Beer v. United States

 United States Court of Appeals for the Federal Circuit __________________________ PETER H. BEER, TERRY J. HATTER, JR., RICHARD A. PAEZ, LAURENCE H. SILBERMAN, A. WALLACE TASHIMA AND U. W. CLEMON, Plaintiffs-Appellants, v. UNITED STATES, Defendant-Appellee. __________________________ 2010-5012 __________________________ Appeal from the United States Court of Federal Claims in No. 09-CV-037, Senior Judge Robert H. Hodges, Jr. _________________________ Decided: October 5, 2012 _________________________ CHRISTOPHER LANDAU, Kirkland & Ellis, LLP, of Washington, DC, argued for plaintiffs-appellants. With him on the brief were JOHN C. O’QUINN and K. WINN ALLEN. BRIAN M. SIMKIN, Assistant Director, Commercial Litigation Branch, Civil Division, United States Depart- ment of Justice, of Washington, DC, argued for defendant- appellee. With him on the brief were STUART F. DELERY, BEER v. US 2 Acting Assistant Attorney General, JEANNE E. DAVIDSON, Director, and MICHAEL S. MACKO, Trial Attorney. JEFFREY A. LAMKEN, MoloLamken LLP, of Washing- ton, DC, for amicus curiae, The Federal Judges Associa- tion. With him on the brief were MARTIN V. TOTARO and LUCAS M. WALKER. AARON M. PANNER, Kellogg, Huber, Hansen, Todd, Evans & Figel, P.L.L.C., of Washington, DC, for amicus curiae, International Municipal Lawyers Association. WILLIAM P. ATKINS, Pillsbury Winthrop Shaw Pittman, LLP, of McLean, Virginia, for amicus curiae, Bar Association of the District of Columbia. Of counsel was ERIN M. DUNSTON, Buchanan Ingersoll & Rooney P.C., of Alexandria, Virginia. LAWRENCE D. ROSENBERG, Jones Day, of Washington, DC, for amicus curiae, American Bar Association. CARTER G. PHILLIPS, Sidley Austin, LLP, of Washing- ton, DC, for amicus curiae, Federal Circuit Bar Associa- tion. With him on the brief was REBECCA K. WOOD. LAWRENCE M. FRIEDMAN, Barnes, Richardson & Colburn, Of Chicago, Illinois, for amicus curiae, Customs and International Trade Bar Association. __________________________ Before RADER, Chief Judge, NEWMAN, MAYER 1 , LOURIE, BRYSON, LINN, DYK, PROST, MOORE, O’MALLEY, REYNA, and WALLACH, Circuit Judges. 1 Judge Mayer participated in the decision on panel rehearing. 3 BEER v. US Opinion for the court filed by Chief Judge RADER, in which Circuit Judges NEWMAN, MAYER, LOURIE, LINN, PROST, MOORE, O’MALLEY, REYNA and WALLACH join. Dissenting opinion filed by Circuit Judge DYK, in which Circuit Judge BRYSON joins. Concurring opinion filed by Circuit Judge O’MALLEY, in which Circuit Judges MAYER and LINN join. Concurring opinion filed by Circuit Judge WALLACH. RADER, Chief Judge. The Constitution erects our government on three foundational corner stones – one of which is an independ- ent judiciary. The foundation of that judicial independ- ence is, in turn, a constitutional protection for judicial compensation. The framers of the Constitution protected judicial compensation from political processes because “a power over a man’s subsistence amounts to a power over his will.” The Federalist No. 79, p. 472 (Alexander Hamil- ton) (Clinton Rossiter ed., 1961). Thus, the Constitution provides that “Compensation” for federal judges “shall not be diminished during their Continuance in Office.” U.S. Const. art. III, § 1 (“Compensation Clause”). This case presents this court with two issues involv- ing judicial independence and constitutional compensa- tion protections – one old and one new. First, the old question: does the Compensation Clause of Article III of the Constitution prohibit Congress from withholding the cost of living adjustments for Article III judges provided for in the Ethics Reform Act of 1989 (“1989 Act”)? To answer this question, this court revisits the Supreme Court’s decision in United States v. Will, 449 U.S. 200 (1980). Over a decade ago in Williams v. United States, 240 F.3d 1019 (Fed. Cir. 2001) (filed with dissenting opinion by Plager, J.), a divided panel of this court found BEER v. US 4 that Will applied to the 1989 Act and concluded that Congress could withdraw the promised 1989 cost of living adjustments. This court en banc now overrules Williams and instead determines that the 1989 Act triggered the Compensation Clause’s basic expectations and protec- tions. In the unique context of the 1989 Act, the Constitu- tion prevents Congress from abrogating that statute’s precise and definite commitment to automatic yearly cost of living adjustments for sitting members of the judiciary. The new issue involves pure statutory interpretation, namely, whether the 2001 amendment to Section 140 of Pub. L. No. 97-92 overrides the provisions of the 1989 Act. This court concludes the 1989 Act was enacted after Section 140, and as such, the 1989 Act’s automatic cost of living adjustments control. I. The 1989 Act overhauled compensation and ethics rules for all three branches of government. With respect to the judiciary, it contained two reciprocal provisions. On the one hand, the 1989 Act limited a federal judge’s ability to earn outside income and restricted the receipt of honoraria. On the other hand, the 1989 Act provided for self-executing and non-discretionary cost of living adjust- ments (“COLA”) to protect and maintain a judge’s real salary. The 1989 Act provides that whenever a COLA for General Schedule federal employees takes effect under 5 U.S.C. § 5303, the salary of judges “shall be adjusted” based on “the most recent percentage change in the [Employment Cost Index] . . . as determined under section 704(a)(1) of the Ethics Reform Act of 1989.” Pub. L. No. 101-194, § 704(a)(2)(A), 103 Stat. 1716, 1769 (Nov. 30, 1989). The Employment Cost Index (“ECI”) is an index of wages and salaries for private industry workers published 5 BEER v. US quarterly by the Bureau of Labor Statistics. Section 704(a)(1) of the 1989 Act calculates COLAs by first de- termining the percent change in the ECI over the previ- ous year. Id. at § 704(a)(1)(B). Next, the statutory formula reduces the ECI percentage change by “one-half of 1 percent . . . rounded to the nearest one-tenth of 1 percent.” Id. However, no percentage change determined under Section 704(a)(1) shall be “less than zero” or “greater than 5 percent.” Id. While the 1989 Act states that judicial salary mainte- nance would only occur in concert with COLAs for Gen- eral Schedule federal employees under 5 U.S.C. § 5303, these General Schedule COLAs are automatic, i.e., they do not require any further congressional action. See 5 U.S.C. § 5303(a). The only limitation on General Sched- ule COLAs is a presidential declaration of a “national emergency or serious economic conditions affecting the general welfare” making pay adjustments “inappropriate.” 5 U.S.C. § 5303(b). Notwithstanding the precise, automatic formula in the 1989 Act, the Legislative branch withheld from the Judicial branch those promised salary adjustments in fiscal years 1995, 1996, 1997, and 1999. During these years, General Schedule federal employees received the adjustments under Section 5303(a), but Congress blocked the adjustments for federal judges. See Pub. L. No. 103- 329, § 630(a)(2), 108 Stat. 2382, 2424 (Sept. 30, 1994) (FY 1995); Pub. L. No. 104-52, § 633, 109 Stat. 468, 507 (Nov. 19, 1995) (FY 1996); Pub. L. No. 104-208, § 637, 110 Stat. 3009, 3009-364 (Sept. 30, 1996) (FY 1997); Pub. L. No. 105-277, § 621, 112 Stat. 2681, 2681-518 (Oct. 21, 1998) (FY 1999). In response to these missed adjustments, several fed- eral judges filed a class action alleging these acts dimin- BEER v. US 6 ished their compensation in violation of Article III. After certifying a class of all federal judges serving at the time (including appellants) and without providing notice or opt-out rights, the district court held that Congress vio- lated the Compensation Clause by blocking the salary adjustments. See Beer v. United States, 671 F.3d 1299, 1308–09 (Fed. Cir. 2012); Williams v. United States, 48 F. Supp. 2d 52 (D.D.C. 1999). On appeal, this court reversed the district court’s judgment. See Williams, 240 F.3d at 1019. This court opined that the Supreme Court’s decision in Will fore- closed the judges’ claim as a matter of law. Id. at 1033, 1035, 1040. According to this court, Will ruled that promised future salary adjustments do not qualify as “Compensation” protected under the Constitution until they are “due and payable.” Id. at 1032 (quoting Will, 440 U.S. at 228). Thus, Congress enjoyed full discretion to revoke any future judicial COLAs previously established by law, no matter how precise or definite, as long as the adjustments had not yet taken effect. Id. at 1039. This court declined to hear the case en banc over the dissent of three judges. See 264 F.3d 1089, 1090–93 (Fed. Cir. 2001) (Mayer, C.J., joined by Newman and Rader, JJ.); id. at 1093–94 (Newman, J., joined by Mayer, C.J. and Rader, J.). The Supreme Court denied certiorari over the dissent of three Justices. See 535 U.S. 911 (2002) (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Following this court’s decision in Williams, Congress amended a 1981 appropriations rider commonly known as Section 140. Section 140 originally read: Notwithstanding any other provision of law or of this joint resolution, none of the funds appropri- ated by this joint resolution or by any other Act 7 BEER v. US shall be obligated or expended to increase, after the date of enactment of this joint resolution, any salary of any Federal judge or Justice of the Su- preme Court, except as may be specifically author- ized by Act of Congress hereafter enacted: Provided, That nothing in this limitation shall be construed to reduce any salary which may be in effect at the time of enactment of this joint resolu- tion nor shall this limitation be construed in any manner to reduce the salary of any Federal judge or of any Justice of the Supreme Court. Pub. L. No. 97-92, § 140, 95 Stat. 1183, 1200 (1981) (codi- fied at 28 U.S.C. § 461 note) (emphasis added). While Section 140 originally expired in 1982, see Williams, 240 F.3d at 1026–27, it was revived by a 2001 amendment that added: “This section shall apply to fiscal year 1981 and each fiscal year thereafter.” Pub. L. No. 107-77, § 625, 115 Stat. 748, 803 (Nov. 28, 2001). Following the Section 140 amendment, Congress en- acted legislation specifically allowing federal judges to receive the salary adjustments mandated by the 1989 Act in fiscal years 2002, 2003, 2004, 2005, 2006, 2008, and 2009. See Barbara L. Schwemle, Congressional Research Service, Legislative, Executive, and Judicial Officials: Process for Adjusting Pay and Current Salaries 2-4 (Feb. 9, 2011). For fiscal years 2007 and 2010, all General Schedule and Executive level federal employees received COLAs under 5 U.S.C. § 5303(a), but federal judges received no adjustments. Congress did not affirmatively authorize judicial COLAs in those years and took the position that, because of the requirements of Section 140, judicial COLAs could not be funded.” The current case results from the combination of the blocking legislation of the 1990s and the amendment to BEER v. US 8 Section 140. Appellants are six current and former Arti- cle III judges, all of whom entered into federal judicial service before 2001. In January 2009, they filed a com- plaint in the United States Court of Federal Claims claiming that Congress violated the Compensation Clause by withholding the salary adjustments established by the 1989 Act. They claimed a deficit resulted not only from the withholding of COLAs in 2007 and 2010, but also the calculation of adjustments due in other years by reference to base compensation that did not include the amounts withheld in 1995, 1996, 1997, and 1999. For relief, they sought back pay for the additional amounts they allegedly should have received during the period covered by the applicable six-year statute of limitations. The Court of Federal Claims dismissed the complaint based on the Williams precedent. On appeal, this court summarily affirmed the judgment, stating that “Williams controls the disposition of this matter.” Beer v. United States, 361 F. App’x. 150, 151–52 (Fed. Cir. 2010). The Supreme Court granted the subsequent petition for certiorari, vacated the judgment, remanded the case for “consideration of the question of preclusion,” and stated that “further proceedings . . . are for the Court of Appeals to determine.” Beer v. United States, 131 S. Ct. 2865 (2011). Specifically, in opposing the petition for certiorari, the Government had argued that Appellants could not litigate anew the issue resolved in Williams because they had been absent members of the class action in Williams. Upon remand, this court unanimously concluded that Appellants were not precluded from bringing their Com- pensation Clause claims in the present case. Beer v. United States, 671 F.3d 1299, 1309 (Fed. Cir. 2012). The district court in Williams had not provided Appellants 9 BEER v. US with notice of the class certification. Thus they were not bound by the result of that earlier litigation. See id. at 1305–09. This court nonetheless continued to feel con- strained by the ultimate conclusion in Williams and affirmed the Court of Federal Claims’ dismissal of the complaint. Id. at 1309. Subsequently, this court granted Appellants’ petition for rehearing en banc. 468 F. App’x 995 (Fed. Cir. 2012). II. This court has jurisdiction over the Court of Federal Claims’ dismissal of the Appellants’ complaint under 28 U.S.C. § 1295(a)(3). This court reviews the decision to dismiss the complaint without deference. Hearts Bluff Game Ranch, Inc. v. United States, 669 F.3d 1326, 1328 (Fed. Cir. 2012); Frazer v. United States, 288 F.3d 1347, 1351 (Fed. Cir. 2002). This court en banc now turns its attention to two pre- liminary issues before addressing the merits of the ap- peal. First, judicial review of laws affecting judicial compensation is not done lightly as these cases implicate a conflict of interest. Will, 449 U.S. at 211–17. After all, judges should disqualify themselves when their impartial- ity might reasonably be questioned or when they have a potential financial stake in the outcome of a decision. See 28 U.S.C. § 455(a). In Will, the Supreme Court applied the time-honored “Rule of Necessity” because if every potentially conflicted judge were disqualified, then plain- tiffs would be left without a tribunal to address their claims. See Will, 449 U.S. at 213–17. The Rule of Neces- sity states that “although a judge had better not, if it can be avoided, take part in the decision of a case in which he has any personal interest, yet he not only may but must do so if the case cannot be heard otherwise.” Id. at 213 (quoting F. Pollack, A First Book of Jurisprudence 270 BEER v. US 10 (6th ed. 1929)) (emphasis added). This court relies on the Supreme Court’s complete analysis of the Rule of Neces- sity and concludes that this en banc court may, indeed must, hear the case. See id. at 211–18. On the other preliminary procedural question, this court deliberately limits the questions under review. To be specific, this court en banc does not overrule the Wil- liams panel’s analysis of Section 140. See 240 F.3d at 1026–27. Furthermore, it does not overrule the Beer panel’s analysis of preclusion. See 671 F.3d 1299. This court adopts the prior panel’s analysis of the preclusion issue in toto. Now the court en banc proceeds to the old and new questions previously set forth. III. At the outset, this court must honor and address the Supreme Court’s decision in Will. As the Williams panel correctly noted, if Will resolves the validity of Congress’ decision to block the COLAs promised in the 1989 Act, then any remedy for salary diminution in this case lies not in this court but in the Supreme Court. See Williams, 240 F.3d at 1035. However, if Will is inapplicable to the statutory scheme at play in this case, then this court has an obligation to resolve the issue. United States v. Will, supra, tested the validity of congressional blocking acts preventing COLAs provided for under the 1975 Adjustment Act (“1975 Act”). The 1975 Act purported to protect judicial salaries with ad- justments calculated under an opaque and indefinite process. Section 5305, as in effect in 1975, directed the President to “carry out the policy stated in section 5301” when giving COLAs to General Schedule federal employ- ees. 5 U.S.C. § 5305(a) (1976). Section 5301 in turn articulated a four-fold policy for setting federal pay: (1) equal pay for equal work; (2) pay distinction based on 11 BEER v. US work and performance distinctions; (3) comparable pay with private sector jobs for comparable work; and (4) interrelated statutory pay levels. 5 U.S.C. § 5301(a) (1976). In furtherance of this policy, the President appointed an agent to prepare an annual report on federal salaries. 5 U.S.C. § 5305(a)(1) (1976). This annual report relied on statistics from the Bureau of Labor Statistics on private sector pay, views of the “Federal Employees Pay Council” about the comparability of private and public sector pay systems, and the views of employee organizations not represented in the Council. 5 U.S.C. § 5305(a)(1) (1976). This report did not and could not mandate the award of COLAs. The President also received a report from “The Advi- sory Committee on Federal Pay.” 5 U.S.C. § 5305(a)(2) (1976). This committee reviewed the report issued by the President’s agent under section 5305(a)(1) and considered further views and recommendations provided by “em- ployee organizations, the President’s agent, other officials of the Government of the United States, and such experts as it may consult.” 5 U.S.C. § 5306(a)–(b) (1976). Based on these reports, the President could provide COLAs to General Schedule federal employees. 5 U.S.C. § 5305(a)(2). If the President decided to recommend an adjustment, he would transmit to Congress the overall adjustment percentage. 5 U.S.C. § 5305(a)(3). Any judi- cial COLAs were pegged to the “overall percentage” in the President’s report to Congress under section 5305. 28 U.S.C. § 461 (1976). Despite the 1975 Act, Congress allowed several COLAs for General Schedule federal employees but denied the increases to judges and other senior officials. The Supreme Court discussed the details of the legislation BEER v. US 12 that blocked these increases. See Will, 449 U.S. at 205– 09. In 1978, a group of federal judges filed suit alleging this blocking legislation was an unconstitutional diminu- tion in salary contrary to Article III. Once the case made its way to the Supreme Court, the Court considered “when, if ever, . . . the Compensation Clause prohibit[s] the Congress from repealing salary increases that other- wise take effect automatically pursuant to a formula previously enacted.” Id. at 221. The Court concluded that Congress could block COLAs due to judges so long as the blocking legislation took effect in the fiscal year prior to the year in which the increase would have become pay- able. Id. at 228–29. According to the Court, “a salary increase ‘vests’ . . . only when it takes effect as part of the compensation due and payable to Article III judges.” Id. at 229. The 1989 Act, informed by the failures of the 1975 Act’s procedure, adopted a different purpose, used a different structure, and created different expectations than the 1975 Act. The 1975 Act “involved a set of inter- locking statutes which, in respect to future cost-of-living adjustments, were neither definite nor precise.” Williams, 535 U.S. at 917 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Instead of being tied to the percent change in a known, published metric of inflation such as the Employment Cost Index, the ad- justments under the 1975 Act depended on the discretion- ary decisions of the President’s agent and the Advisory Committee on Federal Pay. Furthermore, the President was not obligated to award adjustments to General Schedule employees on a specific timeline or even pursu- ant to the suggestions from the agent and the committee. Rather, he only did so if it furthered the policies under- pinning federal pay articulated in 5 U.S.C. § 5301. Thus, the method for calculating COLAs under the 1975 Act 13 BEER v. US was “imprecise as to amount and uncertain as to effect.” Id. By contrast, the 1989 Act promised a mechanical im- plementation of COLAs for judges under the following equation: ⎛ ( ECI Year N −1 ) − ( ECI Year N − 2 ) ⎞ Adjustment Year N = ⎜⎜ ⎟⎟ × (100 ) × (0.995) ⎝ ECI Year N − 2 ⎠ See Pub. L. No. 101-194, § 704(a)(1)(B), 103 Stat. 1716, 1769 (Nov. 30, 1989). The Act contained only two limits: a presidential prohibition (due to national emergency or extreme economic circumstances) and a ceiling (of no more than five percent). Id. In essence, the statutes reviewed in Will required ju- dicial divination to predict a COLA and prevented the creation of firm expectations that judges would in fact receive any inflation-compensating adjustment. In that context, as the Supreme Court noted, no adjustment vested until formally enacted and received. However, the statutes reviewed in Williams and in this case provide COLAs according to a mechanical, automatic process that creates expectation and reliance when read in light of the Compensation Clause. Indeed a prospective judicial nominee in 1989 might well have decided to forego a lucrative legal career, based, in part, on the promise that the new adjustment scheme would preserve the real value of judicial compensation. Aside from their respective differences in methods for calculating COLAs, the 1989 Act’s overall scope and legislative history distinguishes it from the statutory scheme addressed in Will. In fact, the automaticity of the 1989 Act’s COLAs takes on heightened significance in light of the broader statutory scheme because the 1989 BEER v. US 14 Act also banned judges from earning outside income and honoraria. See Brown v. Gardner, 513 U.S. 115, 118 (1994) (“The meaning of statutory language, plain or not, depends on its context.”). In sum, the salary protections in the 1989 Act are only part of a comprehensive codifica- tion of ethical rules, Pub. L. No. 101-194 §§ 301-03, finan- cial reporting requirements, id. at § 202, work rules for senior judges, id. at § 705, and -- perhaps most important -- prohibitions on outside income and honoraria, id. at § 601. Of the 935 active and senior judges in 1987, four hun- dred reported earning outside income from teaching law, speaking fees, and other sources. 135 Cong. Rec. S29,693 (daily ed. Nov. 17, 1989). More than half reported extra earnings from $16,624 to $39,500. Id. The Report by The Bipartisan Task Force on Ethics, which became the basis for the Ethics Reform Act of 1989, noted that the repeated failure to provide recommended salary increases for judges and other executive employees meant increased reliance on “earning honoraria as a supplement to their official salaries.” 135 Cong. Rec. H30,744 (daily ed. Nov. 21, 1989) (Task Force Report). During consideration of the 1989 Act, Congress acknowledged that denying access to outside income would amount to a “pay cut.” 135 Cong. Rec. S29,662 (daily ed. Nov. 17, 1989) (statement of Sen. Dole that removing outside income is a “pay cut”); see also 135 Cong. Rec. H29,488 (daily ed. Nov. 16, 1989) (state- ment of Rep. Fazio), H29,492 (daily ed. Nov. 16, 1989) (statement of Rep. Ford). In that context, reliance on the 1989 Act’s compensation maintenance formula took on added significance. See 135 Cong. Rec. H29,503 (daily ed. Nov. 16, 1989) (statement of Rep. Wolpe) (“[The] pay adjustment provision [is] tied directly to the elimination of all honoraria or speaking fees.”). Indeed, the Task Force Report emphasized that the restrictions and limita- 15 BEER v. US tions on outside earned income, honoraria, and employ- ment made by the Act are conditional on the enactment of the increased pay provisions. 135 Cong. Rec. H30,745 (daily ed. Nov. 21, 1989) (Task Force Report). The dependable COLA system became “a final impor- tant part” of the package designed to remove salaries “from their current vulnerability for political demagogu- ery.” 135 Cong. Rec. H29,483 (Nov. 16, 1989) (statement of Rep. Fazio); H30,753 (Nov. 21, 1989) (Task Force Report). In sum, the 1989 Act reduced judges’ income by banning outside income but promised in exchange auto- matic maintenance of compensation – a classic legislative quid pro quo. 135 Cong. Rec. H29,484 (Nov. 16, 1989) (statement of Rep. Martin stating that the Ethics Reform Act of 1989 is a comprehensive and interrelated package); cf. 135 Cong. Rec. H29,499 (Nov. 16, 1989) (statement of Rep. Crane objecting to the interrelated nature of the package and advocating separate bills for ethics and pay). Thus, the 1989 statutory scheme was a precise legis- lative bargain which gave judges “an employment expec- tation” at a certain salary level. Cf. United States v. Hatter, 532 U.S. 557, 585 (2001) (Scalia, J., concurring in part and dissenting in part) (arguing that the repeal of judges’ exception from Medicare tax constituted a dimin- ishment in compensation because judges had an expecta- tion of an exemption from this tax). Moreover, the 1989 Act COLA provisions were not an increase in judicial pay. If so, the connection with the vesting rule for pay in- creases articulated in Will might be a closer issue. Rather, the statute ensured that real judicial salary would not be reduced in the face of the elimination of outside income and the operation of inflation. See Wil- liams, 535 U.S. at 916 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). BEER v. US 16 The vesting rules considered in Will are not expressly limited to the 1975 Act. However, the Supreme Court had no occasion to draw a distinction between a discretionary COLA scheme and a self-executing, non-discretionary adjustment for inflation coupled with a reduction in judicial compensation via elimination of outside income. For this reason, therefore, this court must examine fur- ther the actual differences in the two statutory schemes. The Supreme Court described the adjustments under the 1975 Act as “automatic.” See Will, 449 U.S. at 203, 223–24. An examination of the 1975 Act, however, shows that the adjustments at issue in Will were automatically operative only “once the Executive had determined the amount.” Id. at 203 (emphasis added). The ways that the Executive determined the amounts under the 1975 Act and the 1989 Act are very different. The former was an uncertain, discretionary process. The latter is precise and definite. While the Supreme Court described the COLAs in Will as “automatic,” the only aspect that was truly auto- matic was the link between judicial and General Schedule employee salaries. Whether General Schedule employees (and judges) would receive COLAs in any given year or whether those COLAs would maintain earning levels was anything but certain under the 1975 Act. Consequently, the only line the Supreme Court could draw in Will was between before and after the COLAs at issue were funded. The 1989 Act’s scheme presents a much different land- scape than the Court confronted in Will. For these rea- sons, Will does not foreclose the relief that the judges seek. Although this court determines that Williams incor- rectly applied Will and other aspects of the law, this determination does not end the inquiry. The court must 17 BEER v. US now examine whether Congress’ decisions to deny the promised COLAs actually violated the Compensation Clause in Article III of the Constitution. The Compensation Clause has two basic purposes. First, it promotes judicial independence by protecting judges from diminishment in their salary by the other branches of Government. The founders of this nation understood the connections amongst protections for Life, Liberty, and the Pursuit of Happiness, protections for judicial independence, and protections for judicial com- pensation. Listed among the colonists’ grievances with the English Crown was that the King “ha[d] made Judges dependent on his Will alone for the Tenure of their Of- fices, and the amount and payment of their salaries.” Decl. of Independence para. 11 (U.S. 1776). As explained in The Federalist Papers, “[n]ext to permanency in office, nothing can contribute more to the independence of the judges than a fixed provision for their support.” The Federalist No. 79, p. 472 (Alexander Hamilton) (Clinton Rossiter ed., 1961). During the Constitutional Convention in 1787, the in- spired draftsmen set out to protect against abuses such as those enumerated in the Declaration of Independence. James Madison of Virginia proposed prohibiting both enhancement and reduction of salary lest judges defer unduly to Congress when that body considered pay in- creases. Will, 449 U.S. at 219–20. Madison urged that variations in the value of money could be “guarded agst. by taking for a standard wheat or some other thing of permanent value.” Id. at 220 (quoting 2 M. Farrand, The Records of the Federal Convention of 1787, p. 45 (1911)). The Convention rejected Madison’s proposal because any commodity chosen as a standard for judicial compensation could also lose value due to inflationary forces, i.e., the value of wheat could also fluctuate. Id. Thus, the Com- BEER v. US 18 pensation Clause did not tie judicial salaries to any com- modity. The framers instead acknowledged that “fluctua- tions in the value of money, and in the state of society, rendered a fixed rate of compensation [for judges] in the Constitution inadmissible.” The Federalist No. 79, supra. The Convention adopted the clause in its current form while voicing, at length, concerns to protect judicial compensation against economic fluctuation and reprisal. The Compensation Clause, as well as promoting judi- cial independence, “ensures a prospective judge that, in abandoning private practice -- more often than not more lucrative than the bench -- the compensation of the new post will not diminish.” Will, 449 U.S. at 221. This expectancy interest attracts able lawyers to the bench and enhances the quality of justice. Id. This expectancy interest does not encompass increases in future salary but contemplates maintenance of that real salary level. Williams, 535 U.S. at 916 (Breyer, J. joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari); The Federalist No. 79, supra, (noting that an Article III judge is assured “of the ground upon which he stands” and that he should “never be deterred from his duty by the appre- hension of being placed in a less eligible situation”). The dual purpose of the Compensation Clause pro- tects not only judicial compensation that has already taken effect but also reasonable expectations of mainte- nance of that compensation level. See Williams, 535 U.S. at 916 (Breyer, J. joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). The 1989 Act prom- ised, in precise and definite terms, salary maintenance in exchange for prohibitions on a judge’s ability to earn outside income. The 1989 Act set a clear formula for calculation and implementation of those maintaining adjustments. Thus, all sitting federal judges are entitled to expect that their real salary will not diminish due to 19 BEER v. US inflation or the action or inaction of the other branches of Government. The judicial officer should enjoy the free- dom to render decisions -- sometimes unpopular decisions -- without fear that his or her livelihood will be subject to political forces or reprisal from other branches of govern- ment. Prospective judges should likewise enjoy the same ex- pectation of independence and protection. A lawyer making a decision to leave private practice to accept a nomination to the federal bench should be entitled to rely on the promise in the Constitution and the 1989 Act that the real value of judicial pay will not be diminished. Will, 449 U.S. at 220–21; cf. United States v. Winstar Corp., 518 U.S. 839, 872 (1996) (recognizing that government promises may give rise to reasonable expectations). To be sure, the Compensation Clause does not require periodic increases in judicial salaries to offset inflation or any other economic forces. As noted before, the Constitu- tional Convention did not tie judicial salaries to a com- modity or other standard of measurement. Will, 449 U.S. at 220. However, when Congress promised protection against diminishment in real pay in a definite manner and prohibited judges from earning outside income and honoraria to supplement their compensation, that Act triggered the expectation-related protections of the Com- pensation Clause for all sitting judges. A later Congress could not renege on that commitment without diminish- ing judicial compensation. That those compensation adjustments would happen in the future does not elimi- nate the reasonableness of the expectations created by the protections in the 1989 Act. Expectancy is, by its very nature, concerned with future events. Congress committed to providing sitting and prospec- tive judges with annual COLAs in exchange for limiting BEER v. US 20 their ability to seek outside income and to offset the effects of inflation. This decision furthered the Founders’ intention of protecting judges against future changes in the economy. Instead of fixing compensation relative to a commodity subject to inflationary pressure, Congress pegged the adjustment to a known measure of change to the economy as a whole, thus protecting the real salary of judges from both inflation and from fickle political will. By enacting blocking legislation in 1995, 1996, 1997, and 1999, Congress broke this commitment and effected a diminution in judicial compensation. Congress is not precluded from amending the 1989 Act. Congress may set up a scheme promising judges a certain pay scale or yearly cost of living increases. How- ever, the Constitution limits those changes. If a future Congress wishes to undo those promises, it may, but only prospectively. Any restructuring of compensation main- tenance promises cannot affect currently-sitting Article III judges. IV. Turning now to the second question, this court deter- mines that the 2001 amendment to Section 140 of Pub. L. 97-92 has no effect on the compensation due to judges. Unlike the preceding discussion of the Compensation Clause, this is a question of statutory interpretation. Without a statutory basis for withholding the COLAs, federal judges should have received the adjustments in 2007 and 2010. These adjustments are payable to the judges regardless of constitutional protections. Congress simply had no statutory authority to deny them. As noted above, Section 140 was part of an appropria- tions bill passed in 1981. It barred judges from receiving additional compensation except as Congress specifically authorized in legislation postdating Section 140. See Pub. 21 BEER v. US L. No. 97-92, § 140, 95 Stat. 1183, 1200 (Dec. 15, 1981). The appropriations act containing Section 140 expired by its terms on September 30, 1982. See Williams, 240 F.3d at 1026. Thus, the rule that judicial pay adjustments had to be “specifically authorized by Act of Congress hereafter enacted” expired in 1982. Of course, in 2001, Congress amended Section 140, purporting to apply it “to fiscal year 1981 and each fiscal year thereafter.” Pub. L. No. 107-77, Title VI, § 625, 115 Stat. 748, 803 (2001). Notably, Congress chose 1981 as the effective date for this extension of Section 140. As shown above, Congress did not explicitly authorize judi- cial compensation adjustments in 2007 and 2010. If Section 140 applied to bar those 2007 and 2010 adjust- ments, the absence of that additional Act of Congress would block -- solely on the basis of this statute -- any adjustments in those years. Section 140, however, by its own terms, did not block the 2007 and 2010 adjustments. Section 140 is straight- forward: it bars judicial salary increases unless (1) “spe- cifically authorized by Act of Congress” and (2) “hereafter enacted.” Pub. L. No. 97-92, § 140. The 1989 Act’s precise and definite promise of COLAs clearly satisfies the first requirement to avoid a Section 140 bar. Williams, 240 F.3d at 1027. The 1989 Act “specifically authorized” the 2007 and 2010 adjustments which occurred under its precise terms. Section 140 was enacted in 1981 and the 1989 Act oc- curred eight years later. Thus, the 1989 Act was “hereaf- ter enacted” within Section 140’s meaning. When Congress amended Section 140 in 2001, it did not wipe the slate clean and set a new benchmark for the “hereaf- ter enacted” requirement. The 2001 amendment makes no reference to its own November 28, 2001, enactment BEER v. US 22 date. Instead, the amendment reiterates the 1981 base- line found elsewhere in the original Section 140, making the provision applicable to “‘fiscal year 1981 and each fiscal year thereafter.’” Pub. L. No. 107-77. An amend- ment referring only to fiscal year 1981 cannot redefine “hereafter” to refer to an entirely different date two decades later. Thus, the “hereafter enacted” requirement remained unchanged setting the “hereafter enacted” trigger date as 1981. In other words, Congress amended the existing Section 140 in 2001, but Section 140 re- mained a part of the Public Law 97-92 enacted in 1981. Furthermore, the amendment did not change Section 140’s enactment date. Indeed the Government agreed at oral argument before this court en banc that the 2001 amendment did not change the “hereafter enacted” clause of Section 140. The 2001 amendment merely erased Section 140’s expiration date, making permanent what- ever effect the provision had when originally enacted. Congress thus expunged this court’s holding in Williams that Section 140 expired in 1982. The 2001 amendment, however, did not change Section 140’s substantive scope. The 1989 Act’s precise, automatic COLAs satisfy the requirements of Section 140 because it was enacted after Section 140. The Government withheld COLAs from judges in 2007 and 2010 solely because the government misinterpreted Section 140 as requiring a separate and additional authorizing enactment to put those adjust- ments into effect. By its own terms, Section 140 did not require that further authorizing legislation because it permitted COLAs under the “hereafter enacted” 1989 Act. V. In this case, Congress’ acts in 1995, 1996, 1997, and 1999 constitute unconstitutional diminishments of judi- cial compensation. Additionally, statutorily promised cost 23 BEER v. US of living adjustments were withheld in 2007 and 2010 based on an erroneous statutory interpretation. Appel- lants’ motion to amend their complaint to include a chal- lenge to the 2010 withholdings is granted. See Mills v. Maine, 118 F.3d 37, 53 (1st Cir. 1997) (“[A]ppellate courts have authority to allow amendments to complaints be- cause ‘[t]here is in the nature of appellate jurisdiction, nothing which forbids the granting of amendments.’”) (quoting Newman-Green, Inc. v. Alfonzo-Larrain, 490 U.S. 826, 834 (1989) (alterations omitted)). The statute of limitations does not bar these claims because, as established in Friedman v. United States, 159 Ct. Cl. 1, 7 (1962) and Hatter v. United States, 203 F.3d 795, 799–800 (Fed. Cir. 2000), aff’d in part, rev’d in part on other grounds, 532 U.S. 557 (2001), the claims are “continuing claims.” As relief, appellants are entitled to monetary damages for the diminished amounts they would have been paid if Congress had not withheld the salary adjustments mandated by the Act. On remand, the Court of Federal Claims shall calculate these damages as the additional compensation to which appellants were entitled since January 13, 2003 – the maximum period for which they can seek relief under the applicable statute of limitations. In making this calculation, the Court of Federal Claims shall incorporate the base salary in- creases which should have occurred in prior years had all the adjustments mandated by the 1989 Act had actually been made. See Hatter, 203 F.3d 795 (applying the “con- tinuing claim” doctrine to calculating wrongful withhold- ing of judicial pay). VI. This court has an “obligation of zealous preservation of the fundamentals of the nation. The question is not how much strain the system can tolerate; our obligation is BEER v. US 24 to deter potential inroads at their inception, for history shows the vulnerability of democratic institutions.” Beer v. United States, 592 F.3d 1326, 1329 (Fed. Cir. 2010) (Newman, J., dissenting from the denial of petition for hearing en banc). The judiciary, weakest of the three branches of government, must protect its independence and not place its will within the reach of political whim. The precise and definite promise of COLAs in the 1989 Act triggered the expectation-related protections of the Compensation Clause. As such, Congress could not block these adjustments once promised. The Court of Federal Claims’ dismissal of Appellants’ complaint is hereby reversed, and the case is remanded for further considera- tion in accordance with this opinion. OVERRULED-IN-PART, VACATED-IN-PART, AND REMANDED United States Court of Appeals for the Federal Circuit __________________________ PETER H. BEER, TERRY J. HATTER, JR., RICHARD A. PAEZ, LAURENCE H. SILBERMAN, A. WALLACE TASHIMA AND U. W. CLEMON, Plaintiffs-Appellants, v. UNITED STATES, Defendant-Appellee. __________________________ 2010-5012 __________________________ Appeal from the United States Court of Federal Claims in No. 09-CV-037, Senior Judge Robert H. Hodges, Jr. __________________________ DYK, Circuit Judge, with whom BRYSON, Circuit Judge, joins, dissenting. The majority opinion brings to mind an exchange be- tween Learned Hand and Justice Holmes. Judge Hand enjoined Justice Holmes to “[d]o justice” on the bench, but the Justice demurred: “That is not my job. My job is to play the game according to the rules.” Learned Hand, A Personal Confession, in The Spirit of Liberty 302, 306-07 (Irving Dilliard ed., 3d ed. 1960). If the Supreme Court must play by the rules, that duty must be doubly binding on subordinate federal courts. Fidelity to this principle BEER v. US 2 mandates adherence to the Supreme Court’s opinion in United States v. Will, 449 U.S. 200 (1980). I While the majority’s approach has much to recom- mend it as a matter of justice to the nation’s underpaid Article III judges, it has nothing to recommend it in terms of the rules governing adjudication. “The criterion of constitutionality is not whether we believe the law to be for the public good,” Adkins v. Children’s Hosp., 261 U.S. 525, 570 (1923) (Holmes, J., dissenting), but whether the law comports with the Supreme Court’s authoritative construction of the Constitution. Here, the issue is the scope of the Supreme Court’s 1980 decision in Will. Will’s holding is squarely on point. The Supreme Court’s fram- ing of the issue was unmistakably clear: “when, if ever, does the Compensation Clause prohibit the Congress from repealing salary increases that otherwise take effect automatically pursuant to a formula previously enacted?” 449 U.S. at 221. The answer was that a future salary increase “becomes irreversible under the Compensation Clause” when it “vests,” id., and that it “‘vests’ for pur- poses of the Compensation Clause only when it takes effect as part of the compensation due and payable to Article III judges,” id. at 228-29. The Court’s opinion in Will is unambiguous that the Court adopted what it has characterized as a “categorical” rule. See, e.g., Plaut v. Spendthrift Farm, Inc., 514 U.S. 211, 239-40 (1995). The Court in Will explained that for two of the years, the statute was passed before the Adjust- ment Act increases had taken effect— before they had become a part of the com- pensation due Article III judges. Thus, the departure from the Adjustment Act policy in no sense diminished the compen- 3 BEER v. US sation Article III judges were receiving; it refused only to apply a previously enacted formula. A paramount—indeed, an indispensa- ble—ingredient of the concept of powers delegated to coequal branches is that each branch must recognize and respect the limits on its own authority and the boundaries of the authority delegated to the other branches. To say that the Con- gress could not alter a method of calculat- ing salaries before it was executed would mean the Judicial Branch could command Congress to carry out an announced future intent as to a decision the Constitution vests exclusively in the Congress. We therefore conclude that a salary increase “vests” for purposes of the Compensation Clause only when it takes effect as part of the compensation due and payable to Arti- cle III judges. 449 U.S. at 228-29 (footnotes omitted). Under Will’s bright-line vesting rule, Congress was free to “abandon” a statutory formula and revoke a planned cost-of-living adjustment (“COLA”), as long as the revoking legislation was enacted into law before the COLA “took effect,” that is, became “due and payable” (i.e., before October 1, the first day of the next fiscal year). Id. at 227-29. In Will Years 1 and 4, Congress missed that deadline, and the Court held that the belated with- drawal of judges’ COLAs violated the Compensation Clause. Id. at 226, 230. But in Will Years 2 and 3, COLA-blocking statutes signed before October 1 were upheld, even though one of those statutes eliminated the BEER v. US 4 promised COLA just a day before it would have taken effect. Id. at 229. Will thus made clear that a future salary increase only becomes protected by the Compensation Clause when it becomes “due and payable”; an increase which is merely anticipated or expected has not vested, and is not pro- tected. By declining to follow Will’s clear vesting rule here, the majority also rejects the carefully crafted panel opinion in Williams v. United States, 240 F.3d 1019, 1039 (Fed. Cir. 2001), reh’g denied, 240 F.3d 1366 (Fed. Cir. 2001) (en banc), whose view of Will was supported at the time by a clear majority of the en banc court. See Wil- liams, 240 F.3d at 1366 (eight judges concurring in the denial of rehearing en banc because “we are duty-bound to enforce [Will’s] rule. If we have incorrectly read the Will opinion, the Supreme Court will have the opportunity to correct the error.”). II The majority attempts to redefine the constitutional test as turning not on “vesting,” but on “reasonable expec- tations,” a concept that appears nowhere in the Will opinion. To justify this shift, the majority seeks to distin- guish Will on its facts, namely on the dubious ground that the “automatic” salary adjustment scheme in Will was different from the “automatic” salary adjustment scheme in place in Williams and here. But even if factual differ- ences were pertinent (which, as we discuss below, could not support a departure from Will’s holding), there is no material difference between the statutes in Will and those in the Williams years (1995, 1996, 1997, and 1999). The Will statutes and the Williams statutes were not different insofar as they tied judicial compensation to General Schedule (“GS”) compensation, nor were they materially different as far as the definiteness of the GS COLA was 5 BEER v. US concerned. Contrary to the majority’s suggestion, under both schemes, the COLA was “required” unless the Presi- dent altered the COLA in response to “national emer- gency” or “economic conditions.” Compare 5 U.S.C. § 5305(c)(1) (1976) with 5 U.S.C. at § 5303(b)(1) (2006). As the House Report to the 1990 Act stated, “[t]he President would have discretion [under the 1990 Comparability Act] to alter this adjustment . . . . This discretion is substan- tially similar to current law,” i.e., the 1975 Act. H.R. Rep. No. 101-906, at 88 (1990). 1 And under both statutory schemes, the GS COLA, once established, would “take effect automatically.” Will, 449 U.S. at 221. 2 Thus, the statutory schemes appear “strikingly similar” for all practical purposes. Williams, 240 F.3d at 1027. Nevertheless, the majority asserts that the expecta- tion of a COLA created by the Williams statutes was significantly more “precise and definite,” Majority Op. 16, because under Will’s more complex scheme, there was greater discretion over the COLA—an assertion which is accurate only insofar as the President’s agent and Advi- sory Committee had greater discretion in setting the initial amount of the GS COLA. Under each statutory scheme, the President’s discretion was the same. 3 1 Plainly Congress saw the references in the 1975 Act to “economic conditions” and in the 1990 Act to “seri- ous economic conditions” as functionally the same, since the President’s discretion was to remain “substantially similar” under the 1990 Act as before. 2 Judge O’Malley’s concurrence misreads the dis- sent in suggesting that we view the COLAs in Will as “automatic” only because “the statutory scheme had run its course” in the disputed years. Concur. Op. 4. 3 Will’s statutory scheme required the President to appoint an ad- justment agent [who] was to compare BEER v. US 6 But whatever the discretion, if the test were “reason- able expectations,” then the key question would not be how the statutory scheme initially determined a COLA, but whether the amount of the COLA had become “precise and definite” at the time the blocking statute thwarted the judges’ expectations. In this respect, Will cannot be distinguished from Williams. For Will Year 3, no “judicial divination,” Majority Op. 13, would have been required: a GS COLA of 5.5% had already been specified in the Presi- dent’s Alternative Plan, 14 Weekly Comp. Pres. Docs. 1480 (Aug. 31, 1978), which was adopted and transmitted to Congress by the President a month before the Year 3 blocking statute was enacted. Will, 449 U.S. at 229. The President had no further discretion to change the amount of the COLA. As the majority notes, “once the Executive had determined the amount,” the adjustments in Will were automatically operative. Majority Op. 16 (quoting Will, 449 U.S. at 203) (internal quotation marks omitted). In the Williams years, at the time the blocking statutes were enacted, the prospective amount of the GS COLA could be calculated based on the Employment Cost Index figures released by the Bureau of Labor Statistics, al- salaries in the civil service with those in the private sector and then recommend an adjustment to an Advisory Committee. Subsequently, the Committee would make its own recommendation to the President, accepting, rejecting, or modifying the agent’s recommendation as the Committee thought desirable. The President would have to accept the Committee’s recom- mendation—unless he determined that national emergency or special economic conditions warranted its rejection. Williams v. United States, 535 U.S. 911, 917 (2002) (Breyer, J., joined by Scalia & Kennedy, JJ., dissenting). 7 BEER v. US though the President generally did not announce a final amount until after the blocking statutes were enacted. 4 Thus, the COLA in Will Year 3 was just as “precise and definite” as the COLAs in the Williams years. Of course, the COLAs remained uncertain in another respect: in both Will and Williams, the presumptive GS COLA could still be overridden by Congressional action, and in fact it was overridden for one of the Williams years. 5 Again, there is no meaningful difference between the situations in Will and Williams. 6 To summarize: in 4 For all the Williams years, GS salary adjustment tables were promulgated by Executive Order in the pre- ceding December. Exec. Order 12944, 60 Fed. Reg. 309 (Dec. 28, 1994); Exec. Order 12984, 61 Fed. Reg. 237 (Dec. 28, 1995); Exec. Order No. 13033, 61 Fed. Reg. 68987 (Dec. 27, 1996); Exec. Order No. 13106, 63 Fed. Reg. 68151 (Dec. 7, 1998). In each year, the judges’ COLAs had been blocked several weeks to months earlier. See Pub. L. 103-329, Title VI, § 630(a)(2), 108 Stat. 2382, 2424 (1994); Pub. L. 104-52, Title VI, § 633, 109 Stat. 468, 507 (1995); Pub. L. 104-208, Title VI, § 637, 110 Stat. 3009-364 (1996); Pub. L. 105-277, Title VI, § 621, 112 Stat. 2681-518 (1998). For one of the Williams years, 1996, the President transmitted an Alternative Plan to Congress setting a 2% GS COLA before the blocking statute was passed. 31 Weekly Comp. Pres. Docs. 1466, 1466-67 (1995). 5 For 1995, Congress reduced the GS COLA to 2%. Pub. L. 103-329, Title VI, § 630(a)(1), 108 Stat. 2382, 2424 (1994). The projected GS COLA had been 2.6%. See Sharon S. Gressle, Cong. Research Serv., Order No. RS20278, Judicial Salary-Setting Policy 6 (March 6, 2003). 6 Under the Will scheme, in addition to enacting separate legislation, Congress could have disapproved the Alternative Plan by a one-house legislative veto. Will, 449 U.S. at 204. But a legislative veto would not have zeroed out the GS COLA; it would have reinstated the amount recommended to the President, id., which was BEER v. US 8 both Will Year 3 and in each of the Williams years, at the time the judges’ COLA was blocked, the amount of the GS COLA had been established, the President retained no discretion to change the GS COLA, and the COLA would have taken effect automatically, absent Congressional intervention. The Supreme Court upheld the blocking statute in Will Year 3. 449 U.S. at 229. Yet the majority maintains that the blocking statutes in Williams offend the Constitution. This distinction is baffling. Finally, the majority here suggests that Will is distin- guishable because the statutes here (unlike the statutes in Will) imposed limits on the judges’ outside income, without “an increase in judicial pay.” Majority Op. 15. But the majority can hardly make a credible claim that judges’ outside compensation is protected by the Compen- sation Clause, and it follows that the reduction of outside compensation cannot create a Compensation Clause issue where none would otherwise exist. 7 III Even if the two statutory schemes were meaningfully different, and the Williams scheme created “reasonable judicial expectation[s] of future compensation” that did not exist in Will, Appellants’ Br. 29-31, that would be quite beside the point. Neither counsel for the appellants higher than the President’s figure in Will Year 3. See 14 Weekly Comp. Pres. Docs. 1480 (Aug. 31, 1978). It is unclear how Congressional action to increase the GS COLA could have made the judges’ expectations of a COLA in Will Year 3 less “precise and definite.” The legislative veto was held unconstitutional after Will and before the Williams years. INS v. Chadha, 462 U.S. 919 (1983). 7 In fact, the 1989 Act did increase judicial pay by 25%, thus offsetting the limitations on outside income. Pub.L. 101-194 § 703(a)(3), 103 Stat. 1716, 1768 (1989). 9 BEER v. US nor the majority is able to explain how that difference authorizes this court to disregard Will’s clear vesting rule. The majority concedes that “the vesting rules considered in Will are not expressly limited to the 1975 Act.” Major- ity Op. 16. There is no basis for concluding that a “rea- sonable expectations” test has supplanted the Will vesting rule as the governing test. Certainly no decision of the Supreme Court has shifted the governing principle from vesting to reasonable expectations. There is not even a claim that subsequent decisions of the Court have some- how “undermine[d] the reasoning” of Will. United States v. Hatter, 532 U.S. 557, 571 (2001) (quoting Will, 449 U.S. at 227 n.31) (internal quotation marks omitted). And even if Will had been undermined, it would not be this court’s prerogative to overrule it. See id. at 567 (noting that because Evans had been undermined but not yet “expressly overrule[d],” the Federal Circuit “was correct in applying Evans” and thereby “invit[ing] us to reconsider” it). So too our job is to follow the holding of Will, not to confine it to its facts. Numerous Supreme Court deci- sions, and our own decisions, have made this clear. As the Supreme Court held in Thurston Motor Lines, Inc. v. Jordan K. Rand, Ltd., a Court of Appeals must not “con- fus[e] the factual contours of [Supreme Court precedent] for its unmistakable holding” in an effort to reach a “novel interpretation” of that precedent. 460 U.S. 533, 534-35 (1983) (per curiam). See also, e.g., Marmet Health Care Ctr., Inc. v. Brown, 132 S. Ct. 1201, 1202 (2012) (per curiam) (a state court “misread[] and disregard[ed] the precedents of this Court” when it held the Federal Arbi- tration Act’s scope to be “more limited than mandated by this Court's previous cases”); Ariad Pharms., Inc. v. Eli Lilly & Co., 598 F.3d 1336, 1347 (Fed. Cir. 2010) (en banc) (“As a subordinate federal court, we may not so easily BEER v. US 10 dismiss [the Supreme Court’s] statements as dicta but are bound to follow them.”). The fact that three Justices of the Court, dissenting from a denial of certiorari, opined that Will might be distinguished from Williams is not authoritative. See Williams, 535 U.S. at 917 (Breyer, J., joined by Scalia & Kennedy, JJ., dissenting). A dissent from a denial of certiorari cannot “destroy[] the precedential effect” of a prior opinion. Teague v. Lane, 489 U.S. 288, 296 (1989). This court has recognized that neither the agreement of three dissenting Justices, nor the approval of their rea- soning by concurring Justices in later cases, can “trans- form a dissent into controlling law.” Prometheus Labs., Inc. v. Mayo Collaborative Servs., 628 F.3d 1347, 1356 n.2 (Fed. Cir. 2010), rev’d on other grounds, Mayo Collabora- tive Servs. v. Prometheus Labs, Inc., 132 S. Ct. 1289 (2012). In short, neither the dissent from denial of certiorari in Williams nor the Supreme Court’s remand in this case can be read as an invitation for this court to perform reconstructive surgery on Will. The Supreme Court may distinguish its own opinions by limiting them to their facts, see, e.g., Williams v. Illinois, 132 S. Ct. 2221, 2242 n.13 (2012), or choose to overrule them, see, e.g., Hatter, 532 U.S. at 567, but that is not an option for this court. We respectfully dissent. 8 8 Appellants also argue that the 2007 and 2010 COLAs were improperly withheld because no blocking legislation was enacted in those years, and Section 140, as amended in 2001, was either inapplicable or unconstitu- tionally discriminated against federal judges under the Supreme Court’s decision in Hatter. While we agree that this issue is not resolved by Will, these statutory and constitutional arguments were not properly raised below, and we decline to address them here. United States Court of Appeals for the Federal Circuit __________________________ PETER H. BEER, TERRY J. HATTER, JR., RICHARD A. PAEZ, LAURENCE H. SILBERMAN, A. WALLACE TASHIMA AND U. W. CLEMON, Plaintiffs-Appellants, v. UNITED STATES, Defendant-Appellee. __________________________ 2010-5012 __________________________ Appeal from the United States Court of Federal Claims in No. 09-CV-037, Senior Judge Robert H. Hodges, Jr. __________________________ O’MALLEY, Circuit Judge, with whom MAYER and LINN, Circuit Judges, join, concurring. I join the majority, both in the judgment it reaches and in its reasoning. I write separately to address two issues. First, I write to explain why I believe that, if United States v. Will, 449 U.S. 200 (1980), must be read as broadly as the dissent and the Williams v. United States, 240 F.3d 1019 (Fed. Cir. 2001) majority believes it must, then Will was wrong and the Supreme Court should say so. Second, I write because I believe that, whatever its BEER v. US 2 current statutory reach, Section 140 is unconstitutional and Congress can no longer rely on it to stagnate judicial compensation. I I first turn to Will. I agree with the majority that Will did not reach the issue presented here and, thus, does not dictate the result we may reach today. The position taken by the dissent, and by the Williams majority before it, is not without some force, however. One cannot deny that the adjudicatory principles upon which they rely are important ones, even if the majority concludes they are not determinative here. If the dissent is correct that we are forced to glean sweeping Compensation Clause princi- ples from Will governing all forms of statutory enact- ments designed to increase judicial pay, we must also be forced to conclude that Will’s analysis is flawed, both jurisprudentially and constitutionally. A. Jurisprudentially I find several aspects of the Will decision problematic. First, a close look at the facts and reasoning in Will reveals its internal inconsistency; neither its analysis nor its ultimate conclusion matches the facts presented. Specifically, while the Court in Will initially characterized the statutory scheme at issue there as “automatic,” 449 U.S. at 223, it later justified its Compensation Clause holding by characterizing congressional action blocking salary increases under the scheme as merely modifying “the formula” by which “future” increases were to be calculated. Id. at 227-28. Next, if the language employed in Will is meant to set down a “vesting” principle applica- ble in all Compensation Clause challenges, I believe the Court both: (1) violated the long-standing principle that courts are to decide only the cases before them and must only reach constitutional issues if and to the extent neces- 3 BEER v. US sary; and (2) landed upon a holding that, taken to its logical extreme, creates absurd results. 1. Use of the Term “Automatic” As the majority notes, the statutory scheme at issue in Will – the Executive Salary Cost-of-Living Adjustment Act of 1975, Pub. L. 94-82, 89 Stat. 419 (Aug. 9, 1975) (“the Adjustment Act”) – was a complex scheme, fraught with discretion and uncertainty. Despite this, Will char- acterized the Adjustment Act as a pay adjustment scheme which contemplated “automatic” pay increases. At issue in Will was the constitutionality of Congress’s decision to enact statutes preventing high-level Executive, Legisla- tive, and Judicial officials, including Article III judges, from receiving COLAs in four consecutive years where General Schedule federal employees received increases. The Court noted that these blocking statutes were de- signed to “stop or to reduce previously authorized cost-of- living increases initially intended to be automatically operative” under the Adjustment Act. Will, 449 U.S. at 203 (emphasis added). The Court then phrased the question presented in Will as: “when, if ever, does the Compensation Clause prohibit the Congress from repeal- ing salary increases that otherwise take effect automati- cally pursuant to a formula previously enacted?” Id. at 221 (emphasis added). As the majority notes, it is hard to understand the Court’s use of the term automatic in the context of the Adjustment Act. Normally, to say something is “auto- matic” is to say it occurs involuntarily or without further debate. See Oxford English Dictionary def. A(1); A(7)(a) (3d ed. June 2011; online version June 2012); see also American Heritage Dictionary 121 (5th ed. 2011) (def. 2a: defining “automatic” as “[a]cting or done without volition or conscious control; involuntary”). Nothing about the BEER v. US 4 judicial salary adjustments at issue in Will was “auto- matic,” however. To the contrary, the adjustments at issue in Will were based on civil service salary adjustments that were en- tirely discretionary. As explained by the majority, whether federal employees would receive a COLA, and in what amount, depended on the initial recommendations of an adjustment agent which were then subject to review by an Advisory Committee, the President, and Congress. This procedure hardly can be described as one that occurs involuntarily. In addition, the statutes setting forth future COLAs were “neither definite nor precise,” and nothing provided that adjustments would be calculated “in a mechanical way.” Williams v. United States, 535 U.S. 911, 917 (2002) (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Be- cause the statutory scheme under the Adjustment Act “was imprecise as to amount and uncertain as to effect,” the Court’s characterization of the increases under the Adjustment Act as “automatic” is difficult to follow. See id. The dissent explains the Court’s mischaracterization of the Adjustment Act’s pay scheme by noting that, for the years in question in Will, the statutory scheme had run its course and resulted in a recommended salary increase by the time Congress acted to block those increases. This, the dissent seems to suggest, explains why the Supreme Court used the term “automatic” to describe what was before it. While that argument has a certain logic to it, it does not explain why the Court’s constitutional analysis focused on the absence of a guarantee under the Adjust- ment Act. According to the Supreme Court, the Adjustment Act did not “alter the compensation of judges; it modified only 5 BEER v. US the formula for determining that compensation.” Will, 449 U.S. at 227 (emphases in original). And, the Court said that the blocking statutes merely represented a decision to “abandon” that “formula.” It then admonished that, “[t]o say that the Congress could not alter a method of calculating salaries before it was executed would mean the Judicial Branch could command Congress to carry out an announced future intent as to a decision the Constitu- tion vests exclusively in the Congress.” Id. at 228 (em- phasis added). It was on this reasoning that the Court concluded that a salary increase does not “vest” for Com- pensation Clause purposes until it becomes part of a judge’s compensation that is due and payable and that Congress had not violated the Compensation Clause when it did not allow certain increases under the Adjustment Act to “vest.” Thus, the Court explained its Compensation Clause decision in Will by saying it was only dealing with a formula regarding an expressed “future intent” to provide increases; the Court did not say at that point that it was addressing increases that had already been decided upon. More importantly, it did not say it was addressing definite increases that had been promised by operation of law; in explaining its assessment of the Act vis-à-vis the Com- pensation Clause, the Court spoke of the scheme under the Adjustment Act as one that promised no more than potential adjustments. And, in discussing the concept of vesting, the Court seemed to back away from the notion that it was dealing with anything one could consider “automatic” in the common sense of that word. How can an increase occur “automatically” if a right to it had not yet “vested”? While I understand why the dissent believes we must assume the Supreme Court meant what it said when it described the Adjustment Act increases as “automatic” BEER v. US 6 ones, that assumption would mean that the Court’s description of the facts presented had little correlation with its reasoning for why those facts did not run afoul of the Compensation Clause. 2. Constitutional Avoidance Next, if we read Will as broadly as Williams did, and the dissent now does, we must assume that, in Will, the Supreme Court violated its own well-established principle of constitutional avoidance. The Supreme Court has long- recognized that “[j]udging the constitutionality of an Act of Congress is ‘the gravest and most delicate duty that this Court is called upon to perform.’” Citizens United v. Fed. Election Comm’n, 558 U.S. 310, 130 S.Ct. 876, 917-18 (2010) (Roberts, C.J., concurring) (quoting Blodgett v. Holden, 275 U.S. 142, 147-48 (1927) (Holmes, J., concur- ring)). The Court’s standard practice, therefore, has been to “refrain from addressing constitutional questions except when necessary to rule on particular claims before [it].” Id. at 918 (citing Ashwander v. TVA, 297 U.S. 288, 346-48 (1936) (Brandeis, J., concurring)). In furtherance of this practice, it has long been the rule that courts should “not ‘formulate a rule of constitutional law broader than is required by the precise facts to which it is to be applied.’” Ashwander, 297 U.S. at 347 (quoting Liverpool, New York & Philadelphia S. S. Co. v. Commissioners of Emigration, 113 U.S. 33, 39 (1885)); see also United States v. Raines, 362 U.S. 17, 21 (1960) (same). Applying this principle in Citizens United, Chief Jus- tice Roberts explained that the Court’s “standard practice of avoiding broad constitutional questions except when necessary” gives rise to an “order of operations,” whereby the Court considers the narrowest claim first before proceeding, if necessary, to any broader claims. 130 S.Ct. at 918. Only if there is no valid narrow constitutional 7 BEER v. US ground available, should the court resolve any broader constitutional question. See id. If we assume that Will is to be read so broadly as to control the result under the very different set of facts presented here, we must also assume the Court spoke to a question not before it. The constitutional question prop- erly raised in Will was whether, under the specific statu- tory scheme set out in the Adjustment Act, the four blocking statutes at issue diminished judicial pay in violation of the Compensation Clause. A fair reading of Will based on “the precise facts to which it [was] applied,” requires limiting the holding to the statutory scheme that was before the Court. See Ashwander, 297 U.S. at 347 (Brandeis, J., concurring) (citation omitted); see also Raines, 362 U.S. at 21. If Will is read to address a ques- tion broader than that presented – one that would govern a host of different congressional efforts to protect judicial pay from diminution in value – then we must conclude that, in Will, the Supreme Court ignored its own govern- ing jurisprudential principles. In its briefing, the government concedes that there was a narrower approach the Court could have taken. Specifically, the government argues that, “even if the Supreme Court in Will could have based its decision upon the ‘discretionary’ character of the then-applicable statu- tory scheme, the Court did not decide the case upon that ground. The Court drew no such distinction.” Appellee’s Br. 26-27. If the government is right on this point, it is the very reason why Will was wrong to make the pro- nouncements upon which the government now relies. If the Court in Will consciously chose not to draw a distinc- tion between a discretionary COLA scheme and a self- executing, non-discretionary one, it: (1) formulated a rule of constitutional law broader than required by the facts presented; and (2) ignored the fundamental precept that BEER v. US 8 judges decide only the cases before them. See Hein v. Freedom from Religion Found., Inc., 551 U.S. 587, 615 (2007) (“Relying on the provision of the Constitution that limits our role to resolving the ‘Cases’ and ‘Controversies’ before us, we decide only the case at hand.”) 3. Absurd Results Finally, the definition of “vesting” Williams gleaned from Will cannot be right. If it were: (1) Congress could do away with judicial retirement benefits for all sitting judges; (2) it would be inconsistent with the way the concept of vesting has been applied to similar pay in- creases for Members of Congress; and (3) it would run afoul of the common law understanding of the way in which future interests “vest” for all other purposes. It necessarily would lead to absurd results. First, if the definition of “vesting” Williams felt bound to under Will is correct, then Congress could eliminate judicial retirement pay for all sitting Article III judges without violating the Compensation Clause. By statute, Article III judges can retire with full pay once they reach a certain combination of age plus years of judicial service. See 28 U.S.C. § 371. Under this system, the Supreme Court has said that the right to receive retirement pay “d[oes] not vest until retirement” and the “system pro- vide[s] nothing for a judge who le[aves] office before age 65.” United States v. Hatter, 532 U.S. 557, 575 (2001). In other words, the Supreme Court has specifically held that retirement benefits do not vest until a judge retires and certain prerequisites are met. In Will, the Court concluded that vesting occurs when a salary increase “takes effect as part of the compensation due and payable to Article III judges.” 449 U.S. at 229. As such, for those years where the COLAs at issue in Will had not yet become “due and payable,” the Court held 9 BEER v. US that the blocking statutes did not violate the Compensa- tion Clause’s prohibition against diminishing judicial pay. See id. If we accept Will’s holding that Congress can abolish judicial salary adjustments at any time before they take effect, it logically follows that Congress would also be free to abolish judicial retirement pay at any time. The practical consequences of Will would place judicial retirement benefits at risk, despite the fact that the Supreme Court itself previously has characterized such benefits as “compensation” under Article III. See Hatter, 532 U.S. at 574 (“the noncontributory pension salary benefits [are] themselves part of the judge’s compensa- tion”). Second, Will’s definition of vesting conflicts with the way in which that concept has been applied in the context of the Twenty-Seventh Amendment. In Boehner v. Ander- son, 30 F.3d 156 (D.C. Cir. 1994), the court addressed whether the 1989 Act (which also applies to Members of Congress) was inconsistent with the Twenty-Seventh Amendment which provides that: “No law, varying the compensation for services of the Senators and Represen- tatives, shall take effect until an election of Representa- tives shall have intervened.” Id. at 159. The court held that the phrase “shall take effect” in the Amendment referred to the date the Ethics Reform Act first became operative – i.e., 1991 – rather than any earlier or later point in time. See id. at 161-62. Because the COLA provision of the Ethics Reform Act took effect in January 1991, after an intervening election in 1990, that provision did not violate the Twenty-Seventh Amendment. Id. at 162. The court also held that: (1) Congress is free to specify a formula for future and continuing salary in- creases; and (2) the COLAs under the 1989 Act were designated to occur automatically each year after 1991, with no additional law necessary. Id. at 162-63. All BEER v. US 10 yearly COLAs beyond 1990 thus became operative and “vested” for Members of Congress when the law was first effective in 1991. 1 In Williams, the appellee-judges relied on the holding in Boehner to contend that the COLA increases for judi- cial officers took effect, or vested, when the law was effective, not when the yearly COLAs became due and payable. Williams, 240 F.3d at 1036. This court recog- nized the holding in Boehner, but distinguished it on grounds that it dealt with a different question limited to Members of Congress. Specifically, the court found that Boehner “has no relevance . . . to the question of whether the judicial pay aspects of the 1989 Act could, consistent with Article III, be revised or abrogated by later Acts of Congress.” Id. at 1037. That question, the Williams court held, was already answered in the affirmative in Will’s holding that “vesting, for federal judges under Article III, occurs only when compensation begins to accrue to the judges, not when a particular adjustment formula is enacted.” Id. at 1036-37. By simply relying on Will to distinguish Boehner, the court in Williams avoided the more difficult task of trying to reconcile two contradictory approaches to what vesting means under the Constitu- tion. We are now faced with two distinct definitions of the constitutionally effective date of congressionally enacted COLAs. While Will provides that, for Article III purposes, 1 In the alternative, the appellant in Boehner ar- gued that, if the court found the COLA provision vested and constitutional, then a later-enacted statute that cancelled a planned COLA absent an intervening election violated the Twenty-Seventh Amendment. 30 F.3d at 162. Although the answer to that question would be of interest to us now, the court declined to address it. See id. at 162-63. 11 BEER v. US a COLA is effective when it becomes “due and payable,” regardless of when the law establishing that COLA was enacted or when it took effect, Boehner states that, for Article I and the Twenty-Seventh Amendment, a COLA vests when the law is first effective, even if not due and payable for years to come. Common sense and basic principles of interpretation counsel against drawing this distinction. While it is certainly true that the operative date of congressionally designated salary increases is not pre- scribed in the Constitution, both the Compensation Clause and the Twenty-Seventh Amendment address the Framers’ concerns with in-term salary changes for the respective branches of government – one with decreases in-term and the other with increases in-term. I see no reason why the concept of vesting should be employed in a way to expand Congress’s ability to decrease judicial salaries under the Compensation Clause and be reframed under the Twenty-Seventh Amendment so as to expand Congress’s ability to increase its own. Finally, the vesting rule articulated in Will is an out- lier. As this court in Williams correctly noted, “[t]ypically, ‘vesting’ of future interests only requires two components: an identification of the future owner, and certainty that the property would transfer.” 240 F.3d at 1032 (citing 2 Blackstone Commentaries 168; Simes & Smith, The Law of Future Interests, § 65, pp. 54-55 (2nd ed. 1956)). This view of vesting of future interests is “more consistent with black-letter [law].” See id. at 1038. The Supreme Court, nevertheless, “departed from traditional vesting rules” for future interests and announced a peculiar “actual posses- sion” rule for Article III. Id. at 1032. Will ignored the standard rule for vesting of future interests and created a unique rule solely for judicial compensation. See id. at 1038. Despite recognition of its illogic, the Williams panel BEER v. US 12 felt compelled to reject the use of traditional vesting rules for Compensation Clause purposes because it found those rules to be “simply contrary to the rule established by the Supreme Court in Will.” Id. at 1033. 2 If we are to believe that Will advanced such an ex- treme vesting rule – one applicable only to the Compensa- tion Clause – then the Court should reexamine that rule and correct its mistake. Had the Supreme Court in Will applied the generally-accepted rule for vesting of future interests to the Adjustment Act, the same one the Boehner court applied to congressional pay increases, then a COLA whose formula was codified by law would vest, at an absolute minimum, once the amount of the COLA was established for a particular year. This ap- proach is grounded in “sound equitable principle[s]” and, as we recognized in Williams, has deep common-law roots. See id. at 1032-33. For the reasons explained in further detail below, as the majority has noted, a more reasonable, consistent, and logical definition of “vesting” under Article III should be governed by the “reasonable expectations” of sitting 2 Indeed, despite awareness of Will, various state courts interpreting analogous provisions of their own constitutions have held that the failure to provide statutorily promised COLAs unconstitutionally dimin- ishes judicial compensation. See e.g., Jorgensen v. Blago- jevich, 811 N.E.2d 652, 664 (Ill. 2004) (noting that the standards for conferring and calculating COLAs, which “were formulated following the United States Supreme Court’s decision in Will, expressly provided that COLAs were to be given on July 1, 1991, and on July 1 of each year thereafter and that such COLAs were to be consid- ered a component of salary fully vested at the time the Compensation Review Board’s report became law”). Will’s “vesting” rule for Compensation Clause challenges – if that is really what it is – stands alone. 13 BEER v. US judicial officers. Put simply, if we are to read Will as broadly as Williams did, and the dissent now does, the Court should revisit Will’s unique vesting rule. B. Constitutionally If Will truly established an “actual possession” vesting rule for Compensation Clause purposes, that holding seems indefensible under the Constitution. The Framers formulated the Compensation Clause for the express purpose of maintaining judicial independence, in part by providing judges with reasonable expectations about their pay and the inability of Congress to reduce it. As inter- preted in Williams, the Will rule defeats the Framers’ intent and threatens the governmental structure around which the Constitution was formulated. 1. Historical Perspective and the Framers’ Intent The Compensation Clause “has its roots in the long- standing Anglo-American tradition of an independent Judiciary.” Will, 449 U.S. at 217. As the Supreme Court has recognized, the “colonists had been subjected to judicial abuses at the hand of the Crown, and the Fram- ers knew the main reasons why: because the King of Great Britain ‘made Judges dependent on his Will alone, for the tenure of their offices, and the amount and pay- ment of their salaries.’” Stern v. Marshall, 131 S. Ct. 2594, 2609 (2011) (quoting the Declaration of Independ- ence, para. 11). Against this backdrop, the Framers designed Article III to protect the public “from a repeat of those abuses.” Id. By giving judges life tenure and pre- venting the other branches from reducing judicial com- pensation, the Framers sought to “preserve the integrity of judicial decisionmaking.” Id. As the majority notes, in Federalist 79, Alexander Hamilton emphasized the importance of protecting judi- BEER v. US 14 cial compensation. Specifically, he argued that, “[n]ext to permanency in office, nothing can contribute more to the independence of the judges than a fixed provision for their support.” The Federalist No. 79 at 385 (Alexander Hamil- ton) (Lawrence Goldman ed., 2008). Hamilton observed that, “[i]n the general course of human nature, a power over a man’s subsistence amounts to a power over his will.” Id. at 386 (emphasis in original). For this reason, the legislative branch must not “change the condition[s] of the [judiciary] for the worse” so that “[a] man may then be sure of the ground upon which he stands, and can never be deterred from his duty by the apprehension of being placed in a less eligible situation.” Id. Hamilton’s concerns, and those of many other Fram- ers, were not merely academic. Indeed, throughout the former colonies, legislatures took retributive actions against judges with whom they disagreed, including attempts to remove judges who declared particular laws unconstitutional and to call judges before the legislature to answer for specific rulings. See Julius Goebel, Jr., Antecedents and Beginnings to 1801, in 1 History of the Supreme Court of the United States, 133-42 (Paul A. Freund ed., 1971). These events further supported the founders’ desire to insulate judges from the influence and control of the other branches of government. The Supreme Court has recognized that the primary purpose of the prohibition against reducing judicial sala- ries is “not to benefit the judges, but . . . to promote that independence of action and judgment which is essential to the maintenance of the guaranties, limitations, and pervading principles of the Constitution.” Evans v. Gore, 253 U.S. 245, 253 (1920), overruled on other grounds by Hatter, 532 U.S. at 571. The Compensation Clause should be “construed, not as a private grant, but as a limitation imposed in the public interest.” Id. It is the public that 15 BEER v. US benefits from a strong, independent judiciary that is free to issue decisions without fear of repercussion. The Framers’ desire to insulate judicial pay from the political process was the subject of much debate and angst. While, given the long tenure judges would be asked to serve, there was no doubt some provision should be made for salary increases, the Framers also feared that, if salary decisions were left entirely to Congress, the judiciary might be forced to curry favor with Congress to secure reasonable compensation increases. See Jonathan L. Entin & Erik M. Jensen, Taxation, Compensation, and Judicial Independence, 56 Case W. Res. L. Rev. 965, 972 (2006). To address this concern, James Madison sug- gested indexing judicial pay to the price of wheat or another stable value. The Framers rejected that idea, however, for fear fluctuations in commodity prices, like inflation, might leave judges undercompensated. See 2 The Records of the Federal Convention of 1787 44-45 (Max Farrand ed., 1911). Thus, while the Framers foresaw a need for in-term increases in judicial salaries and were concerned with leaving the task of providing those increases to Congress, they saw no alternative; no self-executing system they could devise seemed adequate to ensure that, given the dual effects of inflation and rising standards of living, judges would not be left undercompensated. So trust Congress they did, leaving to it the responsibility to guard against real decreases in judicial salary by future legisla- tive enactments. In sum, the Framers intended to provide judges rea- sonable expectations about their pay. The Framers, to be sure, did not contemplate that a judges’ reasonable expec- tation would mean that he or she would become wealthy by taking the bench, or that Congress necessarily would BEER v. US 16 increase judicial salaries. They believed, however, that Congress would assess fairly and periodically the need for increases in judicial compensation, would provide in- creases when appropriate, and that, once it did so, judicial officers thereafter could rely on the fact that Congress could not take such increases away. 2. The Expectations Approach in Practice Courts have long-endorsed this expectations-based approach to the Compensation Clause. Indeed, as Justice Breyer has noted, protecting “a judge’s reasonable expec- tations” is the “basic purposive focus” of the Compensa- tion Clause. Williams, 535 U.S. at 916 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Likewise, Justice Scalia has argued that, when Congress takes away a previously-established component of the federal judicial “employment package,” it reduces compensation and thereby thwarts judicial expectations. See Hatter, 532 U.S. at 585 (Scalia, J., dissenting) (arguing that repeal of federal judges’ exemp- tion from the Medicare tax was a reduction of compensa- tion because those judges “had an employment expectation of a preferential exemption from taxation”). Consistent with this expectations-related focus, the Supreme Court has held that the Compensation Clause forbids laws “which by their necessary operation and effect withhold or take from the judge a part of that which has been promised by law for his services.” O’Donoghue v. United States, 289 U.S. 516, 533 (1933) (quoting Evans v. Gore, 253 U.S. 245, 254 (1920)). Other courts likewise have emphasized judicial expec- tations in their approach to the Compensation Clause. For example, in the early nineteenth century, the Circuit Court for the District of Columbia held that, “if [a judge’s] compensation has once been fixed by law, a subsequent 17 BEER v. US law for diminishing that compensation . . . cannot affect [a sitting judge].” United States v. More, 7 U.S. (3 Cranch) 159, 160 n.2 (1805), writ of error dism’d for want of juris- diction. In More, Congress had enacted and later abol- ished a system of fees for compensating justices of the peace in the District of Columbia. Id. One of the justices of the peace continued to charge fees under the abolished structure, and the government brought an indictment against him. Id. On appeal, the Circuit Court held that: (1) the compensation of justices of the peace was subject to the Compensation Clause; and (2) where a fee structure is set by law, a later-enacted statute diminishing or abolishing that structure violated the Constitution. Id. at 161. Because sitting justices had an expectation that they would receive compensation consistent with the then- existing fee structure, Congress could not take that struc- ture away. In Will, the Supreme Court discarded the long- standing expectations-based approach to the Compensa- tion Clause in favor of its “due and payable” vesting rule, without clear explanation for doing so. In a terse foot- note, the Court distinguished More. See Will, 449 U.S. at 228, n.32. Specifically, the Court claimed that, in More, “the fee system was already in place as part of the jus- tices’ compensation when Congress repealed it” whereas “the increase [via the Adjustment Act] in Year 2 had not yet become part of the compensation of Article III judges” when it was repealed. Id. Careful consideration of the facts in More reveal that this is a distinction without a difference. The justices under the fee system in More were not entitled to compensation until they actually rendered services. See More, 7 U.S. at 160 n.2 (“This compensation is given in the form of fees, payable when the services are rendered.”). At all times, the justices knew the precise amount they could charge for a particu- BEER v. US 18 lar service, but they never knew how much their total compensation would be, for example, in a particular week. In other words, the fee system in More merely set out a structure for calculating the compensation, which was not “due and payable” – to use the Court’s terminology in Will – until the justices performed the affirmative act of ren- dering services. The Adjustment Act formula was no different. In the same way that the justices under the fee system in More did not know how much they would work in a particular year, under the Adjustment Act, Article III judges did not know how much their salary would increase in a particu- lar year, if at all. But they did know that, once the for- mula was enacted for the year, it became part of the compensation due. For example, looking at Year 3 in Will, if we accept the dissent’s proposition that the COLA of 5.5% became automatic once the President’s alternative plan was adopted and transmitted to Congress – which was one month before the Year 3 blocking statute was enacted – then there is no doubt that, as was the case in More, the COLA “was already in place as part of the [judges’] compensation when Congress repealed it.” See Will, 449 U.S. at 228, n.32 (citing More, 3 Cranch at 161). In the same way that Congress was prohibited from abolishing the fee structure in More because it was part of the justices’ compensation, so too should Congress have been prohibited from blocking the COLA for Year 3 in Will. Given these similarities, Will’s dismissal of More is unconvincing. The two opinions are irreconcilable. Either Will is incorrect, or the Court should have said that More was wrong. The Supreme Court should return to the well-established expectations-based approach to the Compensation Clause. 19 BEER v. US 3. The Consequences of Abandoning the Expectations Approach Assuming Will’s vesting rule allows Congress to bar “automatic” COLAs promised by definitive and precise legislative enactment, that rule is contrary to the consti- tutional balance the Framers carefully calibrated – one which, of necessity, delegated control over judicial salaries to the legislature, but did so in a way to guard against congressional retribution for unpopular judicial decisions. So understood, Will’s vesting rule puts at risk the princi- ples the Framers struggled so hard to foster; it threatens to make the judiciary beholden to Congress in ways which undermine its independence. The Supreme Court should rethink such a rule. See e.g., Mistretta v. United States, 488 U.S. 361, 383 (1989) (encouraging vigilance against a “provision of law” that “impermissibly threatens the institutional integrity of the Judicial Branch”) (quoting Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 851 (1986)). The Framers’ concerns were prescient. Statistics demonstrate that the erosion of judicial pay “has reached the level of a constitutional crisis that threatens to un- dermine the strength and independence of the federal judiciary.” Chief Justice John G. Roberts, Jr., 2006 Year- End Report on the Federal Judiciary, 39 The Third Branch 1, 1 (2007). Not only is this not the world the Framers contemplated, it is approaching one they most feared. As Hamilton explained, if judicial independence is “destroyed, the constitution is gone, it is a dead letter; it is vapor which the breath of faction in a moment may dissi- pate.” Commercial Advertiser (Feb. 26, 1802) (reprinted in The Papers of Alexander Hamilton, Volume XXV 525 (Columbia University Press 1977)). BEER v. US 20 III I finally turn to Section 140 of Pub. L. No. 97-92, 95 Stat. 1183, 1200 (1981), and its role in our assessment of the legality of the congressional action challenged here. I agree with the majority that the existence of Section 140 does not change the conclusion that the failure to provide COLAs mandated by the 1989 Act is unconstitutional, whether the withholding occurred before or after Con- gress amended that section in 2001. As the majority explains, by its own terms, Section 140 is not applicable to the salary adjustments contemplated by the 1989 Act. If it were, however, as the government contends it is, we could not enforce it because Section 140 is unconstitu- tional. Section 140 provides as follows: Notwithstanding any other provision of law or of this joint resolution, none of the funds appropriated by this joint resolution or by any other Act shall be obligated or expended to increase, after the date of en- actment of this resolution, any salary of any Federal judge or Justice of the Su- preme Court, except as may be specifically authorized by Act of Congress hereafter enacted . . . . Pub. L. No. 97-92, § 140, 95 Stat. 1183, 1200 (1981). Section 140 was a rider to a Joint Resolution providing continuing appropriations for fiscal year 1982. In Wil- liams, we held that the government could not rely on Section 140 as justification for the blocking statutes passed in 1995, 1996, 1997, and 1999 because Section 140 expired by its own terms on September 30, 1982. Wil- liams, 240 F.3d at 1026 (citing Pub. L. No. 97-161, 96 Stat. 22 (1982) (extending life of provisions from March 21 BEER v. US 31, 1982 to September 30, 1982); Pub. L. No. 97-92, § 102(c), 95 Stat. 1183 (1981)). After Williams, Congress enacted legislation that amended Section 140 to provide that it “shall apply to fiscal year 1981 and each fiscal year thereafter.” Act of Nov. 28, 2001, Pub. L. No. 107-77, § 625, 115 Stat. 803 (“2001 amendment”). Today, the majority assumes that the 2001 amendment supersedes Williams’s holding that Section 140 expired, but agrees with the alternative holding in Williams that, even if not expired, the 1989 Act provides the additional authorization required by Section 140. Were the majority’s conclusion on that point not cor- rect, then we would be forced to conclude that Section 140 violates the Compensation Clause, both because it singles out Article III judges for disadvantageous treatment and because it violates the principle of separation of powers. A. Section 140’s Discriminatory Effect The Supreme Court has held that a law violates the Compensation Clause when it “effectively single[s] out . . . federal judges for unfavorable treatment” in their com- pensation. Hatter, 532 U.S. at 559. In Hatter, the Court struck down a statutory scheme that required sitting federal judges to pay into the Social Security system while other high-level government officials potentially were exempt from making such payments. Id. at 564, 572-73. In finding the denial of the exemption to judges unconsti- tutional, the Court explained that the “practical upshot” of the statutory scheme was to disadvantage judges relative to “nearly every current federal employee.” Id. at 573. 3 3 Justice Scalia did not join in this portion of the Court’s opinion, concurring on grounds that the BEER v. US 22 Section 140 is no different. It only overrides the automatic annual COLAs promised in the 1989 Act for judicial officers. All other federal employees – including high ranking Executive Branch appointees and Members of Congress – remain entitled to those “automatic” ad- justments. Only judicial officers are beholden to Congress for an additional affirmative legislative enactment before they may receive the 1989 Act’s COLAs. Thus, post-2001, Section 140 turns the 1989 Act into a law that provides a financial benefit to all federal employees other than judges and puts the judiciary in the position of annually needing to “curry favor” with the legislature for compen- sation increases, just as the Framers feared. That clearly violates the Compensation Clause. See Hatter, 532 U.S. at 576; Williams, 535 U.S. at 911 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certio- rari) (“[Section 140] refers specifically to federal judges, and it imposes a special legislative burden upon their salaries alone. The singling out of judges must throw the constitutionality of the provision into doubt.”) (citing Hatter, 532 U.S. at 564)). “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.’” Hatter, 532 U.S. at 577 (quoting James Wilson, Lectures on Law (1791), in 1 Works of James Wilson 364 (J. Andrews ed. 1896)). Compensation Clause was violated because the congres- sional action violated the judicial officers’ reasonable expectations about their future income package. Hatter, 532 U.S. at 586 (Scalia, J., concurring in part and dissent- ing in part) (“I disagree with the Court’s grounding of this holding on the discriminatory manner in which the exten- sion occurred.”). The “discrimination” theory, however, received the votes of a majority of the Justices and, there- fore, is binding precedent. 23 BEER v. US The fear of disadvantageous treatment of judges un- der Section 140, as amended, is not hypothetical. Until recently, annual adjustments for federal judges remained in step with those for Executive Branch appointees and Members of Congress. When those groups received auto- matic adjustments under the 1989 Act, Congress also enacted the necessary special legislation to authorize an adjustment for judges. In fiscal year 2007, however, both General Schedule employees and Executive Branch appointees received an automatic adjustment under the 1989 Act, but Congress did not enact special legislation to adjust judicial salaries. The same thing happened in fiscal year 2010. Thus, the link between judicial salary adjustments and those for Executive Branch appointees was severed such that all nonelected federal employees other than Article III judges received COLAs in those years. 4 This is the very sort of individualized treatment of the judiciary that the Supreme Court has characterized as a “disguised legislative effort to influence the judicial will.” See Hatter, 532 U.S. at 571. Little could be more inconsistent with the Framers’ purpose and construct under the Compensation Clause. B. Section 140 and the Separation of Powers Section 140 separately poses a separation of powers problem because it conditions the award of COLAs to judges on the receipt of salary adjustments by Members of Congress. The government argues that, in enacting the 1989 Act, “Congress made clear its intent to maintain a system of salary parity among Federal judges, members of 4 Members of Congress did not receive salary ad- justments in 2007 or 2010 because they affirmatively chose to opt out of their right to receive them under the 1989 Act. That choice was theirs, however, and not one otherwise mandated by preexisting legislation. BEER v. US 24 Congress, and high-level Executive branch officers.” Appellee’s Br. 17 (citing Report of the Bipartisan Task Force on Ethics on H.R. 3660, Government Ethics Reform Act of 1989, 135 Cong. Rec. 30,756 (Nov. 21, 1989)). As noted above, any “parity” objective vis-à-vis Executive Branch officers has been abandoned. And, it is precisely because Congress has continued to use Section 140 to force a parity between judicial salaries and its own that Section 140 violates the principle of separation of powers. The concern with the independence of the judiciary is one which flows directly from the tripartite form of gov- ernment on which the Constitution is structured. In establishing the system of divided powers in the Constitu- tion, the Framers believed it was essential that “the judiciary remain[] truly distinct from both the legislature and the executive.” Stern, 131 S.Ct. at 2608 (quoting The Federalist No. 78, p. 466 (C. Rossiter ed. 1961) (A. Hamil- ton)). Accordingly, as the Supreme Court has noted, the Framers built into the Constitution “a self-executing safeguard against the encroachment or aggrandizement of one branch at the expense of the other.” Mistretta, 488 U.S. at 382 (quoting Buckley v. Valeo, 424 U.S. 1, 122 (1976)). Although the three branches “are not hermeti- cally sealed from one another,” Article III was designed to impose certain “basic limitations that the other branches may not transgress.” Stern, 131 S.Ct. at 2609 (citing Nixon v. Administrator of Gen. Servs., 433 U.S. 425, 443 (1977)). As noted earlier, the compromise the Framers struck under the Compensation Clause was one which would entrust to Congress the power and obligation to ensure reasonable salary adjustments for the judiciary over time. This was a compromise born of necessity, however; this mechanism for judicial salary adjustments was not meant to tie those adjustments to legislative salary changes, or 25 BEER v. US to make them dependent on prevailing political winds. The Framers certainly did not mean to use the Compen- sation Clause to blur the lines between the legislative and judicial branches. That is precisely what Section 140 does, however. Congress has used Section 140 to link judicial pay to its own, affirmatively authorizing judicial compensation increases thereunder only in years where Congress finds it politically palatable to allow increases in its own. By using Section 140 in this way, Congress has ignored its constitutional duty to assess independently the adequacy of judicial compensation. And, it has ignored the obliga- tion entrusted to it by the Framers to jealously guard the independence of the judiciary. “[W]hether the Judiciary is entitled to a compensation increase must be based upon an objective assessment of the Judiciary’s needs if it is to retain its functional and structural independence.” Ma- ron v. Silver, 925 N.E.2d 899, 914 (N.Y. 2010) (finding link between legislative and judicial pay increases uncon- stitutional under New York state constitution). Because Section 140 skirts Congress’s obligations un- der the Compensation Clause and undermines the inde- pendence of the judiciary, it is unconstitutional. The Supreme Court repeatedly has made clear that it is the laws that “threaten[] the institutional integrity of the Judicial Branch” that violate the principle of separation of powers. Mistretta, 488 U.S. at 383 (quoting Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 851 (1986)). Under these well-established guideposts, Section 140 must fail. IV I agree with the majority that the failure to provide COLAs promised by the 1989 Act to the judiciary violates the Compensation Clause. I also agree that Will does not BEER v. US 26 dictate a contrary result. “General propositions do not decide concrete cases.” Lochner v. New York, 198 U.S. 45, 76 (1905) (Holmes, J., dissenting). The general concepts espoused in Will simply do not address the very concrete and different set of facts before us. If the Supreme Court concludes Will must be read as broadly as this Court felt forced to read it in Williams, however, Will must be overruled. To the extent Section 140 plays any role in the Court’s analysis of the issues presented here, moreover, the Supreme Court should address its constitutionality and put its use to rest. United States Court of Appeals for the Federal Circuit __________________________ PETER H. BEER, TERRY J. HATTER, JR., RICHARD A. PAEZ, LAURENCE H. SILBERMAN, A. WALLACE TASHIMA AND U. W. CLEMON, Plaintiffs-Appellants, v. UNITED STATES, Defendant-Appellee. __________________________ 2010-5012 __________________________ Appeal from the United States Court of Federal Claims in No. 09-CV-037, Senior Judge Robert H. Hodges, Jr. __________________________ WALLACH, Circuit Judge, concurring. I concur in the results, and in the reasoning of the de- cision, including the necessity of making this important determination that Congress may not exceed constitu- tional bounds in its relationship with the judiciary. I write separately only to clarify that this decision does not mean that any particular federal judge other than plain- tiffs will necessarily accept accrued back pay. 


[by Rader]

 Opinion for the court filed by Chief Judge RADER, in which Circuit Judges NEWMAN, MAYER, LOURIE, LINN, PROST, MOORE, O’MALLEY, REYNA and WALLACH join. Dissenting opinion filed by Circuit Judge DYK, in which Circuit Judge BRYSON joins. Concurring opinion filed by Circuit Judge O’MALLEY, in which Circuit Judges MAYER and LINN join. Concurring opinion filed by Circuit Judge WALLACH. RADER, Chief Judge. The Constitution erects our government on three foundational corner stones — one of which is an independent judiciary. The foundation of that judicial independence is, in turn, a constitutional protection for judicial compensation. The framers of the Constitution protected judicial compensation from political processes because “a power over a man’s subsistence amounts to a power over his will.” The Federalist No. 79, p. 472 (Alexander Hamilton) (Clinton Rossiter ed., 1961). Thus, the Constitution provides that “Compensation” for federal judges “shall not be diminished during their Continuance in Office.” U.S. Const, art. Ill, § 1 (“Compensation Clause”). This case presents this court with two issues involving judicial independence and constitutional compensation protections— one old and one new. First, the old question: does the Compensation Clause of Article III of the Constitution prohibit Congress from withholding the cost of living adjustments for Article III judges provided for in the Ethics Reform Act of 1989 (“1989 Act”)? To answer this question, this court revisits the Supreme Court’s decision in United States v. Will, 449 U.S. 200 , 101 S.Ct. 471 , 66 L.Ed.2d 392 (1980). Over a decade ago in Williams v. United States, 240 F.3d 1019 (Fed.Cir.2001) (filed with dissenting opinion by Plager, J.), a divided panel of this court found that Will applied to the 1989 Act and concluded that Congress could withdraw the promised 1989 cost of living adjustments. This court en banc now overrules Williams and *1177 instead determines that the 1989 Act triggered the Compensation Clause’s basic expectations and protections. In the unique context of the 1989 Act, the Constitution prevents Congress from abrogating that statute’s precise and definite commitment to automatic yearly cost of living adjustments for sitting members of the judiciary. The new issue involves pure statutory interpretation, namely, whether the 2001 amendment to Section 140 of Pub. L. No. 97-92 overrides the provisions of the 1989 Act. This court concludes the 1989 Act was enacted after Section 140, and as such, the 1989 Act’s automatic cost of living adjustments control. I. The 1989 Act overhauled compensation and ethics rules for all three branches of government. With respect to the judiciary, it contained two reciprocal provisions. On the one hand, the 1989 Act limited a federal judge’s ability to earn outside income and restricted the receipt of honoraria. On the other hand, the 1989 Act provided for self-executing and non-discretionary cost of living adjustments (“COLA”) to protect and maintain a judge’s real salary. The 1989 Act provides that whenever a COLA for General Schedule federal employees takes effect under 5 U.S.C. § 5303 , the salary of judges “shall be adjusted” based on “the most recent percentage change in the [Employment Cost Index] ... as determined under section 704(a)(1) of the Ethics Reform Act of 1989.” Pub. L. No. 101-194, § 704 (a)(2)(A), 103 Stat. 1716 , 1769 (Nov. 30, 1989). The Employment Cost Index (“ECI”) is an index of wages and salaries for private industry workers published quarterly by the Bureau of Labor Statistics. Section 704(a)(1) of the 1989 Act calculates COLAs by first determining the percent change in the ECI over the previous year. Id. at § 704(a)(1)(B). Next, the statutory formula reduces the ECI percentage change by “one-half of 1 percent ... rounded to the nearest one-tenth of 1 percent.” Id. However, no percentage change determined under Section 704(a)(1) shall be “less than zero” or “greater than 5 percent.” Id. While the 1989 Act states that judicial salary maintenance would only occur in concert with COLAs for General Schedule federal employees under 5 U.S.C. § 5303 , these General Schedule COLAs are automatic, i.e., they do not require any further congressional action. See 5 U.S.C. § 5303 (a). The only limitation on General Schedule COLAs is a presidential declaration of a “national emergency or serious economic conditions affecting the general welfare” making pay adjustments “inappropriate.” 5 U.S.C. § 5303 (b). Notwithstanding the precise, automatic formula in the 1989 Act, the Legislative branch withheld from the Judicial branch those promised salary adjustments in fiscal years 1995, 1996, 1997, and 1999. During these years, General Schedule federal employees received the adjustments under Section 5303(a), but Congress blocked the adjustments for federal judges. See Pub. L. No. 103-329, § 630 (a)(2), 108 Stat. 2382 , 2424 (Sept. 30, 1994) (FY 1995); Pub. L. No. 104-52, § 633 , 109 Stat. 468 , 507 (Nov. 19, 1995) (FY 1996); Pub. L. No. 104-208, § 637 , 110 Stat. 3009 , 3009-364 (Sept. 30, 1996) (FY 1997); Pub. L. No. 105-277, § 621 , 112 Stat. 2681 , 2681-518 (Oct. 21, 1998) (FY 1999). In response to these missed adjustments, several federal judges filed a class action alleging these acts diminished their compensation in violation of Article III. After certifying a class of all federal judges serving at the- time (including appellants) and without providing notice or *1178 opt-out rights, the district court held that Congress violated the Compensation Clause by blocking the salary adjustments. See Beer v. United States, 671 F.3d 1299, 1308-09 (Fed.Cir.2012); Williams v. United States, 48 F.Supp.2d 52 (D.D.C.1999). On appeal, this court reversed the district court’s judgment. See Williams, 240 F.3d at 1019 . This court opined that the Supreme Court’s decision in Will foreclosed the judges’ claim as a matter of law. Id. at 1033, 1035, 1040. According to this court, Will ruled that promised future salary adjustments do not qualify as “Compensation” protected under the Constitution until they are “due and payable.” Id. at 1032 (quoting Will, 449 U.S. at 228 , 101 S.Ct. 471 ). Thus, Congress enjoyed full discretion to revoke any future judicial COLAs previously established by law, no matter how precise or definite, as long as the adjustments had not yet taken effect. Id. at 1039. This court declined to hear the case en banc over the dissent of three judges. See 264 F.3d 1089 , 1090-93 (Fed. Cir.2001) (Mayer, C.J., joined by Newman and Rader, JJ.); id. at 1093-94 (Newman, J., joined by Mayer, C.J. and Rader, J.). The Supreme Court denied certiorari over the dissent of three Justices. See 535 U.S. 911 , 122 S.Ct. 1221 , 152 L.Ed.2d 153 (2002) (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Following this court’s decision in Williams, Congress amended a 1981 appropriations rider commonly known as Section 140. Section 140 originally read: Notwithstanding any other provision of law or of this joint resolution, none of the funds appropriated by this joint resolution or by any other Act shall be obligated or expended to increase, after the date of enactment of this joint resolution, any salary of any Federal judge or Justice of the Supreme Court, except as may be specifically authorized by Act of Congress hereafter enacted: Provided,. That nothing in this limitation shall be construed to reduce any salary which may be in effect at the time of enactment of this joint resolution nor shall this limitation be construed in any manner to reduce the salary of any Federal judge or of any Justice of the Supreme Court. Pub. L. No. 97-92, § 140 , 95 Stat. 1183 , 1200 (1981) (codified at 28 U.S.C. § 461 note) (emphasis added). While Section 140 originally expired in 1982, see Williams, 240 F.3d at 1026-27 , it was revived- by a 2001 amendment that added: “This section shall apply to fiscal year 1981 and each fiscal year thereafter.” Pub. L. No. 107-77, § 625 , 115 Stat. 748 , 803 (Nov. 28, 2001). Following the Section 140 amendment, Congress enacted legislation specifically allowing federal judges to receive the salary adjustments mandated by the 1989 Act in fiscal years 2002, 2003, 2004, 2005, 2006, 2008, and 2009. See Barbara L. Schwemle, Congressional Research Service, Legislative, Executive, and Judicial Officials: Process for Adjusting Pay and Current Salaries 2-4 (Feb. 9, 2011). For fiscal years 2007 and 2010, all General Schedule and Executive level federal employees received COLAs under 5 U.S.C. § 5303 (a), but federal judges received no adjustments. Congress did not affirmatively authorize judicial COLAs in those years and took the position that, because of the requirements of Section 140, judicial COLAs could not be funded.” The current case results from the combination of the blocking legislation of the 1990s and the amendment to Section 140. Appellants , are six current and former Article III judges, all of whom entered into federal judicial service before 2001. In January 2009, they filed a complaint in the United States Court of Federal Claims *1179 claiming that Congress violated the Compensation Clause by withholding the salary adjustments established by the 1989 Act. They claimed a deficit resulted not only from the withholding of COLAs in 2007 and 2010, but also the calculation of adjustments due in other years by reference to base compensation that did not include the amounts withheld in 1995, 1996, 1997, and 1999. For relief, they sought back pay for the additional amounts they allegedly should have received during the period covered by the applicable six-year statute of limitations. The Court of Federal Claims dismissed the complaint based on the Williams precedent. On appeal, this court summarily affirmed the judgment, stating that “Williams controls the disposition of this matter.” Beer v. United States, 361 Fed. Appx. 150, 151-52 (Fed.Cir.2010). The Supreme Court granted the subsequent petition for certiorari, vacated the judgment, remanded the case for “consideration of the question of preclusion,” and stated that “further proceedings ... are for the Court of Appeals to determine.” Beer v. United States, — U.S. -, 131 S.Ct. 2865 , 180 L.Ed.2d 909 (2011). Specifically, in opposing the petition for certiorari, the Government had argued that Appellants could not litigate anew the issue resolved in Williams because they had been absent members of the class action in Williams. Upon remand, this court unanimously concluded that Appellants were not precluded from bringing their Compensation Clause claims in the present case. Beer v. United States, 671 F.3d 1299, 1309 (Fed. Cir.2012). The district court in Williams had not provided Appellants with notice of the class certification. Thus they were not bound by the result of that earlier litigation. See id. at 1305-09. This court nonetheless continued to feel constrained by the ultimate conclusion in Williams and affirmed the Court of Federal Claims’ dismissal of the complaint. Id. at 1309. Subsequently, this court granted Appellants’ petition for rehearing en banc. 468 Fed. Appx. 995 (Fed.Cir.2012). II. This court has jurisdiction over the Court of Federal Claims’ dismissal of the Appellants’ complaint under 28 U.S.C. § 1295 (a)(3). This court reviews the decision to dismiss the complaint without deference. Hearts Bluff Game Ranch, Inc. v. United States, 669 F.3d 1326, 1328 (Fed. Cir.2012); Frazer v. United States, 288 F.3d 1347, 1351 (Fed.Cir.2002). This court en banc now turns its attention to two preliminary issues before addressing the merits of the appeal. First, judicial review of laws affecting judicial compensation is not done lightly as these cases implicate a conflict of interest. Will, 449 U.S. at 211-17 , 101 S.Ct. 471 . After all, judges should disqualify themselves when their impartiality might reasonably be questioned or when they have a potential financial stake in the outcome of a decision. See 28 U.S.C. § 455 (a). In Will , the Supreme Court applied the time-honored “Rule of Necessity” because if every potentially conflicted judge were disqualified, then plaintiffs would be left without a tribunal to address their claims. See Will, 449 U.S. at 213-17 , 101 S.Ct. 471 . The Rule of Necessity states that “although a judge had better not, if it can be avoided, take part in the decision of a case in which he has any personal interest, yet he not only may but must do so if the case cannot be heard otherwise.” Id. at 213 , 101 S.Ct. 471 (quoting F. Pollack, A First Book of Jurisprudence 270 (6th ed. 1929)) (emphasis added). This court relies on the Supreme Court’s complete analysis of the Rule of Necessity and concludes that this *1180 en banc court may, indeed must, hear the case. See id. at 211-18 , 101 S.Ct. 471 . On the other preliminary procedural question, this court deliberately limits the questions under review. To be specific, this court en banc does not overrule the Williams panel’s analysis of Section 140. See 240 F.3d at 1026-27 . Furthermore, it does not overrule the Beer panel’s analysis of preclusion. See 671 F.3d 1299 . This court adopts the prior panel’s analysis of the preclusion issue in toto. Now the court en banc proceeds to the old and new questions previously set forth. III. At the outset, this court must honor and address the Supreme Court’s decision in Will . As the Williams panel correctly noted, if Will resolves the validity of Congress’ decision to block the COLAs promised in the 1989 Act, then any remedy for salary diminution in this case lies not in this court but in the Supreme Court. See Williams, 240 F.3d at 1035 . However, if Will is inapplicable to the statutory scheme at play in this case, then this court has an obligation to resolve the issue. United States v. Will, supra, tested the validity of congressional blocking acts preventing COLAs provided for under the 1975 Adjustment Act (“1975 Act”). The 1975 Act purported to protect judicial salaries with adjustments calculated under an opaque and indefinite process. Section 5305, as in effect in 1975, directed the President to “carry out the policy stated in section 5301” when giving COLAs to General Schedule federal employees. 5 U.S.C. § 5305 (a) (1976). Section 5301 in turn articulated a four-fold policy for setting federal pay: (1) equal pay for equal work; (2) pay distinction based on work and performance distinctions; (3) comparable pay with private sector jobs for comparable work; and (4) interrelated statutory pay levels. 5 U.S.C. § 5301 (a) (1976). In furtherance of this policy, the President appointed an agent to prepare an annual report on federal salaries. 5 U.S.C. § 5305 (a)(1) (1976). This annual report relied on statistics from the Bureau of Labor Statistics on private sector pay, views of the “Federal Employees Pay Council” about the comparability of private and public sector pay systems, and the views of employee organizations not represented in the Council. 5 U.S.C. § 5305 (a)(1) (1976). This report did not and could not mandate the award of COLAs. The President also received a report from “The Advisory Committee on Federal Pay.” 5 U.S.C. § 5305 (a)(2) (1976). This committee reviewed the report issued by the President’s agent under section 5305(a)(1) and considered further views and recommendations provided by “employee organizations, the President’s agent, other officials of the Government of the United States, and such experts as it may consult.” 5 U.S.C. § 5306 (a)-(b) (1976). Based on these reports, the President could provide COLAs to General Schedule federal employees. 5 U.S.C. § 5305 (a)(2). If the President decided to recommend an adjustment, he would transmit to Congress the overall adjustment percentage. 5 U.S.C. § 5305 (a)(3). Any judicial COLAs were pegged to the “overall percentage” in the President’s report to Congress under section 5305. 28 U.S.C. § 461 (1976). Despite the 1975 Act, Congress allowed several COLAs for General Schedule federal employees but denied the increases to judges and other senior officials. The Supreme Court discussed the details of the legislation that blocked these increases. See Will, 449 U.S. at 205-09 , 101 S.Ct. 471 . In 1978, a group of federal judges filed suit *1181 alleging this blocking legislation was an unconstitutional diminution in salary contrary to Article III. Once the case made its way to the Supreme Court, the Court considered “when, if ever, ... the Compensation Clause prohibits] the Congress from repealing salary increases that otherwise take effect automatically pursuant to a formula previously enacted.” Id. at 221 , 101 S.Ct. 471 . The Court concluded that Congress could block COLAs due to judges so long as the blocking legislation took effect in the fiscal year prior to the year in which the increase would have become payable. Id. at 228-29 , 101 S.Ct. 471 . According to the Court, “a salary increase ‘vests’ ... only when it takes effect as part of the compensation due and payable to Article III judges.” Id. at 229 , 101 S.Ct. 471 . The 1989 Act, informed by the failures of the 1975 Act’s procedure, adopted a different purpose, used a different structure, and created different expectations than the 1975 Act. The 1975 Act “involved a set of interlocking statutes which, in respect to future cost-of-living adjustments, were neither definite nor precise.” Williams, 535 U.S. at 917 , 122 S.Ct. 1221 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Instead of being tied to the percent change in a known, published metric of inflation such as the Employment Cost Index, the adjustments under the 1975 Act depended on the discretionary decisions of the President’s agent and the Advisory Committee on Federal Pay. Furthermore, the President was not obligated to award adjustments to General Schedule employees on a specific timeline or even pursuant to the suggestions from the agent and the committee. Rather, he only did so if it furthered the policies underpinning federal pay articulated in 5 U.S.C. § 5301 . Thus, the method for calculating COLAs under the 1975 Act was “imprecise as to amount and uncertain as to effect.” Id. By contrast, the 1989 Act promised a mechanical implementation of COLAs for judges under the following equation: [[Image here]] See Pub. L. No. 101-194, § 704 (a)(1)(B), 103 Stat. 1716 , 1769 (Nov. 30, 1989). The Act contained only two limits: a presidential prohibition (due to national emergency or extreme economic circumstances) and a ceiling (of no more than five percent). Id. In essence, the statutes reviewed in Will required judicial divination to predict a COLA and prevented the creation of firm expectations that judges would in fact receive any inflation-compensating adjustment. In that context, as the Supreme Court noted, no adjustment vested until formally enacted and received. However, the statutes reviewed in Williams and in this case provide COLAs according to a mechanical, automatic process that creates expectation and reliance when read in light of the Compensation Clause. Indeed a prospective judicial nominee in 1989 might well have decided to forego a lucrative legal career, based, in part, on the promise that the new adjustment scheme would preserve the real value of judicial compensation. Aside from their respective differences in methods for calculating COLAs, the 1989 Act’s overall scope and legislative history distinguishes it from the statutory scheme addressed in Will . In fact, the automaticity of the 1989 Act’s COLAs takes on heightened significance in light of *1182 the broader statutory scheme because the 1989 Act also banned judges from earning outside income and honoraria. See Brown v. Gardner, 513 U.S. 115, 118 , 115 S.Ct. 552 , 130 L.Ed.2d 462 (1994) (“The meaning of statutory language, plain or not, depends on its context.”). In sum, the salary protections in the 1989 Act are only part of a comprehensive codification of ethical rules, Pub. L. No. 101-194 §§ 301-03, financial reporting requirements, id. at § 202, work rules for senior judges, id. at § 705, and — perhaps most important— prohibitions on outside income and honoraria, id. at § 601. Of the 935 active and senior judges in 1987, four hundred reported earning outside income from teaching law, speaking fees, and other sources. 135 Cong. Rec. S29,693 (daily ed. Nov. 17, 1989). More than half reported extra earnings from $16,624 to $39,500. Id. The Report by The Bipartisan Task Force on Ethics, which became the basis for the Ethics Reform Act of 1989, noted that the repeated failure to provide recommended salary increases for judges and other executive employees meant increased reliance on “earning honoraria as a supplement to their official salaries.” 135 Cong. Rec. H30,744 (daily ed. Nov. 21, 1989) (Task Force Report). During consideration of the 1989 Act, Congress acknowledged that denying access to outside income would amount to a “pay cut.” 135 Cong. Rec. S29,662 (daily ed. Nov. 17, 1989) (statement of Sen. Dole that removing outside income is a “pay cut”); see also 135 Cong. Rec. H29,488 (daily ed. Nov. 16, 1989) (statement of Rep. Fazio), H29,492 (daily ed. Nov. 16, 1989) (statement of Rep. Ford). In that context, reliance on the 1989 Act’s compensation maintenance formula took on added significance. See 135 Cong. Rec. H29,503 (daily ed. Nov. 16, 1989) (statement of Rep. Wolpe) (“[The] pay adjustment provision [is] tied directly to the elimination of all honoraria or speaking fees.”). Indeed, the Task Force Report emphasized that the restrictions and limitations on outside earned income, honoraria, and employment made by the Act are conditional on the enactment of the increased pay provisions. 135 Cong. Rec. H30,745 (daily ed. Nov. 21, 1989) (Task Force Report). The dependable COLA system became “a final important part” of the package designed to remove salaries “from their current vulnerability for political demagoguery.” 135 Cong. Rec. H29,483 (Nov. 16, 1989) (statement of Rep. Fazio); H30,753 (Nov. 21, 1989) (Task Force Report). In sum, the 1989 Act reduced judges’ income by banning outside income but promised in exchange automatic maintenance of compensation — a classic legislative quid pro quo. 135 Cong. Rec. H29,484 (Nov. 16, 1989) (statement of Rep. Martin stating that the Ethics Reform Act of 1989 is a comprehensive and interrelated package); cf. 135 Cong. Rec. H29,499 (Nov. 16, 1989) (statement of Rep. Crane objecting to the interrelated nature of the package and advocating separate bills for ethics and pay). Thus, the 1989 statutory scheme was a precise legislative bargain which gave judges “an employment expectation” at a certain salary level. Cf. United States v. Hatter, 532 U.S. 557, 585 , 121 S.Ct. 1782 , 149 L.Ed.2d 820 (2001) (Scalia, J., concurring in part and dissenting in part) (arguing that the repeal of judges’ exception from Medicare tax constituted a diminishment in compensation because judges had an expectation of an exemption from this tax). Moreover, the 1989 Act COLA provisions were not an increase in judicial pay. If so, the connection with the vesting rule for pay increases articulated in Will might be a closer issue. Rather, the statute ensured that real judicial salary would *1183 not be reduced in the face of the elimination of outside income and the operation of inflation. See Williams, 535 U.S. at 916 , 122 S.Ct. 1221 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). The vesting rules considered in Will are not expressly limited to the 1975 Act. However, the Supreme Court had no occasion to draw a distinction between a discretionary COLA scheme and a self-executing, non-discretionary adjustment for inflation coupled with a reduction in judicial compensation via elimination of outside income. For this reason, therefore, this court must examine further the actual differences in the two statutory schemes. The Supreme Court described the adjustments under the 1975 Act as “automatic.” See Will, 449 U.S. at 203, 223-24 , 101 S.Ct. 471 . An examination of the 1975 Act, however, shows that the adjustments at issue in Will were automatically operative only “once the Executive had determined the amount.” Id. at 203, 101 S.Ct. 471 (emphasis added). The ways that the Executive determined the amounts under the 1975 Act and the 1989 Act are very different. The former was an uncertain, discretionary process. The latter is precise and definite. While the Supreme Court described the COLAs in Will as “automatic,” the only aspect that was truly automatic was the link between judicial and General Schedule employee salaries. Whether General Schedule employees (and judges) would receive COLAs in any given year or whether those COLAs would maintain earning levels was anything but certain under the 1975 Act. Consequently, the only line the Supreme Court could draw in Will was between before and after the COLAs at issue were funded. The 1989 Act’s scheme presents a much different landscape than the Court confronted in Will . For these reasons, Will does not foreclose the relief that the judges seek. Although this court determines that Williams incorrectly applied Will and other aspects of the law, this determination does not end the inquiry. The court must now examine whether Congress’ decisions to deny the promised COLAs actually violated the Compensation Clause in Article III of the Constitution. The Compensation Clause has two basic purposes. First, it promotes judicial independence by protecting judges from diminishment in their salary by the other branches of Government. The founders of this nation understood the connections amongst protections for Life, Liberty, and the Pursuit of Happiness, protections for judicial independence, and protections for judicial compensation. Listed among the colonists’ grievances with the English Crown was that the King “ha[d] made Judges dependent on his Will alone for the Tenure of their Offices, and the amount and payment of their salaries.” Decl. of Independence para. 11 (U.S. 1776). As explained in The Federalist Papers, “[n]ext to permanency in office, nothing can contribute more to the independence of the judges than a fixed provision for their support.” The Federalist No. 79, p. 472 (Alexander Hamilton) (Clinton Rossiter ed., 1961). During the Constitutional Convention in 1787, the inspired draftsmen set out to protect against abuses such as those enumerated in the Declaration of Independence. James Madison of Virginia proposed prohibiting both enhancement and reduction of salary lest judges defer unduly to Congress when that body considered pay increases. Will, 449 U.S. at 219-20 , 101 S.Ct. 471 . Madison urged that variations in the value of money could be “guarded agst. by taking for a standard wheat or some other thing of permanent *1184 value.” Id. at 220 , 101 S.Ct. 471 (quoting 2 M. Farrand, The Records of the Federal Convention of 1787, p. 45 (1911)). The Convention rejected Madison’s proposal because any commodity chosen as a standard for judicial compensation could also lose value due to inflationary forces, i.e., the value of wheat could also fluctuate. Id. Thus, the Compensation Clause did not tie judicial salaries to any commodity. The framers instead acknowledged that “fluctuations in the value of money, and in the state of society, rendered a fixed rate of compensation [for judges] in the Constitution inadmissible.” The Federalist No. 79, supra. The Convention adopted the clause in its current form while voicing, at length, concerns to protect judicial compensation against economic fluctuation and reprisal. The Compensation Clause, as well as promoting judicial independence, “ensures a prospective judge that, in abandoning private practice — more often than not more lucrative than the bench — the compensation of the new post will not diminish.” Will, 449 U.S. at 221 , 101 S.Ct. 471 . This expectancy interest attracts able lawyers to the bench and enhances the quality of justice. Id. This expectancy interest does not encompass increases in future salary but contemplates maintenance of that real salary level. Williams, 535 U.S. at 916 , 122 S.Ct. 1221 (Breyer, J. joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari); The Federalist No. 79, supra, (noting that an Article III judge is assured “of the ground upon which he stands” and that he should “never be deterred from his duty by the apprehension of being placed in a less eligible situation”). The dual purpose of the Compensation Clause protects not only judicial compensation that has already taken effect but also reasonable expectations of maintenance of that compensation level. See Williams, 535 U.S. at 916 , 122 S.Ct. 1221 (Breyer, J. joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). The 1989 Act promised, in precise and definite terms, salary maintenance in exchange for prohibitions on a judge’s ability to earn outside income. The 1989 Act set a clear formula for calculation and implementation of those maintaining adjustments. Thus, all sitting federal judges are entitled to expect that their real salary will not diminish due to inflation or the action or inaction of the other branches of Government. The judicial officer should enjoy the freedom to render decisions — sometimes unpopular decisions — without fear that his or her livelihood will be subject to political forces or reprisal from other branches of government. Prospective judges should likewise enjoy the same expectation of independence and protection. A lawyer making a decision to leave private practice to accept a nomination to the federal bench should be entitled to rely on the promise in the Constitution and the 1989 Act that the real value of judicial pay will not be diminished. Will, 449 U.S. at 220-21 , 101 S.Ct. 471 ; cf. United States v. Winstar Corp., 518 U.S. 839, 872 , 116 S.Ct. 2432 , 135 L.Ed.2d 964 (1996) (recognizing that government promises may give rise to reasonable expectations). To be sure, the Compensation Clause does not require periodic increases in judicial salaries to offset inflation or any other economic forces. As noted before, the Constitutional Convention did not tie judicial salaries to a commodity or other standard of measurement. Will, 449 U.S. at 220 , 101 S.Ct. 471 . However, when Congress promised protection against diminishment in real pay in a definite manner and prohibited judges from earning outside income and honoraria to supplement *1185 their compensation, that Act triggered the expectation-related protections of the Compensation Clause for all sitting judges. A later Congress could not renege on that commitment without diminishing judicial compensation. That those compensation adjustments would happen in the future does not eliminate the reasonableness of the expectations created by the protections in the 1989 Act. Expectancy is, by its very nature, concerned with future events. Congress committed to providing sitting and prospective judges with annual COLAs in exchange for limiting their ability to seek outside income and to offset the effects of inflation. This decision furthered the Founders’ intention of protecting judges against future changes in the economy. Instead of fixing compensation relative to a commodity subject to inflationary pressure, Congress pegged the adjustment to a known measure of change to the economy as a whole, thus protecting the real salary of judges from both inflation and from fickle political will. By enacting blocking legislation in 1995, 1996, 1997, and 1999, Congress broke this commitment and effected a diminution in judicial compensation. Congress is not precluded from amending the 1989 Act. Congress may set up a scheme promising judges a certain pay scale or yearly cost of living increases. However, the Constitution limits those changes. If a future Congress wishes to undo those promises, it may, but only prospectively. Any restructuring of compensation maintenance promises cannot affect currently-sitting Article III judges. IV. Turning now to the second question, this court determines that the 2001 amendment to Section 140 of Pub. L. 97-92 has no effect on the compensation due to judges. Unlike the preceding discussion of the Compensation Clause, this is a question of statutory interpretation. Without a statutory basis for withholding the COLAs, federal judges should have received the adjustments in 2007 and 2010. These adjustments are payable to the judges regardless of constitutional protections. Congress simply had no statutory authority to deny them. As noted above, Section 140 was part of an appropriations bill passed in 1981. It barred judges from receiving additional compensation except as Congress specifically authorized in legislation postdating Section 140. See Pub. L. No. 97-92, § 140 , 95 Stat. 1183 , 1200 (Dec. 15, 1981). The appropriations act containing Section 140 expired by its terms on September 30, 1982. See Williams, 240 F.3d at 1026 . Thus, the rule that judicial pay adjustments had to be “specifically authorized by Act of Congress hereafter enacted” expired in 1982. Of course, in 2001, Congress amended Section 140, purporting to apply it “to fiscal year 1981 and each fiscal year thereafter.” Pub. L. No. 107-77, Title VI, § 625, 115 Stat. 748 , 803 (2001). Notably, Congress chose 1981 as the effective date for this extension of Section 140. As shown above, Congress did not explicitly authorize judicial compensation adjustments in 2007 and 2010. If Section 140 applied to bar those 2007 and 2010 adjustments, the absence of that additional Act of Congress would block — solely on the basis of this statute — any adjustments in those years. Section 140, however, by its own terms, did not block the 2007 and 2010 adjustments. Section 140 is straightforward: it bars judicial salary increases unless (1) “specifically authorized by Act of Congress” and (2) “hereafter enacted.” Pub. L. No. 97-92, § 140 . The 1989 Act’s pre *1186 cise and definite promise of COLAs clearly satisfies the first requirement to avoid a Section 140 bar. Williams, 240 F.3d at 1027 . The 1989 Act “specifically authorized” the 2007 and 2010 adjustments which occurred under its precise terms. Section 140 was enacted in 1981 and the 1989 Act occurred eight years later. Thus, the 1989 Act was “hereafter enacted” within Section 140’s meaning. When Congress amended Section 140 in 2001, it did not wipe the slate clean and set a new benchmark for the “hereafter enacted” requirement. The 2001 amendment makes no reference to its own November 28, 2001, enactment date. Instead, the amendment reiterates the 1981 baseline found elsewhere in the original Section 140, making the provision applicable to “ ‘fiscal year 1981 and each fiscal year thereafter.’ ” Pub. L. No. 107-77. An amendment referring only to fiscal year 1981 cannot redefine “hereafter” to refer to an entirely different date two decades later. Thus, the “hereafter enacted” requirement remained unchanged setting the “hereafter enacted” trigger date as 1981. In other words, Congress amended the existing Section 140 in 2001, but Section 140 remained a part of the Public Law 97-92 enacted in 1981. Furthermore, the amendment did not change Section 140’s enactment date. Indeed the Government agreed at oral argument before this court en banc that the 2001 amendment did not change the “hereafter enacted” clause of Section 140. The 2001 amendment merely erased Section 140’s expiration date, making permanent whatever effect the provision had when originally enacted. Congress thus expunged this court’s holding in Williams that Section 140 expired in 1982. The 2001 amendment, however, did not change Section 140’s substantive scope. The 1989 Act’s precise, automatic COLAs satisfy the requirements of Section 140 because it was enacted after Section 140. The Government withheld COLAs from judges in 2007 and 2010 solely because the government misinterpreted Section 140 as requiring a separate and additional authorizing enactment to put those adjustments into effect. By its own terms, Section 140 did not require that further authorizing legislation because it permitted COLAs under the “hereafter enacted” 1989 Act. V. In this case, Congress’ acts in 1995, 1996, 1997, and 1999 constitute unconstitutional diminishments of judicial compensation. Additionally, statutorily promised cost of living adjustments were withheld in 2007 and 2010 based on an erroneous statutory interpretation. Appellants’ motion to amend their complaint to include a challenge to the 2010 withholdings is granted. See Mills v. Maine, 118 F.3d 37, 53 (1st Cir.1997) (“[Ajppellate courts have authority to allow amendments to complaints because ‘[tjhere is in the nature of appellate jurisdiction, nothing which forbids the granting of amendments.’ ”) (quoting Newman-Green, Inc. v. Alfonzo-Larrain, 490 U.S. 826, 834 , 109 S.Ct. 2218 , 104 L.Ed.2d 893 (1989) (alterations omitted)). The statute of limitations does not bar these claims because, as established in Friedman v. United States, 159 Ct.Cl. 1 , 7, 310 F.2d 381 (1962) and Hatter v. United States, 203 F.3d 795, 799-800 (Fed.Cir. 2000), aff'd in part, rev’d in part on other grounds, 532 U.S. 557 , 121 S.Ct. 1782 , 149 L.Ed.2d 820 (2001), the claims are “continuing claims.” As relief, appellants are entitled to monetary damages for the diminished amounts they would have been paid if Congress had not withheld the salary adjustments mandated by the Act. On *1187 remand, the Court of Federal Claims shall calculate these damages as the additional compensation to which appellants were entitled since January 13, 2003 — the maximum period for which they can seek relief under the applicable statute of limitations. In making this calculation, the Court of Federal Claims shall incorporate the base salary increases which should have occurred in prior years had all the adjustments mandated by the 1989 Act had actually been made. See Hatter, 203 F.3d 795 (applying the “continuing claim” doctrine to calculating wrongful withholding of judicial pay). VI. This court has an “obligation of zealous preservation of the fundamentals of the nation. The question is not how much strain the system can tolerate; our obligation is to deter potential inroads at their inception, for history shows the vulnerability of democratic institutions.” Beer v. United States, 592 F.3d 1326, 1329 (Fed. Cir.2010) (Newman, J., dissenting from the denial of petition for hearing en banc). The judiciary, weakest of the three branches of government, must protect its independence and not place its will within the reach of political whim. The precise and definite promise of COLAs in the 1989 Act triggered the expectation-related protections of the Compensation Clause. As such, Congress could not block these adjustments once promised. The Court of Federal Claims’ dismissal of Appellants’ complaint is hereby reversed, and the case is remanded for further consideration in accordance with this opinion. OVERRULED-IN-PART, VACATED-IN-PART, AND REMANDED 


[Concurrence by O'Malley]

 O’MALLEY, Circuit Judge, with whom MAYER and LINN, Circuit Judges, join, concurring. I join the majority, both in the judgment it reaches and in its reasoning. I write separately to address two issues. First, I write to explain why I believe that, if United States v. Will, 449 U.S. 200 , 101 S.Ct. 471 , 66 L.Ed.2d 392 (1980), must be read as broadly as the dissent and the Williams v. United States, 240 F.3d 1019 (Fed.Cir.2001) majority believes it must, then Will was wrong and the Supreme Court should say so. Second, I write because I believe that, whatever its current statutory reach, Section 140 is unconstitutional and Congress can no longer rely on it to stagnate judicial compensation. I I first turn to Will . I agree with the majority that Will did not reach the issue presented here and, thus, does not dictate the result we may reach today. The position taken by the dissent, and by the Williams majority before it, is not without some force, however. One cannot deny that the adjudicatory principles upon which they rely are important ones, even if the majority concludes they are not determinative here. If the dissent is correct that we are forced to glean sweeping Compensation Clause principles from Will governing all forms of statutory enactments designed to increase judicial pay, we must also be forced to conclude that Witt’s analysis is flawed, both jurisprudentially and constitutionally. A. Jurisprudentially I find several aspects of the Will decision problematic. First, a close look at the facts and reasoning in Will reveals its internal inconsistency; neither its analysis nor its ultimate conclusion matches the facts presented. Specifically, while the Court in Witt initially characterized the statutory scheme at issue there as “automatic,” 449 U.S. at 223 , 101 S.Ct. 471 , it later justified its Compensation Clause holding by characterizing congressional action blocking salary increases under the scheme as merely modifying “the formula” by which “future” increases were to be calculated. Id. at 227-28 , 101 S.Ct. 471 . Next, if the language employed in Witt is meant to set down a “vesting” principle applicable in all Compensation Clause challenges, I believe the Court both: (1) violated the long-standing principle that courts are to decide only the cases before them and must only reach constitutional issues if and to the extent necessary; and (2) landed upon a holding that, taken to its logical extreme, creates absurd results. 1. Use of the Term “Automatic” As the majority notes, the statutory scheme at issue in Witt — the Executive Salary Cosl^of-Living Adjustment Act of 1975, Pub. L. 94-82, 89 Stat. 419 (Aug. 9, 1975) (“the Adjustment Act”) — was a complex scheme, fraught with discretion and uncertainty. Despite this, Will characterized the Adjustment Act as a pay adjust *1193 ment scheme which contemplated “automatic” pay increases. At issue in Will was the constitutionality of Congress’s decision to enact statutes preventing high-level Executive, Legislative, and Judicial officials, including Article III judges, from receiving COLAs in four consecutive years where General Schedule federal employees received increases. The Court noted that these blocking statutes were designed to “stop or to reduce previously authorized cost-of-living increases initially intended to be automatically operative” under the Adjustment Act. Will, 449 U.S. at 203 , 101 S.Ct. 471 (emphasis added). The Court then phrased the question presented in Will as: “when, if ever, does the Compensation Clause prohibit the Congress from repealing salary increases that otherwise take effect automatically pursuant to a formula previously enacted?” Id. at 221 , 101 S.Ct. 471 (emphasis added). As the majority notes, it is hard to understand the Court’s use of the term automatic in the context of the Adjustment Act. Normally, to say something is “automatic” is to say it occurs involuntarily or without further debate. See Oxford English Dictionary def. A(l); A(7)(a) (3d ed. June 2011; online version June 2012); see also American Heritage Dictionary 121 (5th ed. 2011) (def. 2a: defining “automatic” as “[ajcting or done without volition or conscious control; involuntary”). Nothing about the judicial salary adjustments at issue in Will was “automatic,” however. To the contrary, the adjustments at issue in Will were based on civil service salary adjustments that were entirely discretionary. As explained by the majority, whether federal employees would receive a COLA, and in what amount, depended on the initial recommendations of an adjustment agent which were then subject to review by an Advisory Committee, the President, and Congress. This procedure hardly can be described as one that occurs involuntarily. In addition, the statutes setting forth future COLAs were “neither definite nor precise,” and nothing provided that adjustments would be calculated “in a mechanical way.” Williams v. United States, 535 U.S. 911, 917 , 122 S.Ct. 1221 , 152 L.Ed.2d 153 (2002) (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Because the statutory scheme under the Adjustment Act “was imprecise as to amount and uncertain as to effect,” the Court’s characterization of the increases under the Adjustment Act as “automatic” is difficult to follow. See id. The dissent explains the Court’s mischaracterization of the Adjustment Act’s pay scheme by noting that, for the years in question in Will , the statutory scheme had run its course and resulted in a recommended salary increase by the time Congress acted to block those increases. This, the dissent seems to suggest, explains why the Supreme Court used the term “automatic” to describe what was before it. While that argument has a certain logic to it, it does not explain why the Court’s constitutional analysis focused on the absence of a guarantee under the Adjustment Act. According to the Supreme Court, the Adjustment Act did not “alter the compensation of judges; it modified only the formula for determining that compensation.” Will, 449 U.S. at 227 , 101 S.Ct. 471 (emphases in original). And, the Court said that the blocking statutes merely represented a decision to “abandon” that “formula.” It then admonished that, “[t]o say that the Congress could not alter a method of calculating salaries before it was executed would mean the Judicial Branch could command Congress to carry out an announced future intent as to a decision the Constitution vests exclusively in the Con *1194 gress.” Id. at 228 , 101 S.Ct. 471 (emphasis added). It was on this reasoning that the Court concluded that a salary increase does not “vest” for Compensation Clause purposes until it becomes part of a judge’s compensation that is due and payable and that Congress had not violated the Compensation Clause when it did not allow certain increases under the Adjustment Act to “vest.” Thus, the Court explained its Compensation Clause decision in Will by saying it was only dealing with a formula regarding an expressed “future intent” to provide increases; the Court did not say at that point that it was addressing increases that had already been decided upon. More importantly, it did not say it was addressing definite increases that had been promised by operation of law; in explaining its assessment of the Act vis-a-vis the Compensation Clause, the Court spoke of the scheme under the Adjustment Act as one that promised no more than potential adjustments. And, in discussing the concept of vesting, the Court seemed to back away from the notion that it was dealing with anything one could consider “automatic” in the common sense of that word. How can an increase occur “automatically” if a right to it had not yet “vested”? While I understand why the dissent believes we must assume the Supreme Court meant what it said when it described the Adjustment Act increases as “automatic” ones, that assumption would mean that the Court’s description of the facts presented had little correlation with its reasoning for why those facts did not run afoul of the Compensation Clause. 2. Constitutional Avoidance Next, if we read Will as broadly as Williams did, and the dissent now does, we must assume that, in Will , the Supreme Court violated its own well-established principle of constitutional avoidance. The Supreme Court has long-recognized that “[jjudging the constitutionality of an Act of Congress is ‘the gravest and most delicate duty that this Court is called upon to perform.’” Citizens United v. Fed. Election Comm’n, 558 U.S. 310 , 130 S.Ct. 876, 917-18 , 175 L.Edüd 753 (2010) (Roberts, C.J., concurring) (quoting Blodgett v. Holden, 275 U.S. 142, 147-48 , 48 S.Ct. 105 , 72 L.Ed. 206 (1927) (Holmes, J., concurring)). The Court’s standard practice, therefore, has been to “refrain from addressing constitutional questions except when necessary to rule on particular claims before [it].” Id. at 918 (citing Ashwander v. TVA 297 U.S. 288, 346-48 , 56 S.Ct. 466 , 80 L.Ed. 688 (1936) (Brandéis, J., concurring)). In furtherance of this practice, it has long been the rule that courts should “not ‘formulate a rule of constitutional law broader than is required by the precise facts to which it is to be applied.’ ” Ashwander, 297 U.S. at 347 , 56 S.Ct. 466 (quoting Liverpool, New York & Philadelphia S.S. Co. v. Commissioners of Emigration, 113 U.S. 33, 39 , 5 S.Ct. 352 , 28 L.Ed. 899 (1885)); see also United States v. Raines, 362 U.S. 17, 21 , 80 S.Ct. 519 , 4 L.Ed.2d 524 (1960) (same). Applying this principle in Citizens United , Chief Justice Roberts explained that the Court’s “standard practice of avoiding broad constitutional questions except when necessary” gives rise to an “order of operations,” whereby the Court considers the narrowest claim first before proceeding, if necessary, to any broader claims. 130 S.Ct. at 918 . Only if there is no valid narrow constitutional ground available, should the court resolve any broader constitutional question. See id. If we assume that Will is to be read so broadly as to control the result under the very different set of facts presented here, we must also assume the Court spoke to a *1195 question not before it. The constitutional question properly raised in Will was whether, under the specific statutory scheme set out in the Adjustment Act, the four blocking statutes at issue diminished judicial pay in violation of the Compensation Clause. A fair reading of Will based on “the precise facts to which it [was] applied,” requires limiting the holding to the statutory scheme that was before the Court. See Ashwander, 297 U.S. at 347 , 56 S.Ct. 466 (Brandéis, J., concurring) (citation omitted); see also Raines, 362 U.S. at 21 , 80 S.Ct. 519 . If Will is read to address a question broader than that presented — one that would govern a host of different congressional efforts to protect judicial pay from diminution in value— then we must conclude that, in Will , the Supreme Court ignored its own governing jurisprudential principles. In its briefing, the government concedes that there was a narrower approach the Court could have taken. Specifically, the government argues that, “even if the Supreme Court in Will could have based its decision upon the ‘discretionary’ character of the then-applicable statutory scheme, the Court did not decide the case upon that ground. The Court drew no such distinction.” Appellee’s Br. 26-27. If the government is right on this point, it is the very reason why Will was wrong to make the pronouncements upon which the government now relies. If the Court in Will consciously chose not to draw a distinction between a discretionary COLA scheme and a self-executing, non-discretionary one, it: (1) formulated a rule of constitutional law broader than required by the facts presented; and (2) ignored the fundamental precept that judges decide only the cases before them. See Hein v. Freedom from Religion Found., Inc., 551 U.S. 587, 615 , 127 S.Ct. 2553 , 168 L.Ed.2d 424 (2007) (“Relying on the provision of the Constitution that limits our role to resolving the ‘Cases’ and ‘Controversies’ before us, we decide only the case at hand.”) 3. Absurd Results Finally, the definition of “vesting” Williams gleaned from Will cannot be right. If it were: (1) Congress could do away with judicial retirement benefits for all sitting judges; (2) it would be inconsistent with the way the concept of vesting has been applied to similar pay increases for Members of Congress; and (3) it would run afoul of the common law understanding of the way in which future interests “vest” for all other purposes. It necessarily would lead to absurd results. First, if the definition of “vesting” Williams felt bound to under Will is correct, then Congress could eliminate judicial retirement pay for all sitting Article III judges without violating the Compensation Clause. By statute, Article III judges can retire with full pay once they reach a certain combination of age plus years of judicial service. See 28 U.S.C. § 371 . Under this system, the Supreme Court has said that the right to receive retirement pay “d[oes] not vest until retirement” and the “system provide[s] nothing for a judge who le[aves] office before age 65.” United States v. Hatter, 532 U.S. 557, 575 , 121 S.Ct. 1782 , 149 L.Ed.2d 820 (2001). In other words, the Supreme Court has specifically held that retirement benefits do not vest until a judge retires and certain prerequisites are met. In Will , the Court concluded that vesting occurs when a salary increase “takes effect as part of the compensation due and payable to Article III judges.” 449 U.S. at 229 , 101 S.Ct. 471 . As such, for those years where the COLAs at issue in Will had not yet become “due and payable,” the Court held that the blocking statutes did not violate the Compensation Clause’s pro *1196 hibition against diminishing judicial pay. See id. If we accept Will’s holding that Congress can abolish judicial salary adjustments at any time before they take effect, it logically follows that Congress would also be free to abolish judicial retirement pay at any time. The practical consequences of Will would place judicial retirement benefits at risk, despite the fact that the Supreme Court itself previously has characterized such benefits as “compensation” under Article III. See Hatter, 532 U.S. at 574 , 121 S.Ct. 1782 (“the noncontributory pension salary benefits [are] themselves part of the judge’s compensation”). Second, Will’s definition of vesting conflicts with the way in which that concept has been applied in the context of the Twenty-Seventh Amendment. In Boehner v. Anderson, 30 F.3d 156 (D.C.Cir.1994), the court addressed whether the 1989 Act (which also applies to Members of Congress) was inconsistent with the Twenty-Seventh Amendment which provides that: “No law, varying the compensation for services of the Senators and Representatives, shall take effect until an election of Representatives shall have intervened.” Id. at 159 . The court held that the phrase “shall take effect” in the Amendment referred to the date the Ethics Reform Act first became operative — i.e., 1991 — rather than any earlier or later point in time. See id. at 161-62 . Because the COLA provision of the Ethics Reform Act took effect in January 1991, after an intervening election in 1990, that provision did not violate the Twenty-Seventh Amendment. Id. at 162 . The court also held that: (1) Congress is free to specify a formula for future and continuing salary increases; and (2) the COLAs under the 1989 Act were designated to occur automatically each year after 1991, with no additional law necessary. Id. at 162-63 . All yearly COLAs beyond 1990 thus became operative and “vested” for Members of Congress when the law was first effective in 1991. 1 In Williams, the appellee-judges relied on the holding in Boehner to contend that the COLA increases for judicial officers took effect, or vested, when the law was effective, not when the yearly COLAs became due and payable. Williams, 240 F.3d at 1036 . This court recognized the holding in Boehner , but distinguished it on grounds that it dealt with a different question limited to Members of Congress. Specifically, the court found that Boehner “has no relevance ... to the question of whether the judicial pay aspects of the 1989 Act could, consistent with Article III, be revised or abrogated by later Acts of Congress.” Id. at 1037. That question, the Williams court held, was already answered in the affirmative in Will’s holding that “vesting, for federal judges under Article III, occurs only when compensation begins to accrue to the judges, not when a particular adjustment formula is enacted.” Id. at 1036-37. By simply relying on Will to distinguish Boehner , the court in Williams avoided the more difficult task of trying to reconcile two contradictory approaches to what vesting means under the Constitution. We are now faced with two distinct definitions of the constitutionally effective date of congressionally enacted COLAs. While Will provides that, for Article III purposes, a COLA is effective when it becomes “due and payable,” regardless of *1197 when the law establishing that COLA was enacted or when it took effect, Boehner states that, for Article I and the Twenty-Seventh Amendment, a COLA vests when the law is first effective, even if not due and payable for years to come. Common sense and basic principles of interpretation counsel against drawing this distinction. While it is certainly true that the operative date of congressionally designated salary increases is not prescribed in the Constitution, both the Compensation Clause and the Twenty-Seventh Amendment address the Framers’ concerns with in-term salary changes for the respective branches of government — one with decreases in-term and the other with increases in-term. I see no reason why the concept of vesting should be employed in a way to expand Congress’s ability to decrease judicial salaries under the Compensation Clause and be reframed under the Twenty-Seventh Amendment so as to expand Congress’s ability to increase its own. Finally, the vesting rule articulated in Will is an outlier. As this court in Williams correctly noted, “Mypically, ‘vesting’ of future interests only requires two components: an identification of the future owner, and certainty that the property would transfer.” 240 F.3d at 1032 (citing 2 Blackstone Commentaries 168; Simes & Smith, The Law of Future Interests, § 65, pp. 54-55 (2nd ed. 1956)). This view of vesting of future interests is “more consistent with black-letter [law].” See id. at 1038. The Supreme Court, nevertheless, “departed from traditional vesting rules” for future interests and announced a peculiar “actual possession” rule for Article III. Id. at 1032. Will ignored the standard rule for vesting of future interests and created a unique rule solely for judicial compensation. See id. at 1038. Despite recognition of its illogie, the Williams panel felt compelled to reject the use of traditional vesting rules for Compensation Clause purposes because it found those rules to be “simply contrary to the rule established by the Supreme Court in Will.” Id. at 1033. 2 If we are to believe that Will advanced such an extreme vesting rule — one applicable only to the Compensation Clause— then the Court should reexamine that rule and correct its mistake. Had the Supreme Court in Will applied the generally-accepted rule for vesting of future interests to the Adjustment Act, the same one the Boehner court applied to congressional pay increases, then a COLA whose formula was codified by law would vest, at an absolute minimum, once the amount of the COLA was established for a particular year. This approach is grounded in “sound equitable principiéis]” and, as we recognized in Williams, has deep common-law roots. See id. at 1032-33. For the reasons explained in further detail below, as the majority has noted, a more reasonable, consistent, and logical definition of “vesting” under Article III should be governed by the “reasonable expectations” of sitting judicial officers. *1198 Put simply, if we are to read Will as broadly as Williams did, and the dissent now does, the Court should revisit Witt’s unique vesting rule. B. Constitutionally If Witt truly established an “actual possession” vesting rule for Compensation Clause purposes, that holding seems indefensible under the Constitution. The Framers formulated the Compensation Clause for the express purpose of maintaining judicial independence, in part by providing judges with reasonable expectations about their pay and the inability of Congress to reduce it. As interpreted in Williams, the Will rule defeats the Framers’ intent and threatens the governmental structure around which the Constitution was formulated. 1. Historical Perspective and the Framers’ Intent The Compensation Clause “has its roots in the longstanding Anglo-American tradition of an independent Judiciary.” Witt, 449 U.S. at 217 , 101 S.Ct. 471 . As the Supreme Court has recognized, the “colonists had been subjected to judicial abuses at the hand of the Crown, and the Framers knew the main reasons why: because the King of Great Britain ‘made Judges dependent on his Will alone, for the tenure of their offices, and the amount and payment of their salaries.’ ” Stem v. Marshall, — U.S. -, 131 S.Ct. 2594, 2609 , 180 L.Ed.2d 475 (2011) (quoting the Declaration of Independence, para. 11). Against this backdrop, the Framers designed Article III to protect the public “from a repeat of those abuses.” Id. By giving judges life tenure and preventing the other branches from reducing judicial compensation, the Framers sought to “preserve the integrity of judicial decisionmaking.” Id. As the majority notes, in Federalist 79, Alexander Hamilton emphasized the importance of protecting judicial compensation. Specifically, he argued that, “[n]ext to permanency in office, nothing can contribute more to the independence of the judges than a fixed provision for their support.” The Federalist No. 79 at 385 (Alexander Hamilton) (Lawrence Goldman ed., 2008). Hamilton observed that, “[i]n the general course of human nature, a power over a man’s subsistence amounts to a power over his will.” Id. at 386 (emphasis in original). For this reason, the legislative branch must not “change the condition[s] of the [judiciary] for the worse” so that “[a] man may then be sure of the ground upon which he stands, and can never be deterred from his duty by the apprehension of being placed in a less eligible situation.” Id. Hamilton’s concerns, and those of many other Framers, were not merely academic. Indeed, throughout the former colonies, legislatures took retributive actions against judges with whom they disagreed, including attempts to remove judges who declared particular laws unconstitutional and to call judges before the legislature to answer for specific rulings. See Julius Goebel, Jr., Antecedents and Beginnings to 1801, in 1 History of the Supreme Court of the United States, 133-42 (Paul A. Freund ed., 1971). These events further supported the founders’ desire to insulate judges from the influence and control of the other branches of government. The Supreme Court has recognized that the primary purpose of the prohibition against reducing judicial salaries is “not to benefit the judges, but ... to promote that independence of action and judgment which is essential to the maintenance of the guaranties, limitations, and pervading principles of the Constitution.” Evans v. Gore, 253 U.S. 245, 253 , 40 S.Ct. 550 , 64 *1199 L.Ed. 887 (1920), overruled on other grounds by Hatter, 532 U.S. at 571 , 121 S.Ct. 1782 . The Compensation Clause should be “construed, not as a private grant, but as a limitation imposed in the public interest.” Id. It is the public that benefits from a strong, independent judiciary that is free to issue decisions without fear of repercussion. The Framers’ desire to insulate judicial pay from the political process was the subject of much debate and angst. While, given the long tenure judges would be asked to serve, there was no doubt some provision should be made for salary increases, the Framers also feared that, if salary decisions were left entirely to Congress, the judiciary might be forced to curry favor with Congress to secure reasonable compensation increases. See Jonathan L. Entin & Erik M. Jensen, Taxation, Compensation, and Judicial Independence, 56 Case W. Res. L. Rev. 965 , 972 (2006). To address this concern, James Madison suggested indexing judicial pay to the price of wheat or another stable value. The Framers rejected that idea, however, for fear fluctuations in commodity prices, like inflation, might leave judges undercompensated. See 2 The Records of the Federal Convention of 1787 44-45 (Max Farrand ed., 1911). Thus, while the Framers foresaw a need for in-term increases in judicial salaries and were concerned with leaving the task of providing those increases to Congress, they saw no alternative; no self-executing system they could devise seemed adequate to ensure that, given the dual effects of inflation and rising standards of living, judges would not be left undercompensated. So trust Congress they did, leaving to it the responsibility to guard against real decreases in judicial salary by future legislative enactments. In sum, the Framers intended to provide judges reasonable expectations about their pay. The Framers, to be sure, did not contemplate that a judges’ reasonable expectation would mean that he or she would become wealthy by taking the bench, or that Congress necessarily would increase judicial salaries. They believed, however, that Congress would assess fairly and periodically the need for increases in judicial compensation, would provide increases when appropriate, and that, once it did so, judicial officers thereafter could rely on the fact that Congress could not take such increases away. 2. The Expectations Approach in Practice Courts have long-endorsed this expectations-based approach to the Compensation Clause. Indeed, as Justice Breyer has noted, protecting “a judge’s reasonable expectations” is the “basic purposive focus” of the Compensation Clause. Williams, 535 U.S. at 916 , 122 S.Ct. 1221 (Breyer, J„ joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari). Likewise, Justice Scalia has argued that, when Congress takes away a previously-established component of the federal judicial “employment package,” it reduces compensation and thereby thwarts judicial expectations. See Hatter, 532 U.S. at 585 , 121 S.Ct. 1782 (Scalia, J., dissenting) (arguing that repeal of federal judges’ exemption from the Medicare tax was a reduction of compensation because those judges “had an employment expectation of a preferential exemption from taxation”). Consistent with this expeetations-related focus, the Supreme Court has held that the Compensation Clause forbids laws “which by their necessary operation and effect withhold or take from the judge a part of that which has been promised by law for his services.” O’Donoghue v. United States, 289 U.S. 516 , 533, 53 S.Ct. 740 , 77 L.Ed. 1356 (1933) *1200 (quoting Evans v. Gore, 253 U.S. 245, 254 , 40 S.Ct. 550 , 64 L.Ed. 887 (1920)). Other courts likewise have emphasized judicial expectations in their approach to the Compensation Clause. For example, in the early nineteenth century, the Circuit Court for the District of Columbia held that, “if [a judge’s] compensation has once been fixed by law, a subsequent law for diminishing that compensation ... cannot affect [a sitting judge].” United States v. More, 7 U.S. (3 Cranch) 159 , 160 n. 2, 2 L.Ed. 397 (1805), writ of error dism’d for want of jurisdiction. In More, Congress had enacted and later abolished a system of fees for compensating justices of the peace in the District of Columbia. Id. One of the justices of the peace continued to charge fees under the abolished structure, and the government brought an indictment against him. Id. On appeal, the Circuit Court held that: (1) the compensation of justices of the peace was subject to the Compensation Clause; and (2) where a fee structure is set by law, a later-enacted statute diminishing or abolishing that structure violated the Constitution. Id. at 161. Because sitting justices had an expectation that they would receive compensation consistent with the then-existing fee structure, Congress could not take that structure away. In Will , the Supreme Court discarded the longstanding expectations-based approach to the Compensation Clause in favor of its “due and payable” vesting rule, without clear explanation for doing so. In a terse footnote, the Court distinguished More. See Will, 449 U.S. at 228, n. 32, 101 S.Ct. 471 . Specifically, the Court claimed that, in More, “the fee system was already in place as part of the justices’ compensation when Congress repealed it” whereas “the increase [via the Adjustment Act] in Year 2 had not yet become part of the compensation of Article III judges” when it was repealed. Id. Careful consideration of the facts in More reveal that this is a distinction without a difference. The justices under the fee system in More were not entitled to compensation until they actually rendered services. See More, 7 U.S. at 160 n. 2 (“This compensation is given in the form of fees, payable when the services are rendered.”). At all times, the justices knew the precise amount they could charge for a particular service, but they never knew how much their total compensation would be, for example, in a particular week. In other words, the fee system in More merely set out a structure for calculating the compensation, which was not “due and payable” — to use the Court’s terminology in Will — until the justices performed the affirmative act of rendering services. The Adjustment Act formula was no different. In the same way that the justices under the fee system in More did not know how much they would work in a particular year, under the Adjustment Act, Article III judges did not know how much their salary would increase in a particular year, if at all. But they did know that, once the formula was enacted for the year, it became part of the compensation due. For example, looking at Year 3 in Will , if we accept the dissent’s proposition that the COLA of 5.5% became automatic once the President’s alternative plan was adopted and transmitted to Congress — which was one month before the Year 3 blocking statute was enacted — then there is no doubt that, as was the case in More, the COLA “was already in place as part of the [judges’] compensation when Congress repealed it.” See Will, 449 U.S. at 228, n. 32 , 101 S.Ct. 471 (citing More, 3 Cranch at 161 ). In the same way that Congress was prohibited from abolishing the fee structure in More because it was part of the justices’ compensation, so too should Con *1201 gress have been prohibited from blocking the COLA for Year 3 in Will . Given these similarities, Will’s dismissal of More is unconvincing. The two opinions are irreconcilable. Either Will is incorrect, or the Court should have said that More was wrong. The Supreme Court should return to the well-established expectations-based approach to the Compensation Clause. 3. The Consequences of Abandoning the Expectations Approach Assuming Will’s vesting rule allows Congress to bar “automatic” COLAs promised by definitive and precise legislative enactment, that rule is contrary to the constitutional balance the Framers carefully calibrated — one which, of necessity, delegated control over judicial salaries to the legislature, but did so in a way to guard against congressional retribution for unpopular judicial decisions. So understood, Will’s vesting rule puts at risk the principles the Framers struggled so hard to foster; it threatens to make the judiciary beholden to Congress in ways which undermine its independence. The Supreme Court should rethink such a rule. See e.g., Mistretta v. United States, 488 U.S. 361, 383 , 109 S.Ct. 647 , 102 L.Ed.2d 714 (1989) (encouraging vigilance against a “provision of law” that “impermissibly threatens the institutional integrity of the Judicial Branch”) (quoting Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 851 , 106 S.Ct. 3245 , 92 L.Ed.2d 675 (1986)). The Framers’ concerns were prescient. Statistics demonstrate that the erosion of judicial pay “has reached the level of a constitutional crisis that threatens to undermine the strength and independence of the federal judiciary.” Chief Justice John G. Roberts, Jr., 2006 Year-End Report on the Federal Judiciary, 39 The Third Branch 1, 1 (2007). Not only is this not the world the Framers contemplated, it is approaching one they most feared. As Hamilton explained, if judicial independence is “destroyed, the constitution is gone, it is a dead letter; it is vapor which the breath of faction in a moment may dissipate.” Commercial Advertiser (Feb. 26, 1802) (reprinted in The Papers of Alexander Hamilton, Volume XXV 525 (Columbia University Press 1977)). Ill I finally turn to Section 140 of Pub. L. No. 97-92, 95 Stat. 1183, 1200 (1981), and its role in our assessment of the legality of the congressional action challenged here. I agree with the majority that the existence of Section 140 does not change the conclusion that the failure to provide COLAs mandated by the 1989 Act is unconstitutional, whether the withholding occurred before or after Congress amended that section in 2001. As the majority explains, by its own terms, Section 140 is not applicable to the salary adjustments contemplated by the 1989 Act. If it were, however, as the government contends it is, we could hot enforce it because Section 140 is unconstitutional. Section 140 provides as follows: Notwithstanding any other provision of law or of this joint resolution, none of the funds appropriated by this joint resolution or by any other Act shall be obligated or expended to increase, after the date of enactment of this resolution, any salary of any Federal judge or Justice of the Supreme Court, except as may be specifically authorized by Act of Congress hereafter enacted.... Pub. L. No. 97-92, § 140 , 95 Stat. 1183 , 1200 (1981). Section 140 was a rider to a Joint Resolution providing continuing appropriations for fiscal year 1982. In Williams, we held that the government *1202 could not rely on Section 140 as justification for the blocking statutes passed in 1995, 1996, 1997, and 1999 because Section 140 expired by its own terms on September 30, 1982. Williams, 240 F.3d at 1026 (citing Pub. L. No. 97-161, 96 Stat. 22 (1982) (extending life of provisions from March 31, 1982 to September 30, 1982); Pub. L. No. 97-92, § 102 (c), 95 Stat. 1183 (1981)). After Williams, Congress enacted legislation that amended Section 140 to provide that it “shall apply to fiscal year 1981 and each fiscal year thereafter.” Act of Nov. 28, 2001, Pub. L. No. 107-77, § 625 , 115 Stat. 803 (“2001 amendment”). Today, the majority assumes that the 2001 amendment supersedes Williams’s holding that Section 140 expired, but agrees with the alternative holding in Williams that, even if not expired, the 1989 Act provides the additional authorization required by Section 140. Were the majority’s conclusion on that point not correct, then we would be forced to conclude that Section 140 violates the Compensation Clause, both because it singles out Article III judges for disadvantageous treatment and because it violates the principle of separation of powers. A. Section 140’s Discriminatory Effect The Supreme Court has held that a law violates the Compensation Clause when it “effectively single[s] out ... federal judges for unfavorable treatment” in their compensation. Hatter, 532 U.S. at 559 , 121 S.Ct. 1782 . In Hatter , the Court struck down a statutory scheme that required sitting federal judges to pay into the Social Security system while other high-level government officials potentially were exempt from making such payments. Id. at 564, 572-73 , 121 S.Ct. 1782 . In finding the denial of the exemption to judges unconstitutional, the Court explained that the “practical upshot” of the statutory scheme was to disadvantage judges relative to “nearly every current federal employee.” Id. at 573 , 121 S.Ct. 1782 . 3 Section 140 is no different. It only overrides the automatic annual COLAs promised in the 1989 Act for judicial officers. All other federal employees — including high ranking Executive Branch appointees and Members of Congress — remain entitled to those “automatic” adjustments. Only judicial officers are beholden to Congress for an additional affirmative legislative enactment before they may receive the 1989 Act’s COLAs. Thus, post-2001, Section 140 turns the 1989 Act into a law that provides a financial benefit to all federal employees other than judges and puts the judiciary in the position of annually needing to “curry favor” with the legislature for compensation increases, just as the Framers feared. That clearly violates the Compensation Clause. See Hatter, 532 U.S. at 576 , 121 S.Ct. 1782 ; Williams, 535 U.S. at 911 , 122 S.Ct. 1221 (Breyer, J., joined by Scalia and Kennedy, JJ., dissenting from denial of certiorari) (“[Section 140] refers specifically to federal judges, and it imposes a special legislative burden upon their salaries alone. The singling out of judges must throw the constitutionality of the provision into doubt.”) (citing Hatter, 532 U.S. at 564 , 121 S.Ct. 1782 ). *1203 “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.’ ” Hatter, 532 U.S. at 577 , 121 S.Ct. 1782 (quoting James Wilson, Lectures on Law (1791), in 1 Works of James Wilson 364 (J. Andrews ed. 1896)). The fear of disadvantageous treatment of judges under Section 140, as amended, is not hypothetical. Until recently, annual adjustments for federal judges remained in step with those for Executive Branch appointees and Members of Congress. When those groups received automatic adjustments under the 1989 Act, Congress also enacted the necessary special legislation to authorize an adjustment for judges. In fiscal year 2007, however, both General Schedule employees and Executive Branch appointees received an automatic adjustment under the 1989 Act, but Congress did not enact special legislation to adjust judicial salaries. The same thing happened in fiscal year 2010. Thus, the link between judicial salary adjustments and those for Executive Branch appointees was severed such that all nonelected federal employees other than Article III judges received COLAs in those years. 4 This is the very sort of individualized treatment of the judiciary that the Supreme Court has characterized as a “disguised legislative effort to influence the judicial will.” See Hatter, 532 U.S. at 571 , 121 S.Ct. 1782 . Little could be more inconsistent with the Framers’ purpose and construct under the Compensation Clause. B. Section 140 and the Separation of Powers Section 140 separately poses a separation of powers problem because it conditions the award of COLAs to judges on the receipt of salary adjustments by Members of Congress. The government argues that, in enacting the 1989 Act, “Congress made clear its intent to maintain a system of salary parity among Federal judges, members of Congress, and high-level Executive branch officers.” Appellee’s Br. 17 (citing Report of the Bipartisan Task Force on Ethics on H.R. 3660, Government Ethics Reform Act of 1989, 135 Cong. Rec. 30,756 (Nov. 21, 1989)). As noted above, any “parity” objective vis-a-vis Executive Branch officers has been abandoned. And, it is precisely because Congress has continued to use Section 140 to force a parity between judicial salaries and its own that Section 140 violates the principle of separation of powers. The concern with the independence of the judiciary is one which flows directly from the tripartite form of government on which the Constitution is structured. In establishing the system of divided powers in the Constitution, the Framers believed it was essential that “the judiciary remain! ] truly distinct from both the legislature and the executive.” Stem, 131 S.Ct. at 2608 (quoting The Federalist No. 78, p. 466 (C. Rossiter ed. 1961) (A. Hamilton)). Accordingly, as the Supreme Court has noted, the Framers built into the Constitution “a self-executing safeguard against the encroachment or aggrandizement of one branch at the expense of the other.” Mistretta, 488 U.S. at 382 , 109 S.Ct. 647 (quoting Buckley v. Valeo, 424 U.S. 1, 122 , 96 S.Ct. 612 , 46 L.Ed.2d 659 (1976)). Although the three branches “are not hermetically sealed from one another,” Article III was designed to impose certain “basic limitations that the other branches may not transgress.” Stem, 131 S.Ct. at 2609 *1204 (citing Nixon v. Administrator of Gen. Servs., 433 U.S. 425, 443 , 97 S.Ct. 2777 , 53 L.Ed.2d 867 (1977)). As noted earlier, the compromise the Framers struck under the Compensation Clause was one which would entrust to Congress the power and obligation to ensure reasonable salary adjustments for the judiciary over time. This was a compromise born of necessity, however; this mechanism for judicial salary adjustments was not meant to tie those adjustments to legislative salary changes, or to make them dependent on prevailing political winds. The Framers certainly did not mean to use the Compensation Clause to blur the lines between the legislative and judicial branches. That is precisely what Section 140 does, however. Congress has used Section 140 to link judicial pay to its own, affirmatively authorizing judicial compensation increases thereunder only in years where Congress finds it politically palatable to allow increases in its own. By using Section 140 in this way, Congress has ignored its constitutional duty to assess independently the adequacy of judicial compensation. And, it has ignored the obligation entrusted to it by the Framers to jealously guard the independence of the judiciary. “[Wjhether the Judiciary is entitled to a compensation increase must be based upon an objective assessment of the Judiciary’s needs if it is to retain its functional and structural independence.” Maron v. Silver, 14 N.Y.3d 230 , 899 N.Y.S.2d 97 , 925 N.E.2d 899, 914 (2010) (finding link between legislative and judicial pay increases unconstitutional under New York state constitution). Because Section 140 skirts Congress’s obligations under the Compensation Clause and undermines the independence of the judiciary, it is unconstitutional. The Supreme Court repeatedly has made clear that it is the laws that “threaten[] the institutional integrity of the Judicial Branch” that violate the principle of separation of powers. Mistretta, 488 U.S. at 383 , 109 S.Ct. 647 (quoting Commodity Futures Trading Comm’n v. Schor, 478 U.S. 833, 851 , 106 S.Ct. 3245 , 92 L.Ed.2d 675 (1986)). Under these well-established guideposts, Section 140 must fail. IV I agree with the majority that the failure to provide COLAs promised by the 1989 Act to the judiciary violates the Compensation Clause. I also agree that Will does not dictate a contrary result. “General propositions do not decide concrete cases.” Lochner v. New York, 198 U.S. 45, 76 , 25 S.Ct. 539 , 49 L.Ed. 937 (1905) (Holmes, J., dissenting). The general concepts espoused in Will simply do not address the very concrete and different set of facts before us. If the Supreme Court concludes Will must be read as broadly as this Court felt forced to read it in Williams, however, Will must be overruled. To the extent Section 140 plays any role in the Court’s analysis of the issues presented here, moreover, the Supreme Court should address its constitutionality and put its use to rest. WALLACH, Circuit Judge, concurring. I concur in the results, and in the reasoning of the decision, including the necessity of making this important determination that Congress may not exceed constitutional bounds in its relationship with the judiciary. I write separately only to clarify that this decision does not mean that any particular federal judge other than plaintiffs will necessarily accept accrued back pay. . In the alternative, the appellant in Boehner argued that, if the court found the COLA provision vested and constitutional, then a later-enacted statute that cancelled a planned COLA absent an intervening election violated the Twenty-Seventh Amendment. 30 F.3d at 162 . Although the answer to that question would be of interest to us now, the court declined to address it. See id. at 162-63 . . Indeed, despite awareness of Will , various state courts interpreting analogous provisions of their own constitutions have held that the failure to provide statutorily promised COLAs unconstitutionally diminishes judicial compensation. See e.g., Jorgensen v. Blagojevich, 211 Ill.2d 286 , 285 Ill.Dec. 165 , 811 N.E.2d 652, 664 (2004) (noting that the standards for conferring and calculating COLAs, which "were formulated following the United States Supreme Court’s decision in Will , expressly provided that COLAs were to be given on July 1, 1991, and on July 1 of each year thereafter and that such COLAs were to be considered a ' component of salary fully vested at the time the Compensation Review Board's report became law”). Will’s "vesting” rule for Compensation Clause challenges — if that is really what it is — stands alone. . Justice Scalia did not join in this portion of the Court's opinion, concurring on grounds that the Compensation Clause was violated because the congressional action violated the judicial officers' reasonable expectations about their future income package. Hatter, 532 U.S. at 586 , 121 S.Ct. 1782 (Scalia, J., concurring in part and dissenting in part) ("I disagree with the Court’s grounding of this holding on the discriminatory manner in which the extension occurred.”). The "discrimination” theory, however, received the votes of a majority of the Justices and, therefore, is binding precedent. . Members of Congress did not receive salary adjustments in 2007 or 2010 because they affirmatively chose to opt out of their right to receive them under the 1989 Act. That choice was theirs, however, and not one otherwise mandated by preexisting legislation. 


[Dissent by Dyk]

 DYK, Circuit Judge, with whom BRYSON, Circuit Judge, joins, dissenting. The majority opinion brings to mind an exchange between Learned Hand and Justice Holmes. Judge Hand enjoined Justice Holmes to “[d]o justice” on the bench, but the Justice demurred: “That is not my job. My job is to play the game according to the rules.” Learned Hand, A Personal Confession, in The Spirit of Liberty 302, 306-07 (Irving Dilliard ed., 3d ed. 1960). If the Supreme Court must play by the rules, that duty must be doubly binding on subordinate federal courts. Fidelity to this principle mandates adherence to the Supreme Court’s opinion in United States v. Will, 449 U.S. 200 , 101 S.Ct. 471 , 66 L.Ed.2d 392 (1980). I While the majority’s approach has much to recommend it as a matter of justice to the nation’s underpaid Article III judges, it has nothing to recommend it in terms of the rules governing adjudication. “The criterion of constitutionality is not whether we believe the law to be for the public good,” Adkins v. Children’s Hosp., 261 U.S. 525, 570 , 43 S.Ct. 394 , 67 L.Ed. 785 (1923) (Holmes, J., dissenting), but whether the law comports with the Supreme Court’s authoritative construction of the Constitution. Here, the issue is the scope of the Supreme Court’s 1980 decision in Will. Will’s holding is squarely on point. The Supreme Court’s framing of the issue was unmistakably clear: “when, if ever, does the Compensation Clause prohibit the Congress from repealing salary increases that otherwise take effect automatically pursuant to a formula previously enacted?” 449 U.S. at 221 , 101 S.Ct. 471 . The answer was that a future salary increase “becomes irreversible under the Compensation Clause” when it “vests,” id., and that it “ ‘vests’ for purposes of the Compensation Clause only when it takes effect as part of the compensation due and payable to Article III judges,” id. at 228-29 , *1188 101 S.Ct. 471 . The Court’s opinion in Will is unambiguous that the Court adopted what it has characterized as a “categorical” rule. See, e.g., Plant v. Spendthrift Farm, Inc., 514 U.S. 211, 239-40 , 115 S.Ct. 1447 , 131 L.Ed.2d 328 (1995). The Court in Will explained that for two of the years, the statute was passed before the Adjustment Act increases had taken effect — before they had become a part of the compensation due Article III judges. Thus, the departure from the Adjustment Act policy in no sense diminished the compensation Article III judges were receiving; it refused only to apply a previously enacted formula. A paramount — indeed, an indispensable — ingredient of the concept of powers delegated to coequal branches is that each branch must recognize and respect the limits on its own authority and the boundaries of the authority delegated to the other branches. To say that the Congress could not alter a method of calculating salaries before it was executed would mean the Judicial Branch could command Congress to carry out an announced future intent as to a decision the Constitution vests exclusively in the Congress. We therefore conclude that a salary increase “vests” for purposes of the Compensation Clause only when it takes effect as part of the compensation due and payable to Article III judges. 449 U.S. at 228-29 , 101 S.Ct. 471 (footnotes omitted). Under Will’s bright-line vesting rule, Congress was free to “abandon” a statutory formula and revoke a planned cost-of-living adjustment (“COLA”), as long as the revoking legislation was enacted into law before the COLA “took effect,” that is, became “due and payable” (i.e., before October 1, the first day of the next fiscal year). Id. at 227-29 , 101 S.Ct. 471 . In Will Years 1 and 4, Congress missed that deadline, and the Court held that the belated withdrawal of judges’ COLAs violated the Compensation Clause. Id. at 226, 230 , 101 S.Ct. 471 . But in Will Years 2 and 3, COLA-blocking statutes signed before October 1 were upheld, even though one of those statutes eliminated the promised COLA just a day before it would have taken effect. Id. at 229 , 101 S.Ct. 471 . Will thus made clear that a future salary increase only becomes protected by the Compensation Clause when it becomes “due and payable”; an increase which is merely anticipated or expected has not vested, and is not protected. By declining to follow Will’s clear vesting rule here, the majority also rejects the carefully crafted panel opinion in Williams v. United States, 240 F.3d 1019, 1039 (Fed.Cir.2001), reh’g denied, 240 F.3d 1366 (Fed.Cir.2001) (en banc), whose view of Will was supported at the time by a clear majority of the en banc court. See Williams, 240 F.3d at 1366 (eight judges concurring in the denial of rehearing en banc because “we are duty-bound to enforce [Will’s ] rule. If we have incorrectly read the Will opinion, the Supreme Court will have the opportunity to correct the error.”). II The majority attempts to redefine the constitutional test as turning not on “vesting,” but on “reasonable expectations,” a concept that appears nowhere in the Will opinion. To justify this shift, the majority seeks to distinguish Will on its facts, namely on the dubious ground that the “automatic” salary adjustment scheme in Will was different from the “automatic” salary adjustment scheme in place in Williams and here. But even if factual differences were pertinent (which, as we discuss below, could not support a depar *1189 ture from Will’s holding), there is no material difference between the statutes in Will and those in the Williams years (1995, 1996, 1997, and 1999). The Will statutes and the Williams statutes were not different insofar as they tied judicial compensation to General Schedule (“GS”) compensation, nor were they materially different as far as the definiteness of the GS COLA was concerned. Contrary to the majority’s suggestion, under both schemes, the COLA was “required” unless the President altered the COLA in response to “national emergency” or “economic conditions.” Compare 5 U.S.C. § 5305 (c)(1) (1976) with 5 U.S.C. at § 5303(b)(1) (2006). As the House Report to the 1990 Act stated, “[t]he President would have discretion [under the 1990 Comparability Act] to alter this adjustment.... This discretion is substantially similar to current law,” i.e., the 1975 Act. H.R.Rep. No. 101-906, at 88 (1990). 1 And under both statutory schemes, the GS COLA, once established, would “take effect automatically.” Will, 449 U.S. at 221 , 101 S.Ct. 471 . 2 Thus, the statutory schemes appear “strikingly similar” for all practical purposes. Williams, 240 F.3d at 1027. Nevertheless, the majority asserts that the expectation of a COLA created by the Williams statutes was significantly more “precise and definite,” Majority Op. 1183, because under Will’s more complex scheme, there was greater discretion over the COLA — an assertion which is accurate only insofar as the President’s agent and Advisory Committee had greater discretion in setting the initial amount of the GS COLA. Under each statutory scheme, the President’s discretion was the same. 3 But whatever the discretion, if the test were “reasonable expectations,” then the key question would not be how the statutory scheme initially determined a COLA, but whether the amount of the COLA had become “precise and definite” at the time the blocking statute thwarted the judges’ expectations. In this respect, Will cannot be distinguished from Williams . For Will Year 3, no “judicial divination,” Majority Op. 1181, would have been required: a GS COLA of 5.5% had already been specified in the President’s Alternative Plan, 14 Weekly Comp. Pres. Docs. 1480 (Aug. 31, 1978), which was adopted and transmitted to Congress by the President a month before the Year 3 blocking statute was enacted. Will, 449 U.S. at 229 , 101 S.Ct. 471 . The President had no further discretion to change the amount of the COLA. As the majority notes, “once the Executive had determined the amount,” the adjustments in Will were automatically opera *1190 tive. Majority Op. 1183 (quoting Will, 449 U.S. at 203 , 101 S.Ct. 471 ) (internal quotation marks omitted). In the Williams years, at the time the blocking statutes were enacted, the prospective amount of the GS COLA could be calculated based on the Employment Cost Index figures released by the Bureau of Labor Statistics, although the President generally did not announce a final amount until after the blocking statutes were enacted. 4 Thus, the COLA in Will Year 3 was just as “precise and definite” as the COLAs in the Williams years. Of course, the COLAs remained uncertain in another respect: in both Will and Williams , the presumptive GS COLA could still be overridden by Congressional action, and in fact it was overridden for one of the Williams years. 5 Again, there is no meaningful difference between the situations in Will and Williams. 6 To summarize: in both Will Year 3 and in each of the Williams years, at the time the judges’ COLA was blocked, the amount of the GS COLA had been established, the President retained no discretion to change the GS COLA, and the COLA would have taken effect automatically, absent Congressional intervention. The Supreme Court upheld the blocking statute in Will Year 3. 449 U.S. at 229 , 101 S.Ct. 471 . Yet the majority maintains that the blocking statutes in Williams offend the Constitution. This distinction is baffling. Finally, the majority here suggests that Will is distinguishable because the statutes here (unlike the statutes in Will) imposed limits on the judges’ outside income, without “an increase in judicial pay.” Majority Op. 1182. But the majority can hardly make a credible claim that judges’ outside compensation is protected by the Compensation Clause, and it follows that the reduction of outside compensation cannot create a Compensation Clause issue where none would otherwise exist. 7 *1191 Ill Even if the two statutory schemes were meaningfully different, and the Williams scheme created “reasonable judicial expectation[s] of future compensation” that did not exist in Will , Appellants’ Br. 29-31, that would be quite beside the point. Neither counsel for the appellants nor the majority is able to explain how that difference authorizes this court to disregard Will’s clear vesting rule. The majority concedes that “the vesting rules considered in Will are not expressly limited to the 1975 Act.” Majority Op. 1183. There is no basis for concluding that a “reasonable expectations” test has supplanted the Will vesting rule as the governing test. Certainly no decision of the Supreme Court has shifted the governing principle from vesting to reasonable expectations. There is not even a claim that subsequent decisions of the Court have somehow “underminefd] the reasoning” of Will. United States v. Hatter, 532 U.S. 557, 571 , 121 S.Ct. 1782 , 149 L.Ed.2d 820 (2001) (quoting Will, 449 U.S. at 227 n. 31, 101 S.Ct. 471 ) (internal quotation marks omitted). And even if Will had been undermined, it would not be this court’s prerogative to overrule it. See id. at 567 , 121 S.Ct. 1782 (noting that because Evans had been undermined but not yet “expressly overrule^],” the Federal Circuit “was correct in applying Evans” and thereby “invit[ing] us to reconsider” it). So too our job is to follow the holding of Will , not to confine it to its facts. Numerous Supreme Court decisions, and our own decisions, have made this clear. As the Supreme Court held in Thurston Motor Lines, Inc. v. Jordan K. Rand, Ltd., a Court of Appeals must not “confus[e] the factual contours of [Supreme Court precedent] for its unmistakable holding” in an effort to reach a “novel interpretation” of that precedent. 460 U.S. 533, 534-35 , 103 S.Ct. 1343 , 75 L.Ed.2d 260 (1983) (per curiam). See also, e.g., Marmet Health Care Ctr., Inc. v. Brown, — U.S. -, 132 S.Ct. 1201, 1202 , 182 L.Ed.2d 42 (2012) (per curiam) (a state court “misread[ ] and disregarded] the precedents of this Court” when it held the Federal Arbitration Act’s scope to be “more limited than mandated by this Court’s previous cases”); Ariad Pharms., Inc. v. Eli Lilly & Co., 598 F.3d 1336, 1347 (Fed.Cir.2010) (en banc) (“As a subordinate federal court, we may not so easily dismiss [the Supreme Court’s] statements as dicta but are bound to follow them.”). The fact that three Justices of the Court, dissenting from a denial of certiorari, opined that Will might be distinguished from Williams is not authoritative. See Williams, 535 U.S. at 917, 122 S.Ct. 1221 (Breyer, J., joined by Scalia & Kennedy, JJ., dissenting). A dissent from a denial of certiorari cannot “destroy[] the precedential effect” of a prior opinion. Teague v. Lane, 489 U.S. 288, 296 , 109 S.Ct. 1060 , 103 L.Ed.2d 334 (1989). This court has recognized that neither the agreement of three dissenting Justices, nor the approval of their reasoning by concurring Justices in later cases, can “transform a dissent into controlling law.” Prometheus Labs., Inc. v. Mayo Collaborative Servs., 628 F.3d 1347 , 1356 n. 2 (Fed.Cir.2010), rev’d on other grounds, Mayo Collaborative Servs. v. Prometheus Labs., Inc., — U.S. -, 132 S.Ct. 1289 , 182 L.Ed.2d 321 (2012). In short, neither the dissent from denial of certiorari in Williams nor the Supreme Court’s remand in this case can be read as an invitation for this court to perform reconstructive surgery on Will . The Supreme Court may distinguish its own opinions by limiting them to their facts, see, e.g., Williams v. Illinois, — U.S. -, 132 S.Ct. 2221 , 2242 n. 13, 183 L.Ed.2d 89 *1192 (2012), or choose to overrule them, see, e.g., Hatter, 532 U.S. at 567 , 121 S.Ct. 1782 , but that is not an option for this court. We respectfully dissent. 8 . Plainly Congress saw the references in the 1975 Act to "economic conditions” and in the 1990 Act to "serious economic conditions” as functionally the same, since the President’s discretion was to remain "substantially similar” under the 1990 Act as before. . Judge O’Malley's concurrence misreads the dissent in suggesting that we view the COLAs in Will as "automatic” only because "the statutory scheme had run its course” in the disputed years. Concur. Op. 1193. . Will's statutory scheme required the President to appoint an adjustment agent [who] was to compare sala- rles in the civil service with those in the private sector and then recommend an adjustment to an Advisory Committee. Subsequently, the Committee would make its own recommendation to the President, accepting, rejecting, or modifying the agent’s recommendation as the Committee thought desirable. The President would have to accept the Committee’s recommendation— unless he determined that national emergency or special economic conditions warranted its rejection. Williams v. United States, 535 U.S. 911 , 917, 122 S.Ct. 1221 , 152 L.Ed.2d 153 (2002) (Breyer, J., joined by Scalia & Kennedy, JJ., dissenting). . For all the Williams years, GS salary adjustment tables were promulgated by Executive Order in the preceding December. Exec. Order 12944, 60 Fed. Reg. 309 (Dec. 28, 1994); Exec. Order 12984, 61 Fed. Reg. 237 (Dec. 28, 1995); Exec. Order No. 13033, 61 Fed. Reg. 68987 (Dec. 27, 1996); Exec. Order No. 13106, 63 Fed. Reg. 68151 (Dec. 7, 1998). In each year, the judges' COLAs had been blocked several weeks to months earlier. See Pub. L. 103-329, Title VI, § 630(a)(2), 108 Stat. 2382 , 2424 (1994); Pub. L. 104-52, Title VI, § 633, 109 Stat. 468 , 507 (1995); Pub. L. 104-208, Title VI, § 637, 110 Stat. 3009 -364 (1996); Pub. L. 105-277, Title VI, § 621, 112 Stat. 2681 -518 (1998). For one of the Williams years, 1996, the President transmitted an Alternative Plan to Congress setting a 2% GS COLA before the blocking statute was passed. 31 Weekly Comp. Pres. Docs. 1466, 1466-67 (1995). . For 1995, Congress reduced the GS COLA to 2%. Pub. L. 103-329, Title VI, § 630(a)(1), 108 Stat. 2382 , 2424 (1994). The projected GS COLA had been 2.6%. See Sharon S. Gressle, Cong. Research Serv., Order No. RS20278, Judicial Salary-Setting Policy 6 (March 6, 2003). . Under the Will scheme, in addition to enacting separate legislation, Congress could have disapproved the Alternative Plan by a one-house legislative veto. Will, 449 U.S. at 204 , 101 S.Ct. 471 . But a legislative veto would not have zeroed out the GS COLA; it would have reinstated the amount recommended to the President, id., which was higher than the President's figure in Will Year 3. See 14 Weekly Comp. Pres. Docs. 1480 (Aug. 31, 1978). It is unclear how Congressional action to increase the GS COLA could have made the judges’ expectations of a COLA in Will Year 3 less “precise and definite.” The legislative veto was held unconstitutional after Will and before the Williams years. INS v. Chadha, 462 U.S. 919 , 103 S.Ct. 2764 , 77 L.Ed.2d 317 (1983). . In fact, the 1989 Act did increase judicial pay by 25%, thus offsetting the limitations on outside income. Pub. L. 101-194 § 703 (a)(3), 103 Stat. 1716 , 1768 (1989). . Appellants also argue that the 2007 and 2010 COLAs were improperly withheld because no blocking legislation was enacted in those years, and Section 140, as amended in 2001, was either inapplicable or unconstitutionally discriminated against federal judges under the Supreme Court's decision in Hatter . While we agree that this issue is not resolved by Will , these statutory and constitutional arguments were not properly raised below, and we decline to address them here.