PLEs in determining its PLL for 10-year carryback under a “single-entity” approach in which it compared the group’s CNOL and total PLEs to determine the group’s total PLL. In contrast, the Government’s “separate-member” approach compares each affiliate’s STI and PLEs in order to determine whether each affiliate suffers a PLL, and only then combines any PLLs of the individual affiliates to determine a consolidated PLL. Under this approach, PLEs incurred by an affiliate with positive STI cannot contribute to a PLL. In 1986 and 1987, AMCA petitioned the Internal Revenue Service for refunds based on its PLL calculations. The IRS ruled in AMCA’s favor, but was reversed by a joint congressional committee that controls refunds exceeding a certain threshold. AMCA then filed this refund action. The District Court applied AMCA’s single-entity approach, concluding that so long as the affiliated group’s consolidated return reflects CNOL in excess of the group’s aggregate PLEs, the total of those expenses is a PLL that may be carried back. In reversing, the Fourth Circuit applied the separate-member approach.

Held: An affiliated group’s PLL must be figured on a consolidated, single-entity basis, not by aggregating PLLs separately determined company by company. Pp. 5–15. (a) The single-entity approach to calculating an affiliated group’s PLL is straightforward. The first step in applying Sec. 172(j)’s definition of PLL requires a taxpayer filing a consolidated return to calculate an NOL. The Code and regulations governing affiliated groups of corporations filing consolidated returns provide only one definition of NOL: “consolidated” NOL. The absence of a separate NOL for a group member in this context is underscored by the fact that the regulations provide a measure of separate NOL in a different context, for any year in which an affiliated corporation files a separate return. The exclusive definition of NOL as CNOL at the consolidated level is important. Neither the Code nor the regulations indicate that the essential relationship between NOL and PLL for a consolidated group differs from their re

(b) The case for the separate-member approach is not so easily made. Because there is no NOL below the consolidated level, there is nothing for comparison with PLEs to produce a PLL at any stage before the CNOL calculation. Thus, a separate-member proponent must identify some figure in the consolidated return scheme with a plausible analogy to NOL at the affiliated corporations level. An individual member’s STI is not analogous, for it excludes several items that an individual taxpayer would normally count in computing income or loss, but which an affiliated group may tally only at the consolidated level. The “separate net operating loss,” Treas. Reg. Sec. 1.1502–79(a)(3), used by the Fourth Circuit fares no better. Although that figure accounts for some gains or losses that STI does not, Sec. 1.1502–79(a)(3)’s purpose is to allocate CNOL to an affiliate member seeking to carry back a loss to a year in which the member was not part of the consolidated group. Such returns are not at issue here. Pp. 8–11.

(c) Several objections to the single-entity approach—that it allows affiliated groups a double deduction, that the omission of PLEs from the series of items that Treas. Reg. Sec. 1.1502–12 requires to be tallied at the consolidation level indicates that PLEs were not meant to be tallied at that level, and that the single-entity approach would permit significant tax avoidance abuses—are rejected. Pp. 11–15.

208 F.3d 452, reversed and remanded.

2001–44 I.R.B. 379 October 29, 2001

bined at the level of the group filing the single return, where deductions otherwise attributable to one member (say, for a charitable contribution) can offset income received by another (from a capital gain, for example). Treas. Regs. Secs. 1.1502–11(a)(3)–(8); 1.1502–21(f)(2) to (6). A consolidated group’s CTI or CNOL, therefore, is the sum of each member’s STI, plus or minus a handful of items considered on a consolidated basis.

Petitioner United Dominion’s predecessor in interest, AMCA International Corporation, was the parent of an affiliated group of corporations that properly elected to file consolidated tax returns for the years 1983 through 1986. In each of these years, AMCA reported CNOL (the lowest being $85 million and the highest, $140 million) that exceeded the aggregate of its 26 individual members’ PLEs ($3.5 million to $6.5 million). This case focuses on the PLEs of five of AMCA’s member companies, which, together, generated roughly $205,000 in PLEs in 1983, $1.6 million in 1984, $1.3 million in 1985, and $250,000 in 1986. No one disputes these amounts or their characterization as PLEs. See 208 F.3d 452, 453 (CA4 2000) (“The parties agree” with respect to the amount of “the product liability expenses incurred by the five group members in the relevant years”). Rather, the sole question here is whether the AMCA affiliated group may include these amounts on its consolidated return, in determining its PLL for 10-year carryback. The question arises because of the further undisputed fact that in each of the relevant tax years, each of the five companies in question (with minor exceptions not relevant here), reported a positive STI.

AMCA answered this question by following what commentators have called a “single-entity” approach ⁴ to calculating its “consolidated” PLL. For each tax year, AMCA (1) calculated its CNOL pursuant to Treas. Reg. Sec. 1.1502–11(a), and (2) aggregated its individual members’ PLEs. Because, as noted above, for each tax year AMCA’s CNOL was greater

4 Axelrod & Blank, The Supreme Court, Consolidated Returns, and 10-Year Carrybacks, 90 Tax Notes, No. 10, p. 1383 (Mar. 5, 2001) (hereinafter Axelrod & Blank).

SUPREME COURT OF THE

UNITED STATES

No. 00–157

UNITED DOMINION INDUSTRIES,

INC., PETITIONER v . UNITED

STATES

532 U.S. ___ (2001)

ON WRIT OF CERTIORARI TO THE

UNITED STATES COURT OF

APPEALS FOR THE

FOURTH CIRCUIT

June 4, 2001

JUSTICE SOUTER delivered the opinion of the Court.

Under Sec. 172(b)(1)(I) of the Internal Revenue Code of 1954, a taxpayer may carry back its “product liability loss” up to 10 years in order to offset prior years’ income. The issue here is the method for calculating the product liability loss of an affiliated group of corporations electing to file a consolidated federal income tax return. We hold that the group’s product liability loss must be figured on a consolidated basis in the first instance, and not by aggregating product liability losses separately determined company by company.

A “net operating loss” results from deductions in excess of gross income for a given year. 26 U.S.C. Sec. 172(c). ¹

Under Sec. 172(b)(1)(A), a taxpayer may carry its net operating loss either backward to past tax years or forward to future tax years in order to “set off its lean years against its lush years, and to strike something like an average taxable income computed over a period longer than one year,” Libson Shops, Inc. v. Koehler, 353 U.S. 382, 386 (1957).

Although the normal carryback period was at the time three years, in 1978, Congress authorized a special 10-year carryback for “product liability loss[es],” 26 U.S.C. Sec. 172(b)(1)(I), since, it understood, losses of this sort tend to be particularly “large and sporadic.” Joint Committee on Taxation, 95th Cong., General Explanation of the Revenue Act of 1978,

1 Unless otherwise noted, all statutory references are to the Internal Revenue Code of 1954, 26 U.S.C. Sec. 1 et seq . (1982 ed. and Supp. V), as in effect between 1983 and 1986, the tax years here in question.

232 (Comm. Print 1979). The Code defines “product liability loss,” for a given tax year, as the lesser of (1) the taxpayer’s “net operating loss for such year” and (2) its allowable deductions attributable to product liability “expenses.” 26 U.S.C. Sec. 172(j)(1). In other words, a taxpayer’s product liability loss (PLL) is the total of its product liability expenses (PLEs), limited to the amount of its net operating loss (NOL). By definition, then, a taxpayer with positive annual income, and thus no NOL, may have PLEs but can have no PLL. ²

Instead of requiring each member company of “[a]n affiliated group of corporations” to file a separate tax return, the Code permits the group to file a single consolidated return, 26 U.S.C. Sec. 1501, and leaves it to the Secretary of the Treasury to work out the details by promulgating regulations governing such returns, Sec. 1502. Under Treas. Regs. Secs. 1.1502–11(a) and 1.1502–21(f), ³ an affiliated group’s “consolidated taxable income” (CTI), or, alternatively, its “consolidated net operating loss” (CNOL), is determined by “taking into account” several items. The first is the “separate taxable income” (STI) of each group member. A member’s STI (whether positive or negative) is computed as though the member were a separate corporation ( i.e ., by netting income and expenses), but subject to several important “modifications.” Treas. Reg. Sec. 1.1502–12. These modifications require a group member calculating its STI to disregard, among other items, its capital gains and losses, charitable-contribution deductions, and dividends-received deductions. Ibid . These excluded items are accounted for on a consolidated basis, that is, they are com

2 If, for example, a company had $100 in taxable income, $50 in deductible PLEs, and $75 in additional deductions, its NOL would be $25 ( i.e ., $100$50-$75= -$25); it could count only $25 of its $50 in PLEs as PLL. If the company had $100 in income, $50 in PLEs, and $125 in additional deductions, its NOL would be $75, and it could count its entire $50 in PLEs as PLL. And, finally, if the company had $100 in income, $50 in PLEs, and $40 in additional deductions, it would have positive income and, thus, no NOL and no PLL. 3 Unless otherwise noted, Treasury Regulation references are to the regulations in effect between 1983 and 1986, 26 CFR Sec. 1.1502-11 et seq . (19821986).

October 29, 2001 380 2001–44 I.R.B.

than the sum of its members’ PLEs, AMCA treated the full amount of the PLEs as consolidated PLL eligible for 10year carryback. In AMCA’s view, the fact that several member companies throwing off large PLEs also, when considered separately, generated positive taxable income was of no significance.

From the Government’s perspective, however, the fact that the several affiliated members with PLEs also generated positive separate taxable income is of critical significance. According to the Government’s methodology, which we will call the “separate-member” approach, ⁵

PLEs incurred by an affiliate with positive separate taxable income cannot contribute to a PLL eligible for 10-year carryback. Whereas AMCA compares the group’s total income (or loss) and total PLEs in an effort to determine the group’s total PLL, the Government compares each affiliate’s STI and PLEs in order to determine whether each affiliate suffers a PLL, and only then combines any PLLs of the individual affiliates to determine a consolidated PLL amount.

In 1986 and 1987, AMCA petitioned the Internal Revenue Service for refunds of taxes based on its PLL calculations. The IRS first ruled in AMCA’s favor but was reversed by the Joint Committee on Internal Revenue Taxation of the United States Congress, which controls refunds exceeding a certain threshold, 26 U.S.C. Sec. 6405(a). AMCA then filed this refund action in the United States District Court for the Western District of North Carolina. The District Court agreed with AMCA that an affiliated group’s PLL is determined on a single-entity basis, and held that, so long as the group’s consolidated return reflects CNOL in excess of the group’s aggregate PLEs, the total of those expenses (including those incurred by members with positive separate taxable income) is a PLL that “may be carried back the full ten years.” No. 3:95CV-341-MU (June 19, 1998), App. to Pet. for Cert. 39a. The United States Court of Appeals for the Fourth Circuit reversed, and held that “determining ‘product liability loss’ separately for each group member is correct and consistent with [Treasury] regulations.” 208 F.3d, at 458.

Because the Fourth Circuit’s separatemember approach to calculating PLL conflicted with the Sixth Circuit’s adoption of the single-entity approach in Intermet Corp. v. Commissioner, 209 F.3d 901 (CA6 2000), we granted certiorari, 531 U.S. 1009 (2000). ⁶ We now reverse.

III

The case for the single-entity approach to calculating an affiliated group’s PLL is straightforward. Section 172(j)(1) defines a taxpayer’s “product liability loss” for a given tax year as the lesser of its “net operating loss for such year” and its product liability “expenses.” In order to apply this definition, the taxpayer first determines whether it has taxable income or NOL, and in making that calculation it subtracts PLEs. If the result is NOL, the taxpayer then makes a simple comparison between the NOL figure and the total PLEs. The PLE total becomes the PLL to the extent it does not exceed NOL. That is, until NOL has been determined, there is no PLL.

The first step in applying the definition and methodology of PLL to a taxpayer filing a consolidated return thus requires the calculation of NOL. As United Dominion correctly points out, the Code and regulations governing affiliated groups of corporations filing consolidated returns provide only one definition of NOL: “consolidated” NOL, see Treas. Reg. Sec. 1.1502–21(f). There is no definition of separate NOL for a member of an affiliated group. Indeed, the fact that Treasury Regulations do provide a measure of separate NOL in a different context, for an affiliated corporation as to any year in which it filed a separate return, infra, at ___, underscores the absence of such a measure for an affiliated corporation filing as a group member. Given this apparently exclusive definition of NOL as CNOL in the instance of affiliated entities

with a consolidated return (and for reasons developed below, infra, at ___) we think it is fair to say, as United Dominion says, that the concept of separate NOL “simply does not exist.” Brief for Petitioner 15. ⁷ The exclusiveness of NOL at the consolidated level as CNOL is important here for the following reasons. The Code’s authorization of consolidated group treatment contains no indication that for a consolidated group the essential relationship between NOL and PLL will differ from their relationship for a conventional corporate taxpayer. Nor does any Treasury Regulation purport to change the relationship in the consolidated context. If, then, the relationship is to remain essentially the same, the key to understanding it lies in the regulations’ definition of net operating loss exclusively at the consolidated level. Working back from that, PLEs should be considered first in calculating CNOL, and they are: because any PLE of an affiliate affects the calculation of its STI, that same PLE necessarily affects the CTI or CNOL in exactly the same way, dollar for dollar. And because, by definition, there is no NOL measure for a consolidated return group or any affiliate except CNOL, PLEs cannot be compared with any NOL to produce PLL until CNOL has been calculated. Then, and only then in the case of the consolidated filer, can total PLEs be compared with a net operating loss. In sum, comparable treatment of PLL in the instances of the usual corporate taxpayer and group filing a consolidated return can be achieved only if the comparison of PLEs with a limiting loss amount occurs at the consolidated level after CNOL has been determined. This approach resting on comparable treatment has a further

7 In addition to Treas. Reg. Sec. 1.1502-79(a)(3), discussed infra, at ___, two other provisions, 26 U.S.C. Sec. 1503(f)(2) and the current version (though not the version applicable between 1983 and 1986) of Treas. Reg. Sec. 1502-21(b) (2000), refer to separate group members’ NOLs. The parties here have not emphasized those provisions, and with good reason. Not only are they inapplicable to the question before us (either substantively, temporally, or both), but, as one commentator has observed, their references to separate NOLs “stem[] more from careless drafting than meaningful design.” Leatherman, Are Separate Liability Losses Separate for Consolidated Groups?, 52 Tax. Law. 663, 705 (1999) (hereinafter Leatherman, Separate Liability Losses).

Pet. for Cert. 39a. The United States 6 Intermet involved “specified liability losses” Court of Appeals for the Fourth Circuit (SLLs), not PLLs. The difference, however, does reversed, and held that “determining not matter. The PLL was a statutory predecessor to

the SLL, and PLLs were folded into the SLL provi ‘product liability loss’ separately for each

sion in Sec. 11811(b)(1) of the Omnibus Budget

group member is correct and consistent Reconciliation Act of 1990, 104 Stat. 1388-532. with [Treasury] regulations.” 208 F.3d, Thus, “[i]n all relevant respects, the provisions on at 458. [PLLs] and SLLs are the same.” Leatherman,

Current Developments for Consolidated Groups, 486 PLI/Tax 389, 393, n. 5 (2000) (hereinafter 5 Ibid . Leatherman, Current Developments).

2001–44 I.R.B. 381 October 29, 2001

virtue entitled to some weight in case of doubt: it is (relatively) easy to understand and to apply.

The case for the separate-member approach, advanced (in one variant) by the Government and adopted (on a different rationale) by the Court of Appeals, is not so easily made. In the analysis of comparable treatment just set out, of course, there is no NOL below the consolidated level and hence nothing for comparison with PLEs to produce PLL at any stage before the CNOL calculation. At the least, then, a proponent of the separatemember approach must identify some figure in the consolidated return scheme that could have a plausible analogy to NOL at the level of the affiliated corporations. See A. Dubroff, J. Blanchard, J. Broadbent, & K. Duvall, Federal Income Taxation of Corporations Filing Consolidated Returns Sec. 41.04[06], p. 41–75 (2d ed. 2000) (hereinafter Dubroff) (“Even if separate entity treatment was appropriate, it is unclear how a member with [PLEs] would compute its separate NOL”). The Government and the Court of Appeals have suggested different substitute measures. Neither one works.

The Government has argued that an individual group member’s STI, as determined under Treas. Reg. Sec. 1.1502–12, is analogous to a “separate” NOL, so that an affiliate’s STI may be compared with its PLEs in order to determine any separate PLL. An individual member’s PLL would be the amount of its separate PLEs up to the amount of its negative STI; a member having positive STI could have no PLL.

The Government claims that an STIbased comparison places the group member closest to the position it would have occupied if it had filed a separate return. But that is simply not so. We have seen already that the calculation of a group member’s STI by definition excludes several items that an individual taxpayer would normally account for in computing income or loss, but which an affiliated group may tally only at the consolidated level, such as capital gains and losses, charitable-contribution deductions, and dividends-received deductions. Treas. Reg. Secs. 1.1502–12(j) to (n). Owing to these exclusions, an affiliate’s STI will tend to be inflated by eliminating deductions it would have taken if it had filed

separately, or deflated by eliminating an income item like capital gain.

When pushed, the Government concedes that STI is “not necessarily equivalent to the income or [NOL] figure that the corporation would have computed if it had filed a separate return.” Brief for United States 21, n. 14. But, the Government claims, “[t]here has never been a taxpayer with [PLEs] who had a positive

[STI] but a negative separate [NOL].” Tr. of Oral Arg. 27. In other words, the Government says that the deductions excluded from STI have never once made a difference and, therefore, that STI is, in fact, a decent enough proxy for a group member’s “separate” NOL. But whether or not the excluded items have made a difference in the past, or make a difference here, they certainly could make a difference and, given the potential importance of some of the deductions involved (a large charitable contribution, for example), it is not hard to see how the difference could favor the Government.

The Court of Appeals was therefore right to reject the Government’s reliance on STI as a functional surrogate for an affiliate’s “separate” NOL. 208 F.3d, at 459–460. But what the Court of Appeals used in place of STI fares no better. The court relied on Treas. Reg. Sec. 1.1502–79, which contains a definition of “separate net operating loss” that the court believed to be “analogous to an individual’s ‘net operating loss’ on a separate return.” 208 F.3d at 460. Section 1.1502–79(a)(3) provides that, “[f]or purposes of this subparagraph,” the “separate net operating loss of a member of the group shall be determined under Sec. 1.1502–12 . . ., adjusted for the . . . items taken into account in the computation of” the CNOL. As the Court of Appeals said, the directive of Sec. 1.1502–79(a)(3) (unlike the definition of STI) “takes into account, for example, [a] member’s charitable contributions” and other consolidated deductions. 208 F.3d, at 460–461.

But this sounds too good. It is true that, insofar as Sec. 1.1502–79(a)(3) accounts for gains and losses that STI does not, it gets closer to a commonsense notion of a group member’s “separate” NOL than STI does. But the fact that Sec. 1.1502–79(a)(3) improves on STI simply by undoing what Sec. 1.1502–12 requires in defining STI is suspicious, and it turns

out that the suspicion is justified. Section 1.1502–79(a)(3) unbakes the cake for only one reason, and that reason has no application here. The definition on which the Court of Appeals relied applies, by its terms, only “for purposes of” Sec. 1.1502–79(a)(3), and context makes clear that the purpose is to provide a way to allocate CNOL to an affiliate member that seeks to carry back a loss to a “separate return year,” that is, to a year in which the member was not part of the consolidated group. See Treas. Reg. Sec. 1.1502–79 (titled “Separate return years”); Sec. 1.1502–79(a) (titled “Carryover and carryback of [CNOL] to separate return years”); Sec. 1.1502–79(a)(1) (“[i]f a

[CNOL] can be carried . . . to a separate return year . . .”). No separate return years are at issue before us; all NOL carrybacks relevant here apply to years in which the five corporations were affiliated in the group. The Court of Appeals thus applied concepts addressing separate return years to a determination for a consolidated return year, without any statutory or regulatory basis for doing so. Cf. 49 Fed. Reg. 30530 (1984) (“[A]lthough the consolidated net operating loss is apportioned to individual members for purposes of carry backs to separate return years [under Sec. 1.1502–79(a)], the apportioned amounts are not separate NOLs of each member”). Hence, while Sec. 1.1502–79 might not distort an affiliate’s separate NOL in the same way that STI does, the facial inapplicability of that regulation only underscores the exclusive concern of Sec. 1.1502–11(a) with consolidated NOL. In sum, neither method for computing PLL on a separate-member basis squares with the notion of comparability as applied to consolidated return regulations. On the contrary, by expressly and exclusively defining NOL as CNOL, the regulations support the position that group members’ PLEs should be aggregated and the affiliated group’s PLL determined on a consolidated, single-entity basis.

Several objections have been raised to a single-entity approach to calculating PLL that we have not considered yet. First, the Government insists that a single-entity rule allows affiliated groups a “double deduction.” The Government ar

October 29, 2001 382 2001–44 I.R.B.

gues that because PLEs are not included among the specific items (charitable-contribution deductions, etc.) for which consolidated, single-entity treatment is required under Treas. Reg. Sec. 1.1502–12, PLEs are “consumed” or “used up” in computing members’ STIs, which, pursuant to Treas. Regs. Secs. 1.1502–11(a) and 1.1502–21(f), are then used to calculate the group’s CTI or CNOL. According to the Government, to permit the use of PLEs first to reduce an individual member’s STI and then to contribute to an aggregate PLL for carryback purposes would be tantamount to a double deduction.

The double-deduction argument may have superficial appeal, but any appeal it has rests on a fundamental misconception of the function of STI in computing an affiliated group’s tax liability. Calculation of a group member’s STI is not in and of itself the basis for any tax event, and there is no separate tax saving when STI is calculated; that occurs only when deductions on the consolidated return equal income and (if they exceed income and produce a CNOL) are carried back against prior income. STI is merely an accounting construct devised as an interim step in computing a group’s CTI or CNOL; it “has no other purpose.” Intermet, 209 F.3d, at 906 (“A member’s STI is simply a step along the way to calculating the group’s taxable income or CNOL”). The fact that a group member’s PLEs reduce its STI, which in turn either reduces the group’s CTI or contributes to its CNOL “dollar for dollar,” ibid ., is of no other moment. ⁸

If there were anything wrong in what AMCA proposes to do, it would be wrong in relation to AMCA’s CNOL and its use for any carryback. Yet, as noted above, no one here disputes that the group members had PLEs in the total amount claimed or that the AMCA group is entitled to carry back the full amount of its CNOL to offset income in prior years. The only

8 It makes no difference whatsoever whether the affiliate’s PLEs are (1) first netted against each member’s income and then aggregated or (2) first aggregated and then netted against the group’s combined income: under either method, AMCA’s CNOL is the same. See Axelrod & Blank 1394 (noting that this conclusion follows from “the associative principle of arithmetic (which holds that the groupings of items in the case of addition and subtraction have no effect on the result)”).

question is what portion, if any, of AMCA’s CNOL is PLL and, as such, eligible for 10-year, as opposed to 3-year carryback treatment. There is no more of a double deduction with a 10-year carryback than one for three years.

A second objection was the reason that the Court of Appeals rejected the singleentity approach. That court attached dispositive significance to the fact that, while the Treasury Regulation we have discussed, Sec. 1.1502–12, specifically provides that several items (capital gains and losses, charitable-contribution deductions, etc.) shall be accounted for on a consolidated basis, it does not similarly provide for accounting for PLEs on a consolidated basis: “The regulations provide for blending the group members’

[NOLs], and they explicitly define

[CNOL] without an accompanying reference to consolidated [PLEs]. This omission . . . makes clear that blending those expenses is not permitted. . . .” 208 F.3d, at 458.

We think the omission of PLEs from the series of items that Sec. 1.1502–12 requires to be tallied at the consolidated level has no such clear lesson, however. The logic that invests the omission with significance is familiar: the mention of some implies the exclusion of others not mentioned. Leatherman v. Tarrant County Narcotics Intelligence and Coor- dination Unit, 507 U.S. 163, 167 (1993) (“ Expressio unius est exclusio alterius ”). But here, as always, the soundness of that premise is a function of timing: if there was a good reason to consider the treatment of consolidated PLL at the time the regulation was drawn, then omitting PLL from the list of items for consolidated treatment may well have meant something. But if there was no reason to consider PLL then, its omission would mean nothing at all. And in fact, there was no reason. When the consolidated return regulations were first promulgated in 1966, there was no carryback provision pegged to PLEs or PLLs; those notions did not become separate carryback items until 1978, when the 10-year rule was devised. See Revenue Act of 1978, Sec. 371, 92 Stat. 2859; see also Leatherman, Current Developments 393, n. 5. Omission of PLEs or PLLs from the series set out for consolidated treatment in the 1966 regulation therefore meant absolutely

nothing in 1966. The issue, then, is the significance, not of omission, but of failure to include later: has the significance of the earlier regulation changed solely because the Treasury has never amended it, even though PLL is now a separate carryback? We think that is unlikely. The Treasury’s relaxed approach to amending its regulations to track Code changes is well documented. See, e.g., Dubroff 41–72, n. 193; Axelrod & Blank 1391; Leatherman, Separate Liability Losses 708–709. The absence of any amendment to Sec. 1.1502–12 that might have added PLEs or PLLs to the list of items for mandatory single-member treatment therefore is more likely a reflection of the Treasury’s inattention than any affirmative intention on its part to say anything at all.

Last, the Government warns that “[t]he rule that petitioner advocates would permit significant tax avoidance abuses.” Brief for United States 40. Specifically:

“Under petitioner’s approach, a corporation that is currently unprofitable but that had substantial income in prior years could (i) acquire a profitable corporation with product liability expense deductions in the year of acquisition, (ii) file a consolidated return and (iii) thereby create an otherwise nonexistent ‘product liability loss’ for the new affiliated group that would allow the acquiring corporation to claim refunds of the tax it paid in prior years. Ibid .

The Government suggests, for example, that “a manufacturing company (with prior profits and current losses) that has no product liability exposure could purchase a tobacco company (with both prior and current profits) that has significant product liability expenses,” and that “[t]he combined entity could . . . assert a ten-year carryback of ‘product liability losses’ even though the tobacco company has always made a profit and never incurred a ‘loss’ of any type.” Id., at 40–41, n. 27.

There are several answers. First, on the score of tax avoidance, the separate-member approach is no better (and is perhaps worse) than the single-entity treatment; both entail some risk of tax-motivated behavior. See Leatherman, Separate Liabil

2001–44 I.R.B. 383 October 29, 2001

ity Losses 681 (Under the separate-member approach, “[d]espite sound non-tax business reasons, a group may be disinclined to form a new member or transfer assets between members, because it may worry that it would lose the benefit of a ten-year carryback,” and “may be encouraged to transfer assets between members to increase its consolidated [PLL], even when those transfers would otherwise be ill-advised”). Second, the Government may, as always, address tax-motivated behavior under Internal Revenue Code Sec. 269, which gives the Secretary ample authority to “disallow [any] deduction, credit, or other allowance” that results from a transaction “the principal purpose

[of] which . . . is evasion or avoidance of Federal income tax.” 26 U.S.C. Sec. 269(a). And finally, if the Government were to conclude that Sec. 269 provided too little protection and that it simply could not live with the single-entity approach, the Treasury could exercise the authority provided by the Code, 26 U.S.C. Sec. 1502, and amend the consolidated return regulations.

Thus, it is true, as the Government has argued, that “[t]he Internal Revenue Code vests ample authority in the Treasury to adopt consolidated return regulations to effect a binding resolution of the question presented in this case.” Brief for United States 19–20. To the extent that the Government has exercised that authority, its actions point to the single-entity approach as the better answer. To the extent the Government disagrees, it may amend its regulations to provide for a different one.

The judgment of the Court of Appeals is reversed, and the case is remanded for proceedings consistent with this opinion.

It is so ordered.

SUPREME COURT OF THE

UNITED STATES

No. 00–157

FOURTH CIRCUIT

JUSTICE THOMAS, concurring. I agree with the Court that the Internal Revenue Code provision and the corresponding Treasury Regulations that control consolidated filings are best interpreted as requiring a single-entity approach in calculating product liability loss. I write separately, however, because I respectfully disagree with the dissent’s suggestion that, when a provision of the Code and the corresponding regulations are ambiguous, this Court should defer to the Government’s interpretation. See post, at 1–2. At a bare minimum, in cases such as this one, in which the complex statutory and regulatory scheme lends itself to any number of interpretations, we should be inclined to rely on the traditional canon that construes revenue-raising laws against their drafter. See Leavell v. Blades, 237 Mo. 695, 700–701, 141 S.W. 893, 894 (1911) (“When the tax gatherer puts his finger on the citizen, he must also put his finger on the law permitting it”); United States v. Merriam, 263 U.S. 179, 188 (1923) (“If the words are doubtful, the doubt must be resolved against the Government and in favor of the taxpayer”); Bowers v. New York & Albany Literage Co., 273 U.S. 346, 350 (1927) (“The provision is part of a taxing statute; and such laws are to be interpreted liberally in favor of the taxpayers”). Accord American Net & Twine Co. v. Worthing- ton, 141 U.S. 468, 474 (1891); Benziger v. United States, 192 U.S. 38, 55 (1904).

SUPREME COURT OF THE

UNITED STATES

No. 00–157

UNITED DOMINION INDUSTRIES,

INC., PETITIONER v . UNITED

STATES

ON WRIT OF CERTIORARI TO THE

UNITED STATES COURT OF

APPEALS FOR THE

UNITED DOMINION INDUSTRIES,

tential for abuse created by the petitioner’s reading of the statutory scheme and affirm the decision of the Court of Appeals on that basis. ¹

As the majority accurately reports, during the time relevant to this case, Sec. 172(b)(1)(I) of the Internal Revenue Code of 1954 allowed any “taxpayer” who “ha[d] a product liability loss” to carry back its excess product liability losses for 10 years. The resolution of this case turns on whether, when a group of affiliated corporations files a consolidated tax return, the entire group should be considered the “taxpayer” for the purposes of implementing this provision or whether each individual corporation should be seen as a “taxpayer.”

There is no obvious answer to this question. On the one hand, it is generally accepted that the rationale behind the consolidated return regulations is to allow affiliated corporations that are run as a single-entity to elect to be treated for tax purposes as a single-entity. See, e.g., Brief for Petitioner 17–19 (collecting sources in which the Internal Revenue Service so stated). On the other hand, it is quite clear that each corporation in such a group remains in both a legal and a literal sense a “taxpayer,” a status that has important consequences. See Woolford Realty Co. v. Rose, 286 U.S. 319, 328 (1932) (“The fact is not to be ignored that each of two or more corporations joining . . . in a consolidated return is none the less a taxpayer”); 26 U.S.C. Sec. 7701(a)(14) (defining a “taxpayer” as “any person subject to any internal revenue tax,” where a related provision defines “person” to include corporations). As both the group and the individual corpora

1 JUSTICE THOMAS accurately points to a tradition of cases construing “revenue-raising laws” against their drafter. See ante, at 1 (THOMAS, J., concurring). However, when the ambiguous provision in question is not one that imposes tax liability but rather one that crafts an exception from a general revenue duty for the benefit of some taxpayers, a countervailing tradition suggests that the ambiguity should be resolved in the government’s favor. See, e.g., INDOPCO, Inc. v. Commissioner, 503 U.S. 79, 84 (1992); Interstate Transit Lines v. Commissioner, 319 U.S. 590, 593 (1943); Deputy v. Du Pont, 308 U.S. 488, 493 (1940); New Colonial Ice Co. v. Helvering, 292 U.S. 435, 440 (1934); Woolford Realty Co. v. Rose, 286 U.S. 319, 326 (1932).

INC., PETITIONER v . UNITED

STATES

ON WRIT OF CERTIORARI TO THE

June 4, 2001

JUSTICE STEVENS, dissenting. This is a close and difficult case, in which neither the statute nor the regulations offer a definitive answer to the crucial textual question. Absent a clear textual anchor, I would credit the Secretary of the Treasury’s concerns about the po

UNITED STATES COURT OF

APPEALS FOR THE

FOURTH CIRCUIT

June 4, 2001

October 29, 2001 384 2001–44 I.R.B.

tions are considered “taxpayers” in different contexts, the statute presents a genuine ambiguity.

When a provision of the Internal Revenue Code presents a patent ambiguity, Congress, the courts, and the IRS share a preference for resolving the ambiguity via executive action. See, e.g., National Muffler Dealers Assn., Inc. v. United States, 440 U.S. 472, 477 (1979). This is best achieved by the issuing of a Treasury Regulation resolving the ambiguity. Ibid . In this instance, however, the Secretary of the Treasury issued no such regulation. In the absence of such a regulation, the majority has scoured tangentially related regulations, looking for clues to what the Secretary might intend. For want of a more precise basis for resolving this case, that approach is sound.

It is at this point, however, that I part company with the majority’s analysis. The fact that the regulations forward a particular method for calculating a consolidated “net operating loss” (NOL) for a group of affiliated companies, see Treas. Reg. Sec. 1.1502–21(f), tells us how the Secretary wants the NOL to be calculated whenever it is necessary to determine a consolidated NOL, but it does not tell us what provisions of the Code require the calculation of a consolidated NOL. That is a separate and prior question. Even if we were to draw some mild significance from the presence of such a regulation (and the absence, at the time these returns were filed, of a similar regulation for the calculation of corporation-specific NOL’s), the power of that inference is counterbalanced by the fact that the regulations listing deductions that must be reported at the consolidated level makes no mention of product liability expenses. See Treas. Reg. Sec. 1.1502–12; see also H. Enter- prises Int’l, Inc. v. Commissioner, 105 T.C. 71, 85 (1995) (construing Treas. Reg. Sec. 1.1502–80(a) to provide “[w]here the consolidated return regulations do not require that corporations filing such returns be treated differently from the way separate entities would be treated, those corporations shall be treated as separate entities when applying provisions of the Code”). In addition, the subsequent promulgation of a method for calculating a corporation

specific NOL (albeit for a different purpose), see Sec. 1.1502–79(a)(3) (defining “separate net operating loss”), demonstrates that there are no inherent problems implicit in undertaking such a calculation.

In short, I find no answer to this case in the text of the statute or in any Treasury Regulation. ² However, the government does forward a valid policy concern that militates against petitioner’s construction of the statute: the fear of tax abuse. See Brief for United States 40–42. Put simply, the Government fears that currently unprofitable but previously profitable corporations might receive a substantial windfall simply by acquiring a corporation with significant product liability expenses but no product liability losses. See id ., at 40. On a subjective level, I find these concerns troubling. Cf. Woolford Realty Co., 286 U.S., at 330 (rejecting “the notion that Congress in permitting a consolidated return was willing to foster an opportunity for juggling so facile and so obvious”). More importantly, however, I credit the Secretary of the Treasury’s concerns about the potential scope of abuse. Perhaps the Court is correct in suggesting that these concerns can be alleviated through applications of other anti-abuse provisions of the Tax Code, see ante, at 15, but I am not persuaded of my own ability to make that judgment. When we deal “with a subject that is highly specialized and so complex as to be the despair of judges,” Dobson v. Commissioner, 320 U.S. 489, 498 (1943), an ounce of deference is appropriate.

I respectfully dissent. ³

2 I am also in full agreement with the Court’s rejection of the Government’s double-deduction argument. See ante, at 11-12. 3 Because I agree with the majority that the calculation contemplated by Treas. Reg. Sec. 1.150279(a)(3) better approximates the NOL that each company would have had reported if filing individually than the alternative forwarded by the Government, see ante, at 10, I agree with the Court of Appeals’ decision to adopt that measure and would affirm the decision below in its entirety.

Section 3101.—Rate of Tax

Ct. D. 2071

SUPREME COURT OF THE

UNITED STATES

No. 99–1978

UNITED STATES v . HATTER, JUDGE,

UNITED STATES DISTRICT COURT

FOR THE CENTRAL DISTRICT OF

CALIFORNIA, ET AL.

CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR

THE FEDERAL CIRCUIT

May 21, 2001

Syllabus

In 1982, Congress extended Medicare to federal employees. That new law meant, inter alia, that then-sitting federal judges, like all other federal employees and most other citizens, began to have Medicare taxes withheld from their salaries. In 1983, Congress required all newly hired federal employees to participate in Social Security and permitted, without requiring, about 96% of the thencurrently employed federal employees to participate in that program. The remaining 4%—a class consisting of the President, other high-level Government employees, and all federal judges—were required to participate, except that those who contributed to a “covered” retirement program could modify their participation in a manner that left their total payroll deduction for retirement and Social Security unchanged, in effect allowing them to avoid any additional financial obligation as a result of joining Social Security. A “covered” program was defined to include any retirement system to which an employee had to contribute, which did not encompass the noncontributory pension system for federal judges, whose financial obligations (and payroll deductions) therefore had to increase. A number of federal judges appointed before 1983 filed this suit, arguing that the 1983 law violated the Compensation Clause, which guarantees federal judges a “Compensation, which shall not be diminished during their Continuance in Office,” U.S. Const., Art. III, Sec. 1. Initially, the Court of Federal Claims ruled against the judges, but the Federal Circuit reversed. On cer

2001–44 I.R.B. 385 October 29, 2001

tiorari, because some Justices were disqualified and this Court failed to find a quorum, the Federal Circuit’s judgment was affirmed “with the same effect as upon affirmance by an equally divided court.” 519 U.S. 801. On remand, the Court of Federal Claims found that the judges’ Medicare claims were time barred and that a 1984 judicial salary increase promptly cured any violation, making damages minimal. The Federal Circuit reversed, holding that the Compensation Clause prevented the Government from collecting Medicare and Social Security taxes from the judges and that the violation was not cured by the 1984 pay increase.

Held:

The Compensation Clause prevents the Government from collecting Social Security taxes, but not Medicare taxes, from federal judges who held office before Congress extended those taxes to federal employees. Pp. 6–19.

(a) The Court rejects the judges’ claim that the “law of the case” doctrine now prevents consideration of the Compensation Clause because an affirmance by an equally divided Court is conclusive and binding upon the parties. United States v. Pink, 315 U.S. 203, 216, on which the judges rely, concerned an earlier case in which the Court heard oral argument and apparently considered the merits before affirming by an equally divided Court. The law of the case doctrine presumes a hearing on the merits. See, e.g., Quern v. Jordan, 440 U.S. 332, 347, n. 18. When this case previously was here, due to absence of a quorum, the Court could not consider either the merits or whether to consider those merits through a grant of certiorari. This fact, along with the obvious difficulty of finding other equivalent substitute forums, convinces the Court that Pink does not control here. Pp. 6–7.

(b) Although the Compensation Clause prohibits taxation that singles out judges for specially unfavorable treatment, it does not forbid Congress to enact a law imposing a nondiscriminatory tax (including an increase in rates or a change in conditions) upon judges and other citizens. See O’Malley v. Woodrough, 307 U.S. 277, 282. Insofar as Evans v. Gore, 253 U.S. 245, 255, holds to the contrary, that case is overruled. See O’Malley,

supra, at 283. There is no good reason why a judge should not share the tax burdens borne by all citizens. See Evans, supra, at 265, 267 (Holmes, J., dissenting); O’Malley, supra, at 281–283. Although Congress cannot directly reduce judicial salaries even as part of an equitable effort to reduce all Government salaries, a tax law, unlike a law mandating a salary reduction, affects compensation indirectly, not directly. See United States v. Will, 449 U.S. 200, 226. And those prophylactic considerations that may justify an absolute rule forbidding direct salary reductions are absent here, where indirect taxation is at issue. In practice, the likelihood that a nondiscriminatory tax represents a disguised legislative effort to influence the judicial will is virtually nonexistent. Hence the potential threats to judicial independence that underlie the Compensation Clause, see Evans, supra, at 251–252, cannot justify a special judicial exemption from a commonly shared tax, not even as a preventive measure to counter those threats. Because the Medicare tax is nondiscriminatory, the Federal Circuit erred in finding its application to federal judges unconstitutional. Pp. 7–13.

(c) However, because the special retroactivity-related Social Security rules enacted in 1983 effectively singled out then-sitting federal judges for unfavorable treatment, the Compensation Clause forbids the application of the Social Security tax to those judges. Four features of the law, taken together, lead to the conclusion that it discriminates in a manner the Clause forbids. First, the statutory history, context, purpose, and language indicate that the category of “federal employees” is the appropriate class against which the asserted discrimination must be measured. Second, the practical upshot of defining “covered” system in the way the law did was to permit nearly every thencurrent federal employee, but not federal judges, to avoid the newly imposed obligation to pay Social Security taxes. Third, the new law imposed a substantial cost on federal judges with little or no expectation of substantial benefit for most of them. Inclusion meant a deduction of about $2,000 per year, whereas 95% of the thenactive judges had already qualified for Social Security (due to private sector employment) before becoming judges. And

participation would benefit only the minority of judges who had not worked the quarters necessary to be fully insured under Social Security. Fourth, the Government’s sole justification for the statutory distinction between judges and other high-level federal employees— i.e., equalizing the financial burdens imposed by the noncontributory judicial retirement system and the contributory system to which the other employees belonged—is unsound because such equalization takes place not by offering all current federal employees (including judges) the same opportunities, but by employing a statutory disadvantage which offsets an advantage related to those protections afforded judges by the Clause, and because the two systems are not equalized with any precision. Thus, the 1983 law is very different from the nondiscriminatory tax upheld in O’Malley, supra, at 282. The Government’s additional arguments—that Article III protects judges only against a reduction in stated salary, not against indirect measures that only reduce take-home pay; that there is no evidence here that Congress singled out judges for special treatment in order to intimidate, influence, or punish them; and that the law disfavored not only judges but also the President and other high-ranking federal employees— are unconvincing. Pp. 13–19.

The Compensation Clause violation was not cured by the 1984 pay increase for federal judges. The context in which that increase took place reveals nothing to suggest that it was intended to make whole the losses sustained by the pre1983 judges. Rather, everything in the record suggests that the increase was meant to halt a slide in purchasing power resulting from continued and unadjustedfor inflation. Although a circumstancespecific approach is more complex than the Government’s proposed automatic approach, whereby a later salary increase would terminate a Compensation Clause violation regardless of the increase’s purpose, there is no reason why such relief as damages or an exemption from Social Security would prove unworkable. Will, supra, distinguished. Pp. 19–22.

203 F.3d 795, affirmed in part, reversed in part, and remanded.

BREYER, J., delivered the opinion of the Court, in which REHNQUIST, C.J., and KENNEDY, SOUTER, and GINS

October 29, 2001 386 2001–44 I.R.B.

BURG, JJ., joined, and in which SCALIA, J., joined as to Parts I, II, and V. SCALIA, J., filed an opinion concurring in part and dissenting in part. THOMAS, J., filed an opinion concurring in the judgment in part and dissenting in part. STEVENS, J., and O’CONNOR, J., took no part in the consideration or decision of the case.

SUPREME COURT OF THE

UNITED STATES

No. 99–1978

Social Security Reform 2–1, 2–7 (Jan. 1983). In particular, the Commission recommended that Congress require all incoming federal employees (those hired after January 1, 1984) to enter the Social Security system and to pay Social Security taxes. Id ., at 2–7. The Commission emphasized that “present Federal employees will not be affected by this recommendation.” Id ., at 2–8.

In 1983, Congress enacted the Commission’s recommendation into law (effective January 1, 1984) with an important exception. See Social Security Amendments of 1983, Sec. 101(b)(1), 97 Stat. 69 (amending 26 U.S.C. Sections 3121(b)(5), (6)). As the Commission had recommended, Congress required all newly hired federal employees to participate in the Social Security program. It also permitted, without requiring, almost all (about 96%) then-currently employed federal employees to participate.

Contrary to the Commission’s recommendation, however, the law added an exception. That exception seemed to restrict the freedom of choice of the remaining 4% of all current employees. This class consisted of the President, Vice President, high-level Executive Branch employees, Members of Congress, a few other Legislative Branch employees, and all federal judges. See 42 U.S.C. Sections 410(a)(5)(C)–(G); see also H.R. Rep. No. 98–25, p. 39 (1983); H.R. Conf. Rep. No. 98–542, p. 13 (1983) (noting that for these current federal employees “the rules are being changed in the middle of the game”). The new law seemed to require this class of current federal employees to enter into the Social Security program, see 42 U.S.C. Sections 410(a)(5)(C)–(G). But, as to almost all of these employees, the new law imposed no additional financial obligation or burden.

That is because the new law then created an exception to the exception, see Federal Employees’ Retirement Contribution Temporary Adjustment Act of 1983, Secs. 203(a)(2), 208, 97 Stat. 1107, 1111 (codified at note following 5 U.S.C. Sec. 8331). The exception to the exception said that any member of this small class of current high-level officials (4% of all then-current employees) who contributed to a “covered” retirement program nonetheless could choose to modify their participation in a manner that left their

UNITED STATES, PETITIONER v .

TERRY J. HATTER, JR., JUDGE, UNITED STATES DISTRICT COURT

FOR THE CENTRAL DISTRICT OF

CALIFORNIA, ET AL.

532 U.S. ___(2001)

ON WRIT OF CERTIORARI TO THE

UNITED STATES COURT OF APPEALS FOR THE FEDERAL

CIRCUIT

May 21, 2001

JUSTICE BREYER delivered the opinion of the Court.

The Constitution’s Compensation Clause guarantees federal judges a “Compensation, which shall not be diminished during their Continuance in Office.” U.S. Const., Art. III, Sec. 1. The Court of Appeals for the Federal Circuit held that this Clause prevents the Government from collecting certain Medicare and Social Security taxes from a small number of federal judges who held office nearly 20 years ago—before Congress extended the taxes to federal employees in the early 1980’s. In our view, the Clause does not prevent Congress from imposing a “non-discriminatory tax laid generally” upon judges and other citizens, O’Malley v. Woodrough, 307 U.S. 277, 282 (1939), but it does prohibit taxation that singles out judges for specially unfavorable treatment. Consequently, unlike the Court of Appeals, we conclude that Congress may apply the Medicare tax—a nondiscriminatory tax—to then-sitting federal judges. The special retroactivity-related Social Security rules that Congress enacted in 1984, however, effectively singled out then-sitting federal judges for unfavorable treatment. Hence, like the Court of Ap

peals, we conclude that the Clause forbids the application of the Social Security tax to those judges.

The Medicare law before us is straightforward. In 1965, Congress created a Federal Medicare “hospital insurance” program and tied its financing to Social Security. See Social Security Amendments of 1965, 79 Stat. 291. The Medicare law required most American workers (whom Social Security covered) to pay an additional Medicare tax. But it did not require Federal Government employees (whom Social Security did not cover) to pay that tax. See 26 U.S.C. Sections 3121(b)(5), (6) (1982 ed.).

In 1982, Congress, believing that “[f]ederal workers should bear a more equitable share of the costs of financing the benefits to which many of them eventually became entitled,” S. Rep. No. 97–494, pt. 1, p. 378 (1982), extended both Medicare eligibility and Medicare taxes to all currently employed federal employees as well as to all newly hired federal employees, Tax Equity and Fiscal Responsibility Act of 1982, Sec. 278, 96 Stat. 559–563. That new law meant that (as of January 1, 1983) all federal judges, like all other federal employees and most other citizens, would have to contribute between 1.30% and 1.45% of their federal salaries to Medicare’s hospital insurance system. See 26 U.S.C. Sections 3101(b)(4)–(6). The Social Security law before us is more complex. In 1935, Congress created the Social Security program. See Social Security Act, 49 Stat. 620. For nearly 50 years, that program covered employees in the private sector, but it did not cover Government employees. See 26 U.S.C. Sections 3121(b)(5), (6) (1982 ed.) (excluding federal employees); Sec. 3121(b)(7) (excluding state employees). In 1981, a National Commission on Social Security Reform, convened by the President and chaired by Alan Greenspan, noting the need for “action . . . to strengthen the financial status” of Social Security, recommended that Congress extend the program to cover Federal, but not state or local, Government employees. Report of the National Commission on

2001–44 I.R.B. 387 October 29, 2001

total payroll deduction—for retirement and Social Security—unchanged. A “covered” employee paying 7% of salary to a “covered” program could continue to pay that 7% and no more, in effect avoiding any additional financial obligation as a result of joining Social Security.

The exception to the exception defined a “covered” program to include the Civil Service Retirement and Disability System—a program long available to almost all federal employees—as well as any other retirement system to which an employee must contribute. Secs. 203(a)(2)(A), (D). The definition of “covered” program, however, did not encompass the pension system for federal judges—a system that is noncontributory in respect to a judge (but contributory in respect to a spouse).

The upshot is that the 1983 law was specifically aimed at extending Social Security to federal employees. It left about 96% of those who were currently employed free to choose not to participate in Social Security, thereby avoiding any increased financial obligation. It required the remaining 4% to participate in Social Security while freeing them of any added financial obligation (or additional payroll deduction) so long as they previously had participated in other contributory retirement programs. But it left those who could not participate in a contributory program without a choice. Their financial obligations (and payroll deductions) had to increase. And this last mentioned group consisted almost exclusively of federal judges.

This litigation began in 1989, when eight federal judges, all appointed before 1983, sued the Government for “compensation” in the United States Claims Court. They argued that the 1983 law, in requiring them to pay Social Security taxes, violated the Compensation Clause. Initially, the Claims Court ruled against the judges on jurisdictional grounds. 21 Cl. Ct. 786 (1990). The Court of Appeals reversed. 953 F.2d 626 (CA Fed. 1992). On remand, eight more judges joined the lawsuit. They contested the extension to judges of the Medicare tax as well.

The Court of Federal Claims held against the judges on the merits. 31 Fed. Cl. 436 (1994). The Federal Circuit re

versed, ordering summary judgment for the judges as to liability. 64 F.3d 647 (1995). The Government petitioned this Court for writ of certiorari. Some Members of this Court were disqualified from hearing the matter, and we failed to find a quorum of six Justices. See 28 U.S.C. Sec. 1. Consequently, the Court of Appeals’ judgment was affirmed “with the same effect as upon affirmance by an equally divided court.” 519 U.S. 801 (1996); see 28 U.S.C. Sec. 2109.

On remand from the Court of Appeals, the Court of Federal Claims found (a) that the 6-year statute of limitations, see 28 U.S.C. Sections 2401(a), 2501, barred some claims, including all Medicare claims; and (b) that, in any event, a subsequently enacted judicial salary increase promptly cured any violation, making damages minimal. 38 Fed. Cl. 166 (1997). The Court of Appeals (eventually en banc) reversed both determinations. 203 F.3d 795 (CA Fed. 2000). The Government again petitioned for certiorari. It asked this Court to consider two questions:

(1) Whether Congress violated the Compensation Clause when it extended the Medicare and Social Security taxes to the salaries of sitting federal judges; and

(2) If so, whether any such violation ended when Congress subsequently increased the salaries of all federal judges by an amount greater than the new taxes.

Given the specific statutory provisions at issue and the passage of time, seven Members of this Court had (and now have) no financial stake in the outcome of this case. Consequently a quorum was, and is, available to consider the questions presented. And we granted the Government’s petition for writ of certiorari.

At the outset, the judges claim that the “law of the case” doctrine prevents us from now considering the first question presented, namely, the scope of the Compensation Clause. They note that the Government presented that same question in its petition from the Court of Appeals’ earlier ruling on liability. They point out that our earlier denial of that petition for lack of a quorum had the “same effect as” an “affirmance by an equally divided court,” 28 U.S.C. Sec. 2109. And they add that this Court has said that an affir

mance by an equally divided Court is “conclusive and binding upon the parties as respects that controversy.” United States v. Pink, 315 U.S. 203, 216 (1942).

Pink, however, concerned a case, United States v. Moscow Fire Ins. Co., 309 U.S. 624 (1940), in which this Court had heard oral argument and apparently considered the merits prior to concluding that affirmance by an equally divided Court was appropriate. The law of the case doctrine presumes a hearing on the merits. See, e.g., Quern v. Jordan, 440 U.S. 332, 347, n. 18 (1979). This case does not involve a previous consideration of the merits. Indeed, when this case previously was before us, due to absence of a quorum, we could not consider either the merits or whether to consider those merits through grant of a writ of certiorari. This fact, along with the obvious difficulty of finding other equivalent substitute forums, convinces us that Pink’s statement does not control the outcome here, that the “law of the case” doctrine does not prevent our considering both issues presented, and that we should now proceed to decide them.

III

The Court of Appeals upheld the judges’ claim of tax immunity upon the authority of Evans v. Gore, 253 U.S. 245 (1920). That case arose in 1919, when Judge Walter Evans challenged Congress’ authority to include sitting federal judges within the scope of a federal income tax law that the Sixteenth Amendment had authorized a few years earlier. See Revenue Act of 1918, Sec. 213, 40 Stat. 1065 (defining “gross income” to include judicial salaries). In Evans itself, the Court held that the Compensation Clause barred application of the tax to Evans, who had been appointed a judge before Congress enacted the tax. 253 U.S., at 264. A few years later, the Court extended Evans, making clear that its rationale covered not only judges appointed before Congress enacted a tax but also judges whose appointments took place after the tax had become law. See Miles v. Graham, 268 U.S. 501, 509 (1925).

Fourteen years after deciding Miles, this Court overruled Miles . O’Malley v. Woodrough, 307 U.S. 277 (1939). But, as the Court of Appeals noted, this Court did not expressly overrule Evans itself.

October 29, 2001 388 2001–44 I.R.B.

64 F.3d, at 650. The Court of Appeals added that if “changes in judicial doctrine” had significantly undermined Evans ’ holding, this “Court itself would have overruled the case.” Ibid . Noting that this case is like Evans (involving judges appointed before enactment of the tax), not like O’Malley (involving judges appointed after enactment of the tax), the Court of Appeals held that Evans controlled the outcome. 64 F.3d, at 650. Hence application of both Medicare and Social Security taxes to these pre-enactment judges violated the Compensation Clause.

The Court of Appeals was correct in applying Evans to the instant case, given that “it is this Court’s prerogative alone to overrule one of its precedents.” State Oil Co. v. Khan, 522 U.S. 3, 20 (1997); see also Rodriguez de Quijas v. Shearson/American Express, Inc., 490 U.S. 477, 484 (1989). Nonetheless, the court below, in effect, has invited us to reconsider Evans . We now overrule Evans insofar as it holds that the Compensation Clause forbids Congress to apply a generally applicable, nondiscriminatory tax to the salaries of federal judges, whether or not they were appointed before enactment of the tax.

The Court’s opinion in Evans began by explaining why the Compensation Clause is constitutionally important, and we begin by reaffirming that explanation. As Evans points out, 253 U.S., at 251–252, the Compensation Clause, along with the Clause securing federal judges appointments “during good Behavior,” U.S. Const., Art. III, Sec. 1—the practical equivalent of life tenure—helps to guarantee what Alexander Hamilton called the “complete independence of the courts of justice.” The Federalist No. 78, p. 466 (C. Rossiter ed. 1961). Hamilton thought these guarantees necessary because the Judiciary is “beyond comparison the weakest of the three” branches of government. Id ., at 465–466. It has “no influence over either the sword or the purse.” Id ., at 465. It has “no direction either of the strength or of the wealth of the society.” Ibid . It has “neither FORCE nor WILL but merely judgment.” Ibid .

Hamilton’s view, and that of many other Founders, was informed by firsthand experience of the harmful consequences brought about when a King of

England “made Judges dependent on his Will alone, for the tenure of their offices, and the amount and payment of their salaries.” The Declaration of Independence, Par. 11. And Hamilton knew that “ a power over a man’s subsistence amounts to a power over his will .” The Federalist No. 79, at 472. For this reason, he observed, “[n]ext to permanency in office, nothing can contribute more to the independence of the judges than a fixed provision for their support.” Ibid .; see also id ., No. 48 at 310 (J. Madison) (“[A]s the legislative department alone has access to the pockets of the people, and has . . . full discretion . . . over the pecuniary rewards of those who fill the other departments, a dependence is thus created in the latter, which gives still greater facility to encroachments of the former”).

Evans properly added that these guarantees of compensation and life tenure exist, “not to benefit the judges,” but “as a limitation imposed in the public interest.” 253 U.S., at 253. They “promote the public weal,” id ., at 248, in part by helping to induce “learned” men and women “to quit the lucrative pursuits” of the private sector, 1 J. Kent, Commentaries on American Law *294, but more importantly by helping to secure an independence of mind and spirit necessary if judges are “to maintain that nice adjustment between individual rights and governmental powers which constitutes political liberty,” W. Wilson, Constitutional Government in the United States 143 (1911).

Chief Justice John Marshall pointed out why this protection is important. A judge may have to decide “between the Government and the man whom that Government is prosecuting: between the most powerful individual in the community, and the poorest and most unpopular.” Proceedings and Debates of the Virginia State Convention, of 1829–1830, p. 616 (1830). A judge’s decision may affect an individual’s “property, his reputation, his life, his all.” Ibid . In the “exercise of these duties,” the judge must “observe the utmost fairness.” Ibid . The judge must be “perfectly and completely independent, with nothing to influence or contro[l] him but God and his conscience.” Ibid . The “greatest scourge . . . ever inflicted,” Marshall thought, “was an ignorant, a corrupt, or a dependent Judiciary.” Id ., at 619.

Those who founded the Republic recognized the importance of these constitutional principles. See, e.g., Wilson, Lectures on Law (1791), in 1 Works of James Wilson 363 (J. Andrews ed. 1896); (stating that judges should be “completely independent” in “their salaries, and in their offices”); McKean, Debate in Pennsylvania Ratifying Convention, Dec. 11, 1787, in 2 Debates on the Federal Constitution 539 (J. Elliot ed. 1836) (the security of undiminished compensation disposes judges to be “more easy and independent”); see also 1 Kent, supra, at *294 (“permanent support” and the “tenure of their office” “is well calculated . . . to give

[judges] the requisite independence”). They are no less important today than in earlier times. And the fact that we overrule Evans does not, in our view, diminish their importance.

We also agree with Evans insofar as it holds that the Compensation Clause offers protections that extend beyond a legislative effort directly to diminish a judge’s pay, say by ordering a lower salary. 253 U.S., at 254. Otherwise a legislature could circumvent even the most basic Compensation Clause protection by enacting a discriminatory tax law, for example, that precisely but indirectly achieved the forbidden effect.

Nonetheless, we disagree with Evans’ application of Compensation Clause principles to the matter before it—a nondiscriminatory tax that treated judges the same way it treated other citizens. Evans ’ basic holding was that the Compensation Clause forbids such a tax because the Clause forbids “all diminution,” including “taxation,” “whether for one purpose or another.” Id., at 255. The Federal Circuit relied upon this holding. 64 F.3d, at 650. But, in our view, it is no longer sound law.

For one thing, the dissenters in Evans cast the majority’s reasoning into doubt. Justice Holmes, joined by Justice Brandeis, wrote that the Compensation Clause offers “no reason for exonerating” a judge “from the ordinary duties of a citizen, which he shares with all others. To require a man to pay the taxes that all other men have to pay cannot possibly be made an instrument to attack his independence as a judge.” Evans, 253 U.S., at 265. Holmes analogized the “diminution” that a tax might bring about to the burden that a state law might impose upon interstate

2001–44 I.R.B. 389 October 29, 2001

commerce. If “there was no discrimination against such commerce, the tax constituted one of the ordinary burdens of government from which parties were not exempted.” Id ., at 267.

For another thing, this Court’s subsequent law repudiated Evans ’ reasoning. In 1939, 14 years after Miles extended Evans ’ tax immunity to judges appointed after enactment of the tax, this Court retreated from that extension. See O’Mal- ley, 307 U.S., at 283 (overruling Miles ). And in so doing the Court, in an opinion announced by Justice Frankfurter, adopted the reasoning of the Evans dissent. The Court said that the question was whether judges are immune “from the incidences of taxation to which everyone else within the defined classes . . . is subjected.” Id ., at 282. Holding that judges are not “immun[e] from sharing with their fellow citizens the material burden of the government,” ibid ., the Court pointed out that the legal profession had criticized Evans ’ contrary conclusion, and that courts outside the United States had resolved similar matters differently, id ., at 281. And the Court concluded that “a nondiscriminatory tax laid generally on net income is not, when applied to the income of a federal judge, a diminution of his salary within the prohibition of Article III.” Id ., at 282. The Court conceded that Miles had reached the opposite conclusion, but it said that Miles “cannot survive.” 307 U.S., at 283. Still later, this Court noted that “[b]ecause Miles relied on Evans v. Gore, O’Malley must also be read to undermine the reasoning of Evans .” United States v. Will, 449 U.S. 200, 227, n. 31 (1980). Finally, and most importantly, we believe that the reasoning of Justices Holmes and Brandeis, and of this Court in O’Malley, is correct. There is no good reason why a judge should not share the tax burdens borne by all citizens. We concede that this Court has held that the Legislature cannot directly reduce judicial salaries even as part of an equitable effort to reduce all Government salaries. See 449 U.S., at 226. But a tax law, unlike a law mandating a salary reduction, affects compensation indirectly, not directly. See ibid . (distinguishing between measures that directly and those that indirectly diminish judicial compensation). And those prophylactic considerations that may jus

tify an absolute rule forbidding direct salary reductions are absent here, where indirect taxation is at issue. In practice, the likelihood that a nondiscriminatory tax represents a disguised legislative effort to influence the judicial will is virtually nonexistent. Hence, the potential threats to judicial independence that underlie the Constitution’s compensation guarantee cannot justify a special judicial exemption from a commonly shared tax, not even as a preventive measure to counter those threats.

For these reasons, we hold that the Compensation Clause does not forbid Congress to enact a law imposing a nondiscriminatory tax (including an increase in rates or a change in conditions) upon judges, whether those judges were appointed before or after the tax law in question was enacted or took effect. Insofar as Evans holds to the contrary, that case, in O’Malley’s words, “cannot survive.” 307 U.S., at 283.

The Government points out that the Medicare tax is just such a nondiscriminatory tax. Neither the courts below, nor the federal judges here, argue to the contrary. Hence, insofar as the Court of Appeals found that application of the Medicare tax law to federal judges is unconstitutional, we reverse its decision.

The Social Security tax is a different matter. Respondents argue that the 1983 law imposing that tax upon then-sitting judges violates the Compensation Clause, for it discriminates against judges in a manner forbidden by the Clause, even as interpreted in O’Malley, not Evans . Cf. O’Malley, supra, at 282 (stating question as whether judges are immune “from the incidences of taxation to which everyone else within the defined classes . . . is subjected” (emphasis added)). After examining the statute’s details, we agree with the judges that it does discriminate in a manner that the Clause forbids. Four features of the law, taken together, lead us to this conclusion.

First, federal employees had remained outside the Social Security system for nearly 50 years prior to the passage of the 1983 law. Congress enacted the law pursuant to the Social Security Commission’s recommendation to bring those employees within the law. See supra, at 3. And

the law itself deals primarily with that subject. Thus, history, context, statutory purpose, and statutory language, taken together, indicate that the category of “federal employees” is the appropriate class against which we must measure the asserted discrimination.

Second, the law, as applied in practice, in effect imposed a new financial obligation upon sitting judges, but it did not impose a new financial burden upon any other group of (then) current federal employees. We have previously explained why that is so. See supra, at 3–5. The law required all newly hired federal employees to join Social Security and pay related taxes. It gave 96% of all current employees (employed as of January 1, 1984 or earlier) total freedom to enter, or not to enter, the system as they chose. It gave the remaining 4% of all current employees the freedom to maintain their pre1984 payroll deductions, provided that they were currently enrolled in a “covered” system. And it defined “covered” system in a way that included virtually all of that 4%, except for federal judges. See supra, at 4–5. The practical upshot is that the law permitted nearly every current federal employee, but not federal judges, to avoid the newly imposed financial obligation.

Third, the law, by including sitting judges in the system, adversely affected most of them. Inclusion meant a requirement to pay a tax of about $2,000 per year, deducted from a monthly salary check. App. 49. At the same time, 95% of the then-active judges had already qualified for Social Security (due to private sector employment) before becoming judges. See id ., at 115. And participation in Social Security as judges would benefit only a minority. See id ., at 116–119 (reviewing examples of individual judges and demonstrating that participation in Social Security primarily would benefit the minority of judges who had not worked the 40 quarters necessary to be fully insured). The new law imposed a substantial cost on federal judges with little or no expectation of substantial benefit for most of them.

Fourth, when measured against Compensation Clause objectives, the Government’s justification for the statutory distinction (between judges, who do, and other federal employees, who do not,

October 29, 2001 390 2001–44 I.R.B.

produce greater equality when applied to other, less typical examples.

Taken together, these four characteristics reveal a law that is special—in its manner of singling out judges for disadvantageous treatment, in its justification as necessary to offset advantages related to constitutionally protected features of the judicial office, and in the degree of permissible legislative discretion that would have to underlie any determination that the legislation has “equalized” rather than gone too far. For these reasons, the law before us is very different from the “nondiscriminatory” tax that O’Malley upheld. 307 U.S., at 282. Were the Compensation Clause to permit Congress to enact a discriminatory law with these features, it would authorize the Legislature to diminish, or to equalize away, those very characteristics of the Judicial Branch that Article III guarantees—characteristics which, as we have said, see supra, at 9–10, the public needs to secure that judicial independence upon which its rights depend. We consequently conclude that the 1983 Social Security tax law discriminates against the Judicial Branch, in violation of the Compensation Clause.

The Government makes additional arguments in support of reversal. But we find them unconvincing. It suggests that Article III protects judges only against a reduction in stated salary, not against indirect measures that only reduce takehome pay. Brief for United States 28. In O’Malley, however, this Court, when upholding a “nondiscriminatory” tax, strongly implied that the Compensation Clause would bar a discriminatory tax. 307 U.S., at 282. The commentators whose work O’Malley cited said so explicitly. See Fellman, The Diminution of Judicial Salaries, 24 Iowa L. Rev. 89, 99 (1938); see also Hall, Case Comment, 20 Ill. L. Rev. 376, 377 (1925); Corwin, Constitutional Law in 1919–1920, 14 Am. Pol. Sci. Rev. 635, 642 (1920). And in Will, the Court yet more strongly indicated that the Compensation Clause bars indirect efforts to reduce judges’ salaries through taxes when those taxes discriminate. 449 U.S., at 226. Indeed, the Government itself “assume[s] that discriminatory taxation of judges would contravene fundamental principles underlying Article III, if not the [Compensation] Clause itself.” Brief for United States 37, n. 27.

incur additional financial obligations) is unsound. The sole justification, according to the Government, is one of “equaliz[ing]” the retirement-related obligations that pre-1983 law imposed upon judges with the retirement-related obligations that pre-1983 law imposed upon other current high-level federal employees. Brief for United States 40. Thus, the Government says that the new financial burden imposed upon judges was meant to make up for the fact that the judicial retirement system is basically a noncontributory system, while the system to which other federal employees belonged was a contributory system. Id ., at 39–40; Reply Brief for United States 16. This rationale, however, is the Government’s and not necessarily that of Congress, which was silent on the matter. Cf. Motor Vehicle Mfrs. Assn. of United States, Inc. v. State Farm Mut. Automo- bile Ins. Co., 463 U.S. 29, 50 (1983) (expressing concern at crediting post hoc explanation of agency action).

More importantly, the judicial retirement system is noncontributory because it reflects the fact that the Constitution itself guarantees federal judges life tenure— thereby constitutionally permitting federal judges to draw a salary for life simply by continuing to serve. Cf. Booth v. United States, 291 U.S. 339, 352 (1934) (holding that Compensation Clause protects salary of judge who has retired). That fact means that a contributory system, in all likelihood, would not work. And, of course, as of 1982, the noncontributory pension salary benefits were themselves part of the judge’s compensation. The 1983 statute consequently singles out judges for adverse treatment solely because of a feature required by the Constitution to preserve judicial independence. At the same time, the “equaliz[ation]” in question takes place not by offering all current federal employees (including judges) the same opportunities but by employing a statutory disadvantage which offsets a constitutionally guaranteed advantage. Hence, to accept the “justification” offered here is to permit, through similar reasoning, taxes which have the effect of weakening or eliminating those constitutional guarantees necessary to secure judicial independence, at least insofar as similar guarantees are not enjoyed by others. This point

would be obvious were Congress, say, to deny some of the benefits of a tax reduction to those with constitutionally guaranteed life tenure to make up for the fact that other employees lack such tenure. Although the relationships here—among advantages and disadvantages—are less distant and more complex, the principle is similar.

Nor does the statute “equaliz[e]” with any precision. On the one hand, the thencurrent retirement system open to all federal employees except judges required a typical employee to contribute 7% to 8% of his or her annual salary. See generally 5 U.S.C. Sec. 8334(a)(1). In return it provided a Member of Congress, for instance, with a pension that vested after five years and increased in value (by 2.5% of the Member’s average salary) with each year of service to a maximum of 80% of salary, and covered both employee and survivors. See 5 U.S.C. Sections 8339, 8341. On the other hand, the judges’ retirement system (based on life tenure) required no contribution for a judge who retired at age 65 (and who met certain service requirements) to receive full salary. But the right to receive that salary did not vest until retirement. The system provided nothing for a judge who left office before age 65. Nor did the law provide any coverage for a judge’s survivors. Indeed, in 1984, a judge had to contribute 4.5% of annual salary to obtain a survivor’s annuity, which increased in value by 1.25% of the judge’s salary per year to a maximum of 40% of salary. 28 U.S.C. Sections 376(b), ( 1 ) (1982 ed.).

These two systems were not equal either before or after Congress enacted the 1983 law. Before 1983, a typical married federal employee other than a judge had to contribute 7 to 8% of annual salary to receive benefits that were better in some respects (vesting period, spousal benefit) and worse in some respects (80% salary maximum) than his married judicial counterpart would receive in return for a 4.5% contribution. The 1983 law imposed an added 5.7% burden upon the judge, in return for which the typical judge received little, or no, financial benefit. Viewed purely in financial equalization terms, and as applied to typical judges, the new requirement seems to over-equalize, putting the typical married judge at a financial disadvantage—though perhaps it would

2001–44 I.R.B. 391 October 29, 2001

ciently serious to erode the real value of judicial salaries and salary increases insufficient to maintain real salaries or real compensation parity with many other private-sector employees. See Report of 1989 Commission on Executive, Legislative, and Judicial Salaries, Hearings before the Senate Committee on Governmental Affairs, 101st Cong., 1st Sess., 12–13 (1989) (testimony of Lloyd Cutler regarding effect of inflation on judges’ salaries since 1969). For instance, while consumer prices rose 363% between 1969 and 1999, salaries in the private sector rose 421%, and salaries for district judges rose 253%. See American Bar Association, Federal Judicial Pay Erosion 11 (Feb. 2001). These figures strongly suggest that the judicial salary increases simply reflected a congressional effort to restore both to judges and to Members of Congress themselves some, but not all, of the real compensation that inflation had eroded. Those salary increases amounted to a congressional effort to adjust judicial salaries to reflect “fluctuations in the value of money,” The Federalist No. 79, at 473 (A. Hamilton)—the kind of adjustment that the Founders believed “may be requisite,” McKean, Debate in Pennsylvania Ratifying Convention, Dec. 11, 1787, in 2 Debates on the Federal Constitution, at 539; see also Rosenn, The Constitutional Guaranty Against Diminution of Judicial Compensation, 24 UCLA L. Rev. 308, 314–315 (1976). We have found nothing to the contrary. And, we therefore agree with the Court of Appeals’ similar conclusion. 185 F.3d at 1363 (“[E]verything in the record” suggests that the increase was meant to halt “the slide in purchasing power resulting from continued and unadjusted-for inflation”).

The Government says that a circumstance-specific approach may prove difficult to administer. Brief for United States 43. And we concede that examining the circumstances in order to determine whether there is or is not a relation between an earlier violation and a later increase is more complex than the Government’s proposed automatic approach. But we see no reason why such relief as damages or an exemption from Social Security would prove unworkable.

Finally, the Government looks to our decision in Will for support. In that case,

The Government also argues that there is no evidence here that Congress singled out judges for special treatment in order to intimidate, influence, or punish them. But, this Court has never insisted upon such evidence. To require it is to invite legislative efforts that embody, but lack evidence of, some such intent, engendering suspicion among the branches and consequently undermining that mutual respect that the Constitution demands. Cf. Wilson, Lectures on Law, in 1 Works of James Wilson, at 364 (stating that judges “should be removed from the most distant apprehension of being affected, in their judicial character and capacity, by anything, except their own behavior and its consequences”). Nothing in the record discloses anything other than benign congressional motives. If the Compensation Clause is to offer meaningful protection, however, we cannot limit that protection to instances in which the Legislature manifests, say, direct hostility to the Judiciary.

Finally, the Government correctly points out that the law disfavored not only judges but also the President of the United States and certain Legislative Branch employees. As far as we can determine, however, all Legislative Branch employees were free to join a covered system, and the record provides us with no example of any current Legislative Branch employee who had failed to do so. See Tr. of Oral Arg. 16–17, 37–38. The President’s pension is noncontributory. See note following 3 U.S.C. Sec. 102. And the President himself, like the judges, is protected against diminution in his “[c]ompensation.” See U.S. Const., Art. II, Sec. 1. These facts may help establish congressional good faith. But, as we have said, we do not doubt that good faith. And we do not see why, otherwise, the separate and special example of that single individual, the President, should make a critical difference here.

We conclude that, insofar as the 1983 statute required then-sitting judges to join the Social Security System and pay Social Security taxes, that statute violates the Compensation Clause.

The second question presented is whether the

“constitutional violation ended when Congress increased the statutory salaries of federal judges by an amount greater than the amount [of the Social Security] taxes deducted from respondents’ judicial salaries.” Pet. for Cert. (I).

The Government argues for an affirmative answer. It points to a statutory salary increase that all judges received in 1984. It says that this increase, subsequent to the imposition of Social Security taxes on judges’ salaries, cured any earlier unconstitutional diminution of salaries in a lesser amount. Otherwise, if “Congress improperly reduced judges salaries from $140,000” per year “to $130,000” per year, the judges would be able to collect the amount of the improper reduction, here $10,000, forever—even if Congress cured the improper reduction by raising salaries $20,000, to $150,000, a year later. Reply Brief for United States 18. To avoid this consequence, the Government argues, we should simply look to the fact of a later salary increase “whether or not one of Congress’s purposes in increasing the salaries” was “to terminate the constitutional violation.” Ibid .

But how could we always decide whether a later salary increase terminates a constitutional violation without examining the purpose of that increase? Imagine a violation that affected only a few. To accept the Government’s position, would leave those few at a permanent salary disadvantage. If, for example, Congress reduced the salaries of one group of judges by 20%, a later increase of 30% applicable to all judges would leave the first group permanently 20% behind. And a pay cut that left those judges at a permanent disadvantage, would perpetuate the very harm that the Compensation Clause seeks to prevent.

The Court of Appeals consequently examined the context in which the later pay increases took place in order to determine their relation to the earlier Compensation Clause violation. It found “nothing to suggest” that the later salary increase at issue here sought “to make whole the losses sustained by the pre-1983 judges.” 185 F.3d, at 1362–1363. The Government presents no evidence to the contrary.

The relevant economic circumstances surrounding the 1984, and subsequent, salary increases include inflation suffi

October 29, 2001 392 2001–44 I.R.B.

federal judges challenged the constitutionality of certain legislative “freezes” that Congress had imposed upon earlier enacted Government-wide cost-of-living salary adjustments. The Court found a Compensation Clause violation in respect to the freeze for what was designated Year One (where Congress had rescinded an earlier-voted 4.8% salary increase). Will, 449 U.S., at 225–226. The Government points out that the Will Court “noted that Congress, later in that fiscal year, enacted a statutory increase in judges’ salaries that exceeded the salaries that judges would have received” without the rescission. Brief for United States 41. And the Government adds that “it was unquestioned in Will ” that the judges could not receive damages for the time subsequent to this later enactment. Id ., at 41–42.

The Will Year One example, however, shows only that, in the circumstances, and unlike the case before us, the later salary increase was related to the earlier salary diminishment. Regardless, the very fact that the matter was “unquestioned” in Will shows that it was not argued. See 449 U.S., at 206, n. 3 (noting that the judges’ complaint sought relief for Year One’s diminution only up to the moment of the subsequent salary increase). Hence, the Court did not decide the matter now before us.

We conclude that later statutory salary increases did not cure the preceding unconstitutional harm.

Insofar as the Court of Appeals found the application of Medicare taxes to the salaries of judges taking office before 1983 unconstitutional, its judgment is reversed. Insofar as that court found the application of Social Security taxes to the salaries of judges taking office before 1984 unconstitutional, its judgment is affirmed. We also affirm the Court of Appeals’ determination that the 1984 salary increase received by federal judges did not cure the Compensation Clause violation. The case is remanded for further proceedings consistent with this opinion.

It is so ordered.

JUSTICE STEVENS and JUSTICE O’CONNOR took no part in the consideration or decision of this case.

SUPREME COURT OF THE

UNITED STATES

No. 99–1978

UNITED STATES, PETITIONER v .

TERRY J. HATTER, JR., JUDGE, UNITED STATES DISTRICT COURT

FOR THE CENTRAL DISTRICT OF

CALIFORNIA, ET AL.

ON WRIT OF CERTIORARI TO THE

UNITED STATES COURT OF

APPEALS FOR THE FEDERAL CIRCUIT

May 21, 2001

JUSTICE SCALIA, concurring in part and dissenting in part.

I agree with the Court that extending the Social Security tax to sitting Article III judges in 1984 violated Article III’s Compensation Clause. I part paths with the Court on the issue of extending the Medicare tax to federal judges in 1983, which I think was also unconstitutional. ¹

As an initial matter, I think the Court is right in concluding that Evans v. Gore, 253 U.S. 245 (1920)—holding that new taxes of general applicability cannot be applied to sitting Article III judges—is no longer good law, and should be overruled. We went out of our way in O’Malley v. Woodrough, 307 U.S. 277, 280–281 (1939), to catalog criticism of Evans, and subsequently recognized, in United States v. Will, 449 U.S. 200, 227, and n. 31 (1980), that O’Malley had “undermine[d] the reasoning of Evans .” The Court’s decision today simply recognizes what should be obvious: that Evans has not only been undermined, but has in fact collapsed.

My disagreement with the Court arises from its focus upon the issue of discrimination, which turns out to be dispositive with respect to the Medicare tax. The

1 I agree with the Court, see Part II, ante, that the law-of-the-case doctrine does not bar our consideration of the merits. I also join the Court in holding, see Part V, ante, that any constitutional violation was not remedied by subsequent salary increases.

Court holds “that the Compensation Clause does not forbid Congress to enact a law imposing a nondiscriminatory tax . . . upon judges, whether those judges were appointed before or after the tax law in question was enacted or took effect.” Ante, at 12. Since “the Medicare tax is just such a nondiscriminatory tax,” the Court concludes that “application of [that] tax law to federal judges is [c]onstitutional.” Ante, at 12–13.

But we are dealing here with a “Compensation Clause,” not a “Discrimination Clause.” See U.S. Const., Art III, Sec. 1 (“The Judges . . . shall, at stated Times, receive for their Services, a Compensation, which shall not be diminished during their Continuance in Office”). As we have said, “the Constitution makes no exceptions for ‘nondiscriminatory’ reductions” in judicial compensation, Will, supra, at 226. A reduction in compensation is a reduction in compensation, even if all federal employees are subjected to the same cut. The discrimination criterion that the Court uses would make sense if the only purpose of the Compensation Clause were to prevent invidious (and possibly coercive) action against judges. But as the Court acknowledges, the Clause “‘promote[s] the public weal’ . . . by helping to induce ‘learned’ men and women to ‘quit the lucrative pursuits’ of the private sector,” ante, at 9 (quoting Evans, supra, at 248; 1 J. Kent, Commentaries on American Law *294). That inducement would not exist if Congress could cut judicial salaries so long as it did not do so discriminatorily.

What the question comes down to, then, is (1) whether exemption from a certain tax can constitute part of a judge’s “compensation,” and (2) if so, whether exemption from the Medicare tax was part of the judges’ compensation here. The answer to the more general question seems to me obviously yes. Surely the term “compensation” refers to the entire “package” of benefits—not just cash, but retirement benefits, medical care, and ex- emption from taxation if that is part of the employment package . It is simply unreasonable to think than “$150,000 a year tax-free” (if that was the bargain struck) is not higher compensation than “$150,000 a year subject to taxes.” Ask the employees of the World Bank.

2001–44 I.R.B. 393 October 29, 2001

The more difficult question—though far from an insoluble one—is when an exemption from tax constitutes compensation. In most cases, the presence or absence of taxation upon wages, like the presence or absence of many other factors within the control of government— inflation, for example, or the rates charged by government-owned utilities, or import duties that increase consumer prices—affects the value of compensation, but is not an element of compensation itself. The Framers had this distinction well in mind. Hamilton, for example, wrote that as a result of “the fluctuations in the value of money,” “[i]t was . . . necessary to leave it to the discretion of the legislature to vary its provisions” for judicial compensation. The Federalist No. 79, p. 473 (C. Rossiter, ed. 1961); see also Will, supra, at 227 (the Constitution “placed faith in the integrity and sound judgment of the elected representatives to enact increases” in judicial salaries to account for inflation). Since Hamilton thought that the Compensation Clause “put it out of the power of [Congress] to change the condition of the individual [judge] for the worse,” The Federalist No. 79, at 473, he obviously believed that inflation does not diminish compensation as that term is used in the Constitution.

This distinction between Government action affecting compensation and Government action affecting the value of compensation was the basis for our statement in O’Malley, 307 U.S., at 282, that “[t]o subject [judges] to a general tax is merely to recognize that judges are also citizens, and that their particular function in government does not generate an immunity from sharing with their fellow citizens the material burden of the government. . . .” I agree with the Court, therefore, that Evans was wrongly decided—not, however, because in Evans there was no discrimination, but because in Evans the universal application of the tax demonstrated that the Government was not reducing the compensation of its judges but was acting as sovereign rather than employer, imposing a general tax.

But just as it is clear that a federal employee’s sharing of a tax-free status that all citizens enjoy is not compensation (and elimination of that tax-free status

not a reduction in compensation), so also it is clear that a tax-free status conditioned on federal employment is compensation, and its elimination a reduction. The Court apparently acknowledges that if a tax is imposed on the basis of federal employment (an income tax, for example, payable only by federal judges) it would constitute a reduc- tion in compensation. It is impossible to understand why a tax that is suspended on the basis of federal employment (an exemption from federal income tax for federal judges) does not constitute the conferral of compensation—in which case its elimination is a reduction, whether or not federal judges end up being taxed just like other citizens. Only converting the Compensation Clause into a Discrimination Clause can explain a contrary conclusion.

And this, of course, is what has been achieved by the targeted extension of the Medicare tax to federal employees who were previously exempt. It may well be that, in some abstract sense, they are not being “discriminated against,” since they end up being taxed like other citizens; but this does not alter the fact that, since exemption from the tax was part of their employment package— since they had an employment expectation of a preferential exemption from taxation—their compensation was being reduced. One of the benefits of being a federal judge (or any federal employee) had, prior to 1982, been an exemption from the Medicare tax. This benefit Congress took away, much as a private employer might terminate a contractual commitment to pay Medicare taxes on behalf of its employees. The latter would clearly be a cut in compensation, and so is the former. ² Had Congress simply imposed the Medicare tax on its

2 As the Court explains, the purpose of the Medicare tax extension was to ensure that federal workers “bear a more equitable share of the costs of financing the benefits to which many of them eventually became entitled” by reason of their own or their spouses’ private-sector employment. Ante, at 2 (internal quotation marks and citation omitted). As with the Social Security tax, therefore, the Medicare tax aspect of this case does not present the situation in which a tax exemption has been eliminated in return for some other benefit, different in kind but equivalent in value. Cf. Ante, at 14 (“[P]articipation in Social Security as judges would benefit only a minority”).

own employees (including judges) at the time it introduced that tax for other working people, no benefit of federal employment would have been reduced, because, with respect to the newly introduced tax, none had ever existed. But an extension to federal employees of a tax from which they had previously been exempt by reason of their employ- ment status, seems to me a flat-out reduction of federal employment compensation.

III

As should be clear from the above, though I agree with the Court that the extension of the Social Security tax to federal judges runs afoul of the Compensation Clause, I disagree with the Court’s grounding of this holding on the discriminatory manner in which the extension occurred. In this part of its opinion, however, the Court’s antidiscrimination rationale is slightly different from that which appeared in its discussion of the Medicare tax. There, the focus was on discrimination compared with ordinary citizens; here, the focus is on discrimination vis-a-vis other federal employees. (As the Court explains, federal judges, unlike nearly all other federal employees, were not given the opportunity to opt out of paying the tax). On my analysis, it would not matter if every federal employee had been made subject to the Social Security tax along with judges, so long as one of the previous entitlements of their federal employment had been exemption from that tax. Federal judges, unlike all other federal employees except the President, see Art. II, Sec. 1, cl. 7, cannot, consistent with the Constitution, have their compensation diminished. If this case involved salary cuts to pay for Social Security, rather than taxes to pay for Social Security, the irrelevance of whether other federal employees were covered by the operative legislation would be clear.

I join in the judgment that extension of the Social Security tax to sitting Article III judges was unconstitutional. I would affirm the Federal Circuit’s holding that extension of the Medicare tax was unconstitutional as well.

October 29, 2001 394 2001–44 I.R.B.

SUPREME COURT OF THE

UNITED STATES

No. 99–1978

UNITED STATES, PETITIONER v .

TERRY J. HATTER, JR., JUDGE, UNITED STATES DISTRICT COURT

FOR THE CENTRAL DISTRICT OF

JUSTICE THOMAS, concurring in the judgment in part and dissenting in part.

I believe this Court was correct in Evans v. Gore, 253 U.S. 245 (1920), when it held that any tax that reduces a judge’s net compensation violates Article III of the Constitution. Accordingly, I would affirm the judgment of the Court of Appeals in its entirety.

CALIFORNIA, ET AL.

ON WRIT OF CERTIORARI TO THE

UNITED STATES COURT OF

APPEALS FOR THE FEDERAL CIRCUIT

May 21, 2001

2001–44 I.R.B. 395 October 29, 2001