dissenting:
I cannot subscribe to the majority’s evisceration of the Supreme Court’s decision in Farrar v. Hobby, — U.S. -, 113 S.Ct. 566, 121 L.Ed.2d 494 (1992). The language of Farrar is plain: “[T]o qualify as a prevailing party ... [t]he plaintiff must obtain an enforceable judgment .,. or comparable relief through a consent decree or settlement.” Id. at -, 113 S.Ct. at 573. “Whatever relief the plaintiff secures must directly benefit him at the time of the judgment or settlement. ” Id. (emphasis added). “Only under these circumstances can civil rights litigation effect ‘the material alteration of the legal relationship of the parties’ and thereby transform the plaintiff into a prevailing party.” Id. (emphasis added). “No material alteration of the legal relationship between the parties occurs until the plaintiff becomes entitled to enforce a judgment, consent decree, or settlement against the defendant.” Id. at -, 113 S.Ct. at 574 (emphasis added).1
Each of the above statements is now rendered meaningless by the majority’s direction to district courts to “look outside the final judgment” and beyond “the merits of the basic controversy” in awarding attorneys’ fees under § 1988. Majority Op. at 164. With respect, the resemblance between the Supreme Court’s view of fee recovery and that of the majority quite eludes me. Farrar states that without a demonstrated legal entitlement, plaintiffs cannot be deemed to have prevailed. The majority states that, notwithstanding Farrar, vague “catalyst” theories of causation continue to control who is a prevailing party under 42 U.S.C. § 1988. There is no way, however, that Farrar and a broad “catalyst theory” of attorneys’ fees recovery *169can be reconciled.2 I shall first address the particulars of this case, and then turn to a broader critique of the catalyst theory upon which the majority rests its case.
I.
Plaintiffs in this case fall short of meeting the rule laid down in Farrar. They obtained a judgment on the merits of their claim, but that judgment is no longer enforceable. The judgment’s vacatur on appeal renders it a legal nullity. See S-1 v. Spangler, 832 F.2d 294 (4th Cir.1987). Farrar is explicit on this point: “[t]he plaintiff must obtain an enforceable judgment against the defendant from whom fees are sought ” — U.S. at -, 113 S.Ct. at 573 (emphasis added). Plaintiffs certainly have prevailed against the City Board via their settlement, but the State Board has not settled and there is no judgment remaining against it.
In crediting plaintiffs’ vacated judgment against the State Board, the majority effectively deprives the state defendants in this case of their right to an appeal. And by adopting a porous standard for prevailing party status — that a plaintiffs position be “arguably supported by case or statutory law,” Majority Op. at 165 — the majority subjects these defendants to liability for attorneys’ fees without offering them the opportunity to refute the claims asserted against them. There is a virtue in the recognition that the vindication of legal rights is the proper end of all litigation. Without an enforceable judgment or settlement, however, it becomes difficult to discern what legal rights of plaintiffs have been violated or vindicated. This court properly recognized in its earlier decision that defendants had raised a constitutional issue of real “difficulty and sensitivity” involving “the power of the federal courts, acting within the constraints of the eleventh amendment, to interfere by injunction with the internal process of a state administrative agency.” Spangler, 832 F.2d at 298. The defendants in this case continue to insist that their view of the Individuals with Disabilities Education Act is the legally correct one. The majority never says otherwise — in its view all that is required to award attorneys’ fees against these defendants is for plaintiffs to state a “colorable claim.” Majority Op. at 166.
The majority attempts to explain away plaintiffs’ failure to secure an enforceable judgment by writing off as “dicta” the rule laid down in Farrar, Majority Op. at 167. The majority instead clings to a broad catalyst standard for recovery of attorneys’ fees under 42 U.S.C. § 1988. The causal story recounted by the majority in this case demonstrates the tenuous connection between a lawsuit and changes in conduct that will suffice to subject defendants to liability for attorneys’ fees under catalyst theory. Plaintiffs are found here to have been a “catalyst” for changes made by the State because (1) OSERS agreed with plaintiffs’ legal position in a Letter Ruling in an unrelated case, and (2) the State amended its laws three years after plaintiffs’ case was mooted, in response to the threatened loss of five million dollars in federal funding.
Even under the expansive standards of catalyst theory, this evidence of causation is astonishingly thin. Plaintiffs’ lawsuit lost *170whatever force it had as a catalyst the day it was mooted by this court. I do not understand how moot lawsuits, in which the possibility of relief against a party is by definition foreclosed, continue nonetheless to function as catalysts for change and for the recovery of attorneys’ fees. In this sense, the majority has stood finality on its head — the “prevailing party” standard in 42 U.S.C. § 1988 requires that we consult the confines of a case, not plumb the legal hereafter.3
II.
Quite apart from the difficulties of applying catalyst theory to this particular case, I cannot agree with the majority’s vocal endorsement of catalyst theory as a means of determining prevailing party status under § 1988. The problems with that theory are prodigious.
A.
First, the catalyst theory of fee recovery conflicts with the plain language of § 1988 itself. Section 1988 allows awards of reasonable attorneys’ fees to a “prevailing party, ” not simply to anyone who influences a change in behavior or policy. 42 U.S.C. § 1988 (emphasis added). For individuals to recover fees, they must prevail in their status as parties, not in their role as agents of reform. And to prevail for § 1988 purposes, parties must succeed on the merits of a claim. Farrar, — U.S. at -, 118 S.Ct. at 573. In other words, success must be something buttressed by a court’s authority or required by a rule of law. The lawsuit must materially alter the “legal relationship” between plaintiffs and defendants. See Texas State Teachers Ass’n v. Garland Indep. Sch. Dist., 489 U.S. 782, 792-93, 109 S.Ct. 1486, 1493, 103 L.Ed.2d 866 (1989). Farrar makes plain that “[n]o material alteration of the legal relationship between the parties occurs until the plaintiff becomes entitled to enforce a judgment, consent decree, or settlement against the defendant.” — U.S. at -, 113 S.Ct. at 574.
B.
Second, catalyst theory ignores the Supreme Court’s recent decisions. The majority says it is inconceivable that Farrar ever intended to disturb something so “well-established” as the catalyst theory for the recovery of attorneys’ fees. Majority Op. at 167.4 Far from overturning a “well-established” theory, however, Farrar’s specification of judgments, consent decrees, and settlement agreements as the exclusive avenues to prevailing party status under § 1988 follows from prior Supreme Court caselaw. Before Farrar, the Court "had noted a supposed fourth avenue. In Hewitt v. Helms, 482 U.S.
*171755, 760-61, 107 S.Ct. 2672, 2675-76, 96 L.Ed.2d 654 (1987), the Court observed that a § 1988 fee award could be justified if a lawsuit produced a voluntary “change in conduct” by a defendant that redressed a plaintiffs grievance. With this language, however, the Court did not endorse catalyst theory; it specifically reserved judgment on that question. 482 U.S. at 763, 107 S.Ct. at 2677. Nor did the Court’s language extend the ways to attain prevailing party status beyond the three situations later listed in Farrar. Farrar’s inclusion of settlements between parties encompasses both monetary settlements and agreements to a change in conduct. See Farrar, — U.S. at -, 113 S.Ct. at 573; see also Hewitt, 482 U.S. at 760, 107 S.Ct. at 2675. A plaintiff securing such a settlement can certainly be deemed to have prevailed in the absence of a formal judgment. See Hewitt, 482 U.S. at 761, 107 S.Ct. at 2676. Farrar merely clarified Hewitt by stressing that a voluntary change in conduct must be formalized in a legally enforceable settlement agreement to transform a plaintiff into a prevailing party for purposes of § 1988.
Other cases have similarly clarified Hewitt. For example, Hewitt stated that the change in a defendant’s conduct must redress a grievance. 482 U.S. at 761, 107 S.Ct. at 2676. Cases since Hewitt have narrowed this idea of grievance to make it a legal grievance. See Garland, 489 U.S. at 792-93 (focusing on the change in “legal relationship”); Rhodes v. Stewart, 488 U.S. 1, 3-4, 109 S.Ct. 202, 203, 102 L.Ed.2d 1 (1988) (per curiam) (holding that plaintiffs did not prevail when they were no longer in custody and judgment provided them with no legal relief). In other words, the change in conduct must be enforceable in court, not merely gratuitous. Farrar’s specification of the three avenues by which a party can prevail for § 1988 purposes does no more than continue the process of refining Hewitt. In its embrace of catalyst theory, the majority not only ignores Farrar; it bucks a clear decisional trend.
C.
Third, the catalyst theory of fee recovery engenders confusion and unnecessary litigation. By providing a clear rule for achieving prevailing party status, Farrar promises to reduce litigation over attorneys’ fees. The majority’s concern over the “specter of satellite litigation,” is well-founded, Majority Op. at 164, but the catalyst theory adopted by the majority then proceeds to exacerbate that very problem. Too frequently, legal battles over attorneys’ fees merely add another round of protracted litigation to what already has been protracted litigation on the merits of a claim. Compare Hensley v. Eckerhart, 461 U.S. 424, 437, 103 S.Ct. 1933, 1941, 76 L.Ed.2d 40 (1983) (“A request for attorney’s fees should not result in a second major litigation.”). This collateral litigation over attorneys’ fees is often more heated, more arcane, and over far higher monetary stakes than the underlying lawsuit. The relationship of all of this activity to the larger public good is becoming increasingly difficult to discern. Farrar’s crucial insights are that the refuge from such litigation lies in a clearly established rule for fee recovery and that the catalyst-based approach to fee applications has left us utterly at sea.
The majority’s standards for fee recovery under catalyst theory fail to chart a clear course through this litigious fog. We are told that the catalyst inquiry is “a pragmatic one of both fact and law that will ordinarily range outside the merits of the basic controversy ... to whether as a quite practical matter the outcome, in whatever form it is realized, is one to which the plaintiff fee claimant’s efforts contributed in a significant way,” and that “[courts] should look outside the final judgment to voluntary actions taken by a defendant so long as those actions are causally connected to the litigation.” Majority Op. at 164 (citations omitted). How far outside the merits should courts look? What is a significant contribution? When is a lawsuit causally connected and when is it not? What is the meaning of “in whatever form” an outcome is realized? The majority does not provide any guidance — only criteria that invite further litigation. In contrast to the majority’s approach, Farrar’s clear rule ensures that parties deserving attorneys’ fees will obtain them without forcing district courts to untangle a web of supposed causal *172connections to determine whether a party prevailed.
D.
Finally, catalyst theory discourages public officials from taking initiatives to revise outmoded ordinances or to improve institutional conditions, because such theory expressly recognizes “voluntary actions taken by a defendant” as a proper basis for a fee award. Majority Op. at 164. With its reliance on a simple chronology of events to show causation, catalyst theory empowers courts to award fees for any change in behavior that occurs after the filing óf a lawsuit, whether or not the court could have ordered that change in conduct. In this way, catalyst theory serves to disable public officials, who may come to fear that worthwhile changes may be retroactively linked to a lawsuit and result in a hefty bill for attorneys’ fees. It is counterproductive to convert § 1988 into a penalty on state and local officials for making salutary changes. At the same time, catalyst theory provides incentives for filing marginal, even frivolous, lawsuits. Any change in conduct by the defendant, for whatever reason, may offer a promising payout to attorneys who file a complaint, whether or not that complaint has any ultimate legal merit. Section 1988 should not be a license to shake down government officials, nor was it ever intended to be simply “a relief Act for lawyers.” Farrar, — U.S. at -, 113 S.Ct. at 578 (O’Connor, J., concurring), quoting Riverside v. Rivera, 477 U.S. 561, 588, 106 S.Ct. 2686, 2701, 91 L.Ed.2d 466 (1986) (Rehnquist, J., dissenting).
Catalyst theory is so open to fee awards because it assumes that enlightened social change automatically results from lawyers pursuing litigation. Unconstrained catalyst theory betrays the disquieting presumption that litigation must somehow lead the march of social progress. With § 1988, however, Congress did not grant parties a license to collect fees for pursuing personal visions of the public good. Rather, § 1988 compensates parties for attorneys’ fees incurred in bringing and successfully concluding lawsuits. Congress recognizes litigation under 42 U.S.C. § 1983 as a desired vehicle to vindicate the violation of legal rights, not as a means of effecting social change of the generic sort. In our system of government, we have looked first to elected officials to decide what change may or may not be considered enlightened. Section 1988 does not alter the basic democratic premise. Congress has said that attorneys’ fees may be awarded to prevailing parties who have established legal claims. That requires a judgment, consent decree, or settlement.
III.
Under Farrar, plaintiffs with valid civil rights claims will continue to have their day in court, assisted by counsel. Citizens may still “assert their civil rights,” and those who would violate such basic rights shall not “proceed with impunity,” because parties retain the chance “to recover what it costs them to vindicate these rights in court. ” S.Rep. No. 1011, 94th Cong., 2d Sess. 2 (1976), reprinted in 1976 U.S.C.C.A.N. 5908, 5910 (emphasis added). Farrar assures, however, that when a court awards “fees to a prevailing plaintiff, it is awarding them against a violator of federal law,” Christiansburg Garment Co. v. EEOC, 434 U.S. 412, 418, 98 S.Ct. 694, 699, 54 L.Ed.2d 648 (1978), not against an entity that improves a facility or changes a policy on its own initiative. By contrast, a theory of fee-shifting based upon changes in conduct that judges might consider constructive transfers the evaluation of many public actions to the unaccountable preferences of federal courts.
Plaintiffs in this case have not secured an enforceable judgment, consent decree, or settlement agreement. I would not award them attorneys’ fees on the basis that their mooted lawsuit may have operated as a catalyst to subsequent, changes in defendants’ conduct. With all respect to my colleagues, we have in Farrar v. Hobby a Supreme Court decision that controls this ease. We should follow it.
. The majority discards as "dicta” and dismisses as "cryptic” this extended discussion on the part of the Supreme Court whose meaning could not be more clear. See Majority Op. at 166-67 n. 2. We are entitled to disregard it, the majority says, because the Supreme Court failed to utter the words "catalyst theory” in the course of its discussion or to elucidate its reasoning in a manner sufficiently "careful” to satisfy the majority. See id. The majority makes no attempt to reconcile the Supreme Court’s opinion with the continued existence of "catalyst theory.” With all respect, this is a nervy course — while a circuit court is necessarily entrusted with the interpretation and application of Supreme Court pronouncements, it is not empowered, sua sponte, to set them aside.
. The cases from other circuits cited by the majority are no more persuasive in resuscitating catalyst theory. With all due respect, those cases fail to come to grips with Farrar's ruling that "[n]o material alteration of the legal relationship between the parties occurs until the plaintiff becomes entitled to enforce a judgment, consent decree, or settlement against the defendant." - U.S. at -, 113 S.Ct. at 574. Paris v. HUD simply ignores Farrar's requirement of a legal victory, citing the case only in passing. 988 F.2d 236, 238 (1st Cir.1993). Citizens Against Tax Waste v. Westerville City School quotes Farrar ’s standard in a footnote, but nonetheless invokes Farrar in support of a pure catalyst theory to reverse a district court denial of fees. 985 F.2d 255, 258 and n. 3 (6th Cir.1993). In sum, the cases cited by the majority do not come to terms with the inconsistency between Farrar and catalyst theory. I would not follow their example.
Finally, in Craig v. Gregg County, the court acknowledged the conflict between Farrar and catalyst theory, but skirted the issue by finding as a matter of fact that plaintiff had not been a catalyst for the county's reconfiguration of a voting precinct. 988 F.2d 18, 21 (5th Cir.1993) ("A more precise reading of Farrar ... might suggest that a party may prevail [via catalyst theory]," (emphasis added) but plaintiff "has not shown that his suit caused [the defendant’s change in conduct].").
.The "voluntary compliance” cases cited by the majority, see Majority Op. at 167 n. 3, are easily distinguishable from the instant case. The cases relied on by the majority reflect only the principle that when a plaintiff is successful in obtaining a preliminary injunction based on its probability of success, the defendant's voluntary cessation of unlawful conduct need not deprive plaintiffs of prevailing party status. That is quite different from our case, where the change in conduct occurred in response to a third-party’s view after we declared the lawsuit moot, where we have never adjudged the state's conduct to be unlawful, and where we explicitly credited defendants with raising a complex question of federal law which we then declined to address. The inappropriateness of an attorney's fee award against defendants in this latter set of circumstances should be apparent.
The majority purports to see some inconsistency between my position here and my joining the earlier remand to the district court to determine whether fees should be awarded. See id. The veiy terms of the remand presupposed, however, that the basis for any award would return here for review. Further, the earlier decision itself emphasized that “[b]y remanding we of course express no view on the parents’ entitlement to costs and attorney’s fees against the state defendants.” 832 F.2d at 298 n. 3. I make no apologies for following a Supreme Court decision that has issued between the last appeal and the disposition of this one. Indeed, if the majority believed Fanar to be persuasive authority, it presumably would follow it too.
.In this context, the majority’s reference to Hensley v. Eckerhart, 461 U.S. 424, 433, 103 S.Ct. 1933, 1939, 76 L.Ed.2d 40 (1983), as being a "case remanded specifically to consider whether claimant's lawsuit was not a catalyst' for achieving a favorable result,” Majority Op. at 167, is anything but an endorsement of catalyst theory. That case involved an award to a "partially prevailing plaintiff” who had actually obtained a judgment on the merits. Hensley, 461 U.S. at 426-28, 103 S.Ct. at 1935-36. The issue of prevailing party status simply was not raised.