EASTERBROOK, Circuit Judge.
Now 16 years old, this case is making its third appearance in this court — and the parties’ current dispute arises out of the conflicting decisions of the first two panels. We meet in banc to consider whether a district court should require a unit of state or local government to abide by a consent decree that does not serve any federal interest. The answer is No, and the injunction based on the parties’ agreement therefore must be vacated.
I
During the late 1970s and eai’ly 1980s, the rate of interest Chicago paid on judgments (6% per annum for municipal governments, 735 ILCS 5/2-1303) was substantially less than the market cost of credit. Judgment debtors had every reason to postpone payment as long as this imbalance persisted. Instead of borrowing in the market at 15%, or raising taxes, Chicago borrowed from its judgment creditors. It paid tort judgments of $1,000 or less, and all contract judgments, quickly. The City Council did not appropriate funds sufficient to pay other judgments in the year the courts entered them. By 1979 plaintiffs without the political clout to jump the queue had to wait on average 47 months for payment. An active secondary market in judgments against Chicago developed. About 80% of judgment holders sold in this market, accepting a discount of approximately 25% off the face value of their awards. Evans v. Chicago, 689 F.2d 1286, 1290 (7th Cir.1982) (Evans I).
Suits challenging Chicago’s delay in payment were filed, certified as class actions, and consolidated. By summary judgment the district court held that Chicago’s practice of paying small judgments quickly while deferring payment of larger ones violated the due process and equal protection clauses of the fourteenth amendment: due process because it deprived judgment creditors of a “property” interest in immediate payment created by state law, and equal protection because there was no rational basis for distinguishing large from small awards. A panel of this court affirmed the portion of the judgment that rested on the equal protection clause, id. at 1299-1300, while vacating the due process aspect as premature. Id. at 1296-99. Dicta in the opinion strongly imply that the City deprived the plaintiffs of due process of law. Id. at 1297-98.
On remand the district judge told the parties and their lawyers that the combination of his opinion and ours condemned Chicago’s practices en toto. He directed the parties to negotiate appropriate relief. They complied, and the judge approved a consent decree eliminating the distinction between large and small judgments, requiring all judgments to be paid in order of their entry, and providing that the Mayor must ask the City Council to appropriate enough money to pay all judgments promptly. The litigants could not agree whether the plaintiffs were entitled to damages for delay in payment. The district court concluded that they were, under the equal protection clause — both the plaintiffs and the district court deeming the due process theory surplusage in light of the equal protection holding. Another panel of this court reversed, overruling Evans I. Evans v. City of Chicago, 873 F.2d 1007 (7th Cir.[476]*4761989) (Evans II). The second panel (with the acquiescence of the full court, see id. at 1008 n. *) concluded that the City had a rational basis, if only administrative convenience and the placation of the more numerous holders of small awards, to pay little judgments before big ones. Id. at 1015-18. The second panel wrapped up: “The district judge advanced some worthy ideas, but they are for the City’s self-determination, and will not be imposed by this court. It is regrettable that this matter which has lingered so long now takes a new and possibly unexpected turn, but what we now view as error must be arrested even at so late a date. Improvements in this situation are better left to the state and municipal governments.” Id. at 1017-18.
Chicago then asked the district court to vacate the injunction and return the subject to- “the City’s self-determination”. Interest rates have changed; today the legal rate approximates the market rate, so Chicago would pay promptly even if the injunction were not in force. Still, it wants to redeem its governmental powers, now in hock in a district court. The motion invoked Fed. R.Civ.P. 60(b)(5), which permits a court to modify a judgment when “a prior judgment upon which it is based has been reversed or otherwise vacated, or it is no longer equitable that the judgment should have prospective application”. Evans I, which led to the settlement, was gone; the City added that in its view it is also “no longer equitable” that a judgment affecting its legislative functions remain in force when the judgment does not serve a federal interest. Rule 60(b)(5) applies to consent decrees as well as to other judgments. Rufo v. Inmates of Suffolk County Jail, — U.S. -, -, 112 S.Ct. 748, 757, 116 L.Ed.2d 867 (1992). Recognizing that the decree’s foundation had been washed away, the district court set it aside. Rufo, which holds that consent decrees regulating the conduct of state or local govern-' ments may be modified more freely than those entered by private litigants, supports this disposition.
On motion for reconsideration, however, the district judge reversed course. He observed that Evans II addressed only the plaintiffs’ equal protection theory- — -for the plaintiffs had not urged the panel to affirm the award of damages on due process grounds. 873 F.2d at 1012 n. 11, .1018 n. 15. Because Evans II did not bear on the due process theory, there had not been any change of law affecting this aspect of the plaintiffs’ case. When Chicago agreed to pay judgments faster, the district court believed, it was settling the due process claim along with the equal protection claim. The district judge therefore revised the injunction to delete only so much of the relief as Evans II had expressly held inappropriate. The sole provision satisfying that description was the portion forbidding Chicago to distinguish judgments according to size. Deleting that element was nugatory because of the separate provision compelling the City to pay judgments promptly and strictly in order of their entry. Stuck exactly where it was before Evans II, Chicago asks us to annul the consent decree and the injunction implementing it.
II
No one doubts that Rufo and Rule 60(b)(5) entitle Chicago to relief to the extent the decree rests on the equal protection theory. The injunction intrudes on the internal operations of the City, telling the Mayor what items must be in the annual budget. Evans II rightly said that, although changes may be beneficent, the political rather than the judicial process is responsible for the subject.
Governments may undertake to do more than the Constitution requires. Rufo, - U.S. at - - -, 112 S.Ct. at 760-63; see also Firefighters v. Cleveland, 478 U.S. 501, 525, 106 S.Ct. 3063, 3076, 92 L.Ed.2d 405 (1986). If there were something to the due process theory (a big if, discussed in Part III below), a decision removing all but this one theory would leave room for settlement, as the district court said. The initial problem with its decision, however, is that Chicago did not settle any aspect of the merits. It litigated the merits to the nines and settled only the details of relief.
Rule 60(b)(5) calls on the court to ascertain whether a judgment is “based” on some earlier decision. This consent decree is “based” [477]*477on Evans I rather than an independent compromise of the due process claim. Plaintiffs wanted money. The City refused to pay, and after issuing an injunction the district judge held a trial on damages. Chicago appealed from an adverse decision, leading to Evans II. If the consent decree really represented a compromise of the merits — even of the due process theory alone — there could not have been an Evans II. Yet on appeal in Evans II the plaintiffs did not contend that the merits had been compromised; it was plain to all that the merits (and damages) were still in dispute. All that had been compromised was prospective relief. Chicago “agreed” to do what Evans I implied that it must do. “ ‘Consent’ that is no more than knuckling under to the inevitable is more like an adjudication than a contract.” People Who Care v. Rockford Board of Education, 961 F.2d 1335, 1338 (7th Cir.1992), quoted in United States v. Chicago, 978 F.2d 325, 333 (7th Cir.1992).
Before accepting the parties’ agreement, the district judge made clear his view that Evans I (and his own prior decision) resolved the merits, leaving for decision only issues about relief. If Evans I condemned the City’s approach to paying judgments, the parties had no need to settle the due process claim. After a decision on one claim resolves the ease, neither judge nor litigants needs to address additional legal theories. The district judge told the parties in no uncertain terms that Evans I resolved the merits, and that nothing remained but to select the appropriate relief. When Chicago bridled, the judge announced from the bench on October 26, 1983:
Now, I have got to have an order from the City and the plaintiffs. I have got to have an agreed order that will bind the City to pass a budget sufficient to pay tort judgments.
If by November 25th, approximately a month from now, I do not have in my hands an agreed order that calls upon the City and requires the City to budget annually an amount sufficient to pay tort judgments along the lines which we have been discussing in great detail at our recent and not so recent conferences in chambers, then I am going to go ahead and enter my own order.
* * * * * *
I see no need for any further conferences with counsel that go to basic philosophy or go to what I have been saying this afternoon. If there is some mechanical thing you want to talk about that I can help you with, I will be happy to sit down, but just to rehash this question about whether we are going to do something in this case that the Court of Appeals ordered done a long time ago, the time for discussion has passed.
The judge believed that Evans I dictated ulhat to do, but not hoiv. Mechanical details remained for decision. And the details of prospective relief are all the parties compromised. The merits were not settled. They were litigated, twice. The City lost in Evans I and won the rematch in Evans II.
Ill
Let us suppose, however, that this understanding is inaccurate. The question remains whether in the language of Rule 60(b)(5) “it is no longer equitable that the judgment should have prospective application”. The district court treated this case the same way it would if the defendant were a private merchant that had agreed to sell the plaintiff a ton of steel: a contract’s a contract, and pacta sunt servanda. Yet this is no ordinary contract. First, it requires continuing supervision by the district court. Judges need a good reason, one in addition to the parties’ say so, before diverting attention from other business in this fashion. System Federation v. Wright, 364 U.S. 642, 651, 81 S.Ct. 368, 373, 5 L.Ed.2d 349 (1961) (“The parties cannot, by giving each other consideration, purchase from a court of equity a continuing injunction.”); Firefighters v. Cleveland, 478 U.S. at 525, 106 S.Ct. at 3077 (A “federal court is more than ‘a recorder of contracts’ from whom parties can purchase injunctions.”). Second, the decree entangles an arm of federal government in the administration of another sovereign, monitoring the budgetary decisions of a Mayor and City [478]*478Council rather than enforcing strictly legal rights. Again such a step may be taken with justification, cf. Missouri v. Jenkins, 495 U.S. 33, 110 S.Ct. 1651, 109 L.Ed.2d 31 (1990), but the parties’ consent is not automatically sufficient.
Especially not when one of “the parties” did not consent. This is the third, and most potent, objection to the district court’s approach. “Chicago” did not reach a settlement with the plaintiffs. The consent decree was entered in 1984. Negotiations were conducted on Chicago’s behalf by its corporation counsel, who we may suppose acted with the approval of Harold Washington, then Chicago’s mayor. Although the decree purports to last for all time — and the district court’s decision refusing to vacate the decree even after Evans II reflects a belief that the commitments ought to run perpetually — democracy does not permit public officials to bind the polity forever. What one City Council enacts, another may repeal; what one mayor decrees during his four-year term, another may revoke. Today’s lawmakers have just as much power to set public policy as did their predecessors. “Chicago” speaks through its elected representatives, and the people are free to upset even the most enlightened policies of earlier times. The current mayor wants to be free of his predecessor’s commitment, concluding that more flexibility over budgets will promote the public welfare. People of good will could be on either side of this disagreement; each mayor may have correctly perceived the needs of the moment.
Governments are in this respect unlike corporations or other contracting parties. A corporate board of directors may enter into commitments that continue after new directors take office; a legislature may not. True, governments may form contracts (for example, to build a new road or repay a loan) and must keep these commitments by virtue of the contract clause of the Constitution, Art. I, § 10, cl. 1. See United States Trust Co. v. New Jersey, 431 U.S. 1, 97 S.Ct. 1505, 52 L.Ed.2d 92 (1977). But temporary officeholders may not contract away the basic powers of government to enact laws — or in this case to adopt budgets — in the same way natural persons may make enduring promises about their own future behavior. Wilbur v. United States, 281 U.S. 206, 217, 50 S.Ct. 320, 324, 74 L.Ed. 809 (1930); Stone v. Mississippi, 101 U.S. 814, 817-18, 25 L.Ed. 1079 (1879); Charles River Bridge Co. v. Warren Bridge Co., 36 U.S. (11 Pet.) 420, 9 L.Ed. 773 (1837). Why then should things differ if the parties choose not the device of a seal (or even of a statute) but the imprimatur of a district judge? See Michael W. McConnell, Why Hold Elections? Using Consent Decrees to Insulate Policies from Political Change, 1987 U.Chi. Legal Forum 295; cf. Peter M. Shane, Federal Policy Making by Consent Decree: An Analysis of Agency and Judicial Discretion, id. at 241, 269 n. 105.
Consent alone is insufficient to support a commitment by a public official that ties the hands of his successor. Accord, League of United Latin American Citizens v. Clements, 999 F.2d 831, 846 (5th Cir.1993) (in banc) (“Courts must be especially cautious when parties seek to achieve by consent decree what they cannot achieve by their own authority. Consent is not enough when litigants seek to grant themselves powers they do not hold outside of court.”); United States v. Miami, 664 F.2d 435, 440 (11th Cir.1981) (in banc). Thus the answer to the question “why should consent decrees be enforced when contracts out of court are not?” must concentrate on legal rules that shape the parties’ agreement. A promise by a public official is of limited utility when only the official’s word supports the commitment. A consent decree may be — under System Federation must be — based on more than consent. It depends on rules of law that govern the public official’s conduct. “ ‘[T]he District Court’s authority to adopt a consent decree comes only from the statute which the decree is intended to enforce,’ not from the parties’ consent to the decree.” Firefighters v. Stotts, 467 U.S. 561, 576 n. 9, 104 S.Ct. 2576, 2586 n. 9, 81 L.Ed.2d 483 (1984), quoting from System Federation, 364 U.S. at 651, 81 S.Ct. at 373. A state official’s promise to follow a rule of federal law l’etains its force because of the continuing effect of the law, which the state cannot alter. And a settlement of a dispute about the meaning of that law may be enforced if the agreement com[479]*479promises genuine uncertainties, for then the public official actually may be enhancing or preserving the powers of the democratic branch (by avoiding a worse outcome after trial) rather than ceding the powers of the government.
This method of justifying the implementation of consent decrees implies, however, that the court must ensure that there is a substantial federal claim, not only when the decree is entered but also when it is enforced, and that the obligations imposed by the decree rest on this rule of federal law rather than the bare consent of the officeholder. When making these inquiries, courts are bound by principles of federalism (and by the fundamental differences between judicial and political branches of government) to preserve the maximum leeway for democratic governance. Over the last decade a series of decisions in this circuit has emphasized the district court’s responsibility to identify the rule of federal law supporting a consent decree binding the political arms of government, and the corresponding obligation to permit new public officials to set their own policy within the limits established by federal law. We mention only a sample of these cases, beginning with another decision by the full court.
As part of a settlement of litigation protesting investigations calculated to stifle dissident groups, the Department of Justice promised to continue using a set of investigatory techniques specified in guidelines developed by Attorney General Levi in 1976. In 1983 Attorney General Smith promulgated a new set of guidelines, which like the Levi guidelines fully complied with all constitutional requirements but which differed from the Levi guidelines in respects the plaintiffs deemed material. Relying on the consent decree, the district court forbade implementation of the Smith guidelines to the extent they differed from the Levi guidelines. This court held, however, that despite one Administration’s promise, Attorneys General are entitled to reach, and implement, different conclusions about appropriate investigative techniques, so long as the new rules adhere to all legal norms. Alliance to End Repression v. Chicago, 742 F.2d 1007 (7th Cir.1984) (in banc). Although the case involved interpretation rather than modification of the decree, the court emphasized the importance of allowing the political branches of government to reach, and update, independent conclusions about law enforcement techniques. Doubts were to be resolved in favor of leeway for the political branches: “We doubt that in agreeing to the consent decree the Justice Department tied its hands to such an extent; for if it did, it was trifling with the public safety ... and maybe even violating the President’s constitutional obligation to ‘take Care that the Laws be faithfully executed.’ ” 742 F.2d at 1014. “And what did [it] get in return for abandoning that duty, if that is what it did? [The plaintiffs] would not have gotten a broader injunction if they had pressed the case to trial and won than they got in the consent decree.... The decree as it stands is a remarkable judicial intervention in vital executive functions; a proper decree formulated after a trial would not have been more Draconian.” Id. at 1016. Such considerations led the court to infer that the executive branch had not contracted away its powers after all. See also id. at 1020. Pretty much the same approach — with the same result of permitting the political branches of government to go on formulating new policies — -was applied in White v. Roughton, 689 F.2d 118 (7th Cir.1982) (permitting a town to enact new welfare legislation, although an earlier consent decree seemed to commit it to a different program in perpetuity, when the due process clause of the Constitution did not lock the town into the choice expressed in the decree). Cf. Secretary of Labor v. Fitzsimmons, 805 F.2d 682, 695-97 (7th Cir.1986) (in banc).
Consider, too, a dispute about using consent decrees to affect the budget. In order to induce Chicago to agree to a school desegregation remedy, the federal government promised financial aid. According to the consent decree, the United States would give Chicago priority in the allocation of available funds. A new Administration in Washington allocated Chicago less than it believed the agreement required, and six years of litigation ensued. We finally concluded that the consent decree should not be read to require automatic funding, or to compel the Presi[480]*480dent to seek additional appropriations, but only to require the Secretary of Education to exercise discretion in good faith within a range preserved by statutes and regulations. United States v. Board of Education, 799 F.2d 281 (7th Cir.1986). We adopted this reading in substantial measure because of concern that judges should not take control of the budgetary process even with the consent of the parties — and we were reluctant to enforce even such a limited commitment except for the fact that the federal government’s promise induced Chicago to undertake a program otherwise beyond its means, a program that protected the constitutional rights of children. Take away the detrimental reliance, and the bona fide constitutional claim, and there would have been little support for even the modest control of budgetary priorities reflected in that decree. See 799 F.2d at 297-98. In today’s case, by contrast, the decree governs the size and allocation of the City’s budget, did not engender reliance, and does not protect anyone’s constitutional rights.
Let us turn to a few cases about the entry of consent decrees. Electoral fraud has been a persistent blemish on the political system of Chicago. The Board of Election Commissioners agreed to resolve constitutional litigation by adopting a new regulatory system, including different dates for registration and canvasses, that it believed would reduce the amount of fraud. We held that entry of such a decree would be improvident, because regulation of election procedures should be left to the political process unless a particular procedure is essential to. cure an ongoing violation of federal law, and because “[a] federal court must preserve the appropriate relation between state and national power.” Kasper v. Board of Election Commissioners, 814 F.2d 332, 340 (7th Cir.1987). And we reminded the public officials of Chicago who tried to enter into this consent decree: “The Commissioners are agents, not principals; they need their principals’ approval to alter the terms of their agency. An alteration of the [state’s] statutory scheme may not be based on consent alone; it depends on an exercise of federal power, which in turn depends on a violation of federal law. The district court therefore properly insisted on a demonstration of at least a probable violation of that law as a condition to the entry of this decree.” Id. at 342. See also Dunn v. Carey, 808 F.2d 555, 559-60 (7th Cir.1986); Duran v. Elrod, 760 F.2d 756 (7th Cir.1985). Finding no probable violation of federal law that would be rectified by the parties’ agreement, this court held that there was no basis for entering a consent decree. People Who Care v. Rockford Board of Education, 961 F.2d 1335 (7th Cir.1992), similarly insists that the litigants establish a probable violation of federal law before the court may enter a consent decree that affects the rights of third parties. That is a fair description of the consent decree in our case, for a binding promise to pay all tort judgments promptly affects the allocation of the budget. Someone else must wait for payment; alternatively the City must raise taxes (or cut back on other things, such as police or education) to pay all creditors immediately. In any of these cases the decree affects strangers’ interests. See Larry Kramer, Consent Decrees and the Rights of Third Parties, 87 Mich.L.Rev. 321 (1988).
All of these cases illustrate the principle we recognize today: entry and continued enforcement of a consent decree regulating the operation of a governmental body depend on the existence of a substantial claim under federal law. Unless there is such a claim, the consent decree is no more than a contract, whose enforcement cannot be supported by the diversity jurisdiction and that has in court no more force than it would have outside of court. Accord, League of United Latin American Citizens, 999 F.2d at 847. Cf. Rizzo v. Goode, 423 U.S. 362, 379-80, 96 S.Ct. 598, 608-09, 46 L.Ed.2d 561 (1976). Prospective enforcement therefore is “inequitable” within the meaning of Rule 60(b)(5).
Evans I identifies two federal claims: one, under the equal protection clause, which it adjudicated in plaintiffs’ favor, and another, under the due process clause, on which it reserved judgment. The decree was properly supported when the district court entered the injunction in 1984. But Evans II properly overruled the equal protection holding of Evans I, and there can be little doubt that [481]*481Evans II would have repulsed a due process argument had the plaintiffs presented it for decision. The two lines of argument were conceived as complements, and the conclusion of Evans II that budgeting to pay tort judgments is a subject for political rather than judicial decision undermines all of plaintiffs’ theories.
At the time of Evans I, plaintiffs’ due process theory was that tort judgments are “property” and that state law calls for prompt payment. Failure to follow state law thus deprived them of the “property” right in prompt payment. Deferring payment indeed is problematic under Illinois law, which requires partial payment if full payment is postponed. 745 ILCS 10/9-104. But litigants aggrieved by Chicago’s failure to adhere to state law must take their claims to state court. Kasper, 814 F.2d at 340. The Constitution does not authorize federal judges to superintend state and local governments’ compliance with their own laws. Nordlinger v. Hahn, — U.S. -, - n. 8, 112 S.Ct. 2326, 2335 n. 8, 120 L.Ed.2d 1 (1992); id. at -, 112 S.Ct. at 2339-41 (Thomas, J., concurring); DeShaney v. Winnebago County Department of Social Services, 489 U.S. 189, 202, 109 S.Ct. 998, 1006, 103 L.Ed.2d 249 (1989); Snowden v. Hughes, 321 U.S. 1, 11, 64 S.Ct. 397, 402, 88 L.Ed. 497 (1944); Archie v. Racine, 847 F.2d 1211, 1215-18 (7th Cir.1988) (in banc). Quite the contrary, principles of respect for a coordinate sovereign (and in some cases the eleventh amendment) mean that federal courts should refrain from adjudicating claims under state law, whether raised directly or whether used as the springboards for other theories. “[I]t is difficult to think of a greater intrusion on state sovereignty than when a federal court instructs state officials on how to conform their conduct to state law.” Pennhurst State School & Hospital v. Halderman, 465 U.S. 89, 106, 104 S.Ct. 900, 911, 79 L.Ed.2d 67 (1984). State law may create property interests that give rise to federal procedural entitlements, see Archie, 847 F.2d at 1217, but establishing a budget is a legislative rather than a person-specific decision. A government need not offer individual hearings before adopting legislation with widespread effects. Atkins v. Parker, 472 U.S. 115, 129-31, 105 S.Ct. 2520, 2528-30, 86 L.Ed.2d 81 (1985).
Instead of contending that the due process clause requires state and local governments to follow their own law, plaintiffs might contend that the disparity between the statutory 6% rate of interest and the market rate of interest means that delay diminishes the value of the judgments, violating the due process clause or, more accurately, the takings clause applied to the states through the due process clause. Chicago, Burlington & Quincy R.R. v. Chicago, 166 U.S. 226, 17 S.Ct. 581, 41 L.Ed. 979 (1897). These arguments, too, belong in state court, which may compensate litigants for the erosion in the value of their judgments. Williamson County Regional Planning Commission v. Hamilton Bank, 473 U.S. 172, 105 S.Ct. 3108, 87 L.Ed.2d 126 (1985); Gamble v. Eau Claire County, 5 F.3d 285 (7th Cir.1993).
Recasting the due process argument as an objection to erosion of a judgment’s value during delay encounters additional obstacles. State and local law determines the damages awarded in tort litigation under these legal systems. There would be no constitutional obstacle to a rule saying, for example, that the court shall calculate the income the plaintiff lost because of the defendant’s tort and award 80% of that amount as damages. Workers’ compensation systems use this principle, and many other rules of law limit the damages recoverable in tort. Limitations on the damages recoverable from state and local governments in tort litigation are common. If a state may reduce the award directly, it may do so indirectly by computing-damages at 100% of loss and deferring payment while interest accrues at less than the market rate.
If the statutory interest rate matches the market rate, judgment holders receive full compensation for delay and can sell judgments in the secondary market for their face value, just as people buy and sell 30-year municipal bonds. A court that would not dream of declaring, on constitutional grounds, that a 6% post-judgment interest rate is unconstitutionally low or that a cap on awards in tort cases violates the due process [482]*482clause has no greater warrant for declaring that the judgment debtor is taking too long to pay; the calculation of the award, the interest rate, and the delay in payment are three facets of the same thing. (The post-judgment rate of interest in Illinois is 6% if the defendant is a unit of local government and 9% otherwise. 735 ILCS 5/2-1303. This difference does not pose any problem under the due process clause, and plaintiffs do not bother to contest it.)
Federal law excluding municipalities from involuntary bankruptcy, 11 U.S.C. § 904, may make cities more able (and thus more willing) to defer payment, but a decision by Congress to limit the scope of federal remedies does not justify stretching the due process clause. Some other elements of national law may bear on Chicago’s budgetary decisions. State and local governments do not determine the level of damages in cases arising under federal law, which implies a correspondingly reduced entitlement to diminish the value of these awards by delay. Chicago’s payment of contract before tort judgments may disfavor holders of federal judgments (most federal judgments against municipalities rest on 42 U.S.C. § 1983 and other civil rights statutes), which might offend the supremacy clause of the Constitution. And if Chicago pays interest on federal judgments at the local rate rather than the federal coupon issue yield equivalent rate, see 28 U.S.C. § 1961(a), then Chicago is in hot water under federal statutory law. Cf. Evans II, 873 F.2d at 1011 n. 7. Plaintiffs do not pursue any of these claims based on the difference between federal and state judgments, however, and the difference could not support the full scope of the relief awarded in this case.
To the objection that our approach interferes with the settlement of litigation, we have two replies. First, settlement is not an end in itself. It is a means of resolving disputes harmoniously. Many things are more important: preserving democratic governance, separating the judicial and political spheres, respecting state autonomy in the absence of a federal rule. These interests may elude a hectored district judge, eager to reach the next case in the queue, but to the political society whose long-term good the judge serves they are vital. Settlements purchased at the cost of putting the court in control of state and local budgets come at too high a cost. Second, the premise is incorrect. Attempts to enforce this consent decree have not produced the peace that settlement brings. There have been 16 years of noisome litigation. See also United States v. Board of Education, 799 F.2d at 288-89, 296-98. Politics is unruly and often unpleasant; judicial regulation of political affairs does not end the conflict but only shifts the venue.
Recognizing that a substantial federal claim must undergird a consent decree does not make plaintiffs less willing to settle: the decree still provides relief, which may be tailored more closely to the parties’ circumstances than a remedy of the judge’s devising could be. Plaintiffs’ alternative remains a trial, at which they might lose everything or obtain less suitable relief. From defendants’ perspective, the approach we have taken actually may make settlement more rather than less attractive. Knowledge that a change in the course of judicial decisions permits modification or withdrawal of the decree may make a responsible public official more willing to consent to relief based on the state of the law at the time. Public officeholders whose objectives include ensuring that their policies outlast the terms to which they had been elected will have less reason to settle, but, as we have explained, this is not a proper inducement to resolve litigation.
District judges need a substantial measure of discretion to deal with consent decrees, and with discretion comes deferential appellate review. Nothing in our discussion is designed to detract from these principles. What we have emphasized is a rule of law that the district judges must take into account when deciding which settlements to accept, and when to vacate consent decrees. Unless there is a substantial claim under federal law, the district judge should not enter or continue to enforce a consent decree affecting the operation of a governmental body. The district court did not find that after Evans II plaintiffs have a substantial claim under the due process clause, and any [483]*483such finding would have been legally erroneous. Thus the consent decree must be vacated, and the City’s budgetary decisions restored to the political process.
Reversed.