EASTERBROOK, Circuit Judge.
Rate regulation for electricity rests on the theory that generating stations and transmission lines are natural monopolies. Unless constrained, suppliers charge monopoly prices in order to satisfy their investors. States hold electric utilities to the cost of service (including a competitive rate of return on invested capital); the Federal Energy Regulatory Commission does the same for wholesale electricity transmitted across state borders, preempting state regulation. 16U.S.C. § 824(b); FPCv. Southern California Edison Co., 376 U.S. 205, 84 S.Ct. 644, 11 L.Ed.2d 638 (1964). When the electric utility is owned by its customers, however, there is no reason to think that the seller will charge a monopoly price — or that if it does the customers will care. If the firm charges a monopoly price, the profits return to the customers as dividends, deferred rebates on the purchase price.
The natural-monopoly explanation for regulation of investor-owned utilities has been challenged. E.g., George J. Stigler & Claire Friedland, What Can Regulators Regulate?: The Case of Electricity, 5 J.L. & Econ. 1 (1962), reprinted in Stigler, The Citizen and the State 61 (1975); see also Paul L. Joskow & Richard Schmalensee, Markets for Power: An Analysis of Electrical Utility Deregulation (1983). For customer-owned utilities there is no equivalent debate: either way, the customers win. Most states accordingly do not regulate the [448]*448prices charged by these firms. When the federal Rural Electrification Administration lends money to a utility — which it does at bargain rates for customer-owned firms — the Federal Energy Regulatory Commission drops its own system of price control, believing that customers of such firms don’t need “protection” from their own managers’ decisions. Dairyland Power Cooperative, 37 F.P.C. 12 (1967). See also Salt River Project Agricultural Improvement & Power District v. FPC, 391 F.2d 470 (D.C.Cir.1968).
Yet 11 states do regulate the prices charged by cooperative electric utilities. Arkansas Electric Cooperative Corp. v. Arkansas Public Service Commission, 461 U.S. 375, 103 S.Ct. 1905, 76 L.Ed.2d 1 (1983), holds that neither the hands-off attitude of the FERC nor any generally-applicable rule established by the REA preempts state regulation. Why might states choose to regulate the prices charged by customer-owned utilities? We can think of two answers, both of which start from the premise that the equity investors (shareholders, members) of a corporation are not its sole “owners”. Other persons also invest: creditors invest money in exchange for a promise of interest plus repayment; employees invest specialized skills that may not be so useful at other firms. These investors obtain contractual rights. The equity investors “own” only the rights to control and profit not assigned by contract to other investors. Electric utilities commonly depend more on debt financing than on equity financing. So perhaps the “real” owners — at least the effective controllers — of an electric co-op are the debt investors, who would be tempted to raise prices past the competitive level in order to make their loans more safe or hike up the rate of interest. When creditors control the firm, they may be inclined to set a monopoly price and not rebate the profits to the equity-investor-customers. Regulation then would play the same role it does in traditional theories.
If one possibility is that regulation prevents the debt investors from exploiting the equity investors, the other possibility is that regulation enables the equity investors to exploit the debt investors. Customer-investors — residents of the state who wield political influence there — may seek regulation of “their own” utility’s prices in order to depress revenues so greatly that they no longer can pay off the debt investors. They hope to emerge from bankruptcy with control of the firm and entitled to keep all future revenues for themselves. Without regulation, the debt investors may have contractual rights to force price increases sufficient to pay the debt in full; regulation plus bankruptcy law may provide a one-two punch allowing the customer-cum-equity-investor to escape these contractual provisions yet remain in control of the utility-
Neither the customer-exploits-cr editor hypothesis nor the creditor-exploits-customer hypothesis is troubling if the firm depends on a flow of fresh capital under new contracts; the need to return to capital markets frequently ensures that contractual terms are competitive, and anyone who violates commitments made then cannot be expected to be invited back. When, however, an event sufficiently large in the life of the firm or the industry occurs, it may be worth accepting the injury to one’s reputation for square dealing. In this “last period”, the gains from breach exceed any loss from fetters on future bargains.
I
According to the Rural Electrification Administration, the Wabash Valley Power Association has entered that last period and is doing its best to extract the value of loans the REA made. Wabash is a nonprofit generation-and-transmission utility whose customers operate in Indiana, Michigan, and Ohio. Indiana and Michigan regulate the rate customer-owned electric utilities may charge for power. Over the years the REA loaned Wabash a sum that, with unpaid interest and further advances to prevent other lenders from foreclosing, comes to about a billion dollars. Wabash promised the REA to set rates high enough to ensure that the principal and interest would be repaid. To make doubly sure, before lending money the REA required [449]*449Wabash to obtain state agencies’ approval for its projects. On top of that, the REA takes a security interest in a co-op’s assets, including its contracts with customers— contracts that require customers to take their requirements of electricity from the co-op at rates high enough to repay the loans. The REA thought that this combination of contractual devices would protect it. It has been proved wrong.
During the 1970s one of Wabash’s neighbors, the for-profit Public Service Company of Indiana, decided to construct the Marble Hill nuclear generating station. Wabash purchased a 17% interest in the project. To finance its share it borrowed $480 million on the strength of the REA’s guarantees. In January 1984 Public Service of Indiana abandoned construction of the Marble Hill station because costs had risen so high that further work would throw good money after bad. Wabash was out the money, and unless it raised its rates could not repay the lenders. It did not raise its rates and defaulted; the lenders exercised their rights under the REA’s guarantees, compelling it to pay in Wabash’s stead. The REA, pointing to its contract, instructed Wabash to file for rate increases sufficient to assure repayment. Wabash complied with transparent insincerity, making it clear to the state regulators that it would be delighted if they said no. (So the REA contends, and given the posture of this case we must take facts and inferences in the REA’s favor. Because as a practical matter Wabash serves the interests of its owner-customers, this interpretation of its behavior before the state agencies is plausible.)
Public Service of Indiana applied for its own rate increase and emerged empty-handed. Until 1984 the Indiana Utility Regulatory Commission believed that an investment prudent when made may be included in the rate base even though things turn out poorly. When Northern Indiana Public Service Co. (NIPSCO) applied for a rate increase to cover the expense of its abandoned Bailly N-l plant, the Commission granted the request. In re NIPSCO, No. 36689 (Aug. 11, 1982). The court of appeals reversed, Citizens Action Coalition v. NIPSCO, 472 N.E.2d 938 (2d Dist.1984), affirmed, 485 N.E.2d 610 (Ind.1985). Indiana’s courts concluded that unless an investment is “used and useful” the utility is not entitled to a return on its investment. Public Service of Indiana eventually agreed not to include Marble Hill in its rate base. In re Public Service Co. of Indiana, No. 37414 (Mar. 7, 1986).
Such a limitation is the automatic result of competition, for a firm that builds a useless plant cannot charge more than its rivals who have invested astutely. If all firms overbuild, then there will be “ruinous competition” as price falls to marginal cost, and no one can recover the cost of capital investment. Risk of ruination make capital more expensive (investors demand compensation for the chance that they will receive nothing). A risky business will have a higher normal rate of return on investment than a utility protected against failure by the exclusion of business rivals. So as Duquesne Light Co. v. Barasch, 488 U.S. 299, 109 S.Ct. 609, 617-18 n. 7, 102 L.Ed.2d 646 (1989), observes, a policy of disallowing a return on a utility’s unusable projects logically entails a higher rate of return on usable investments, and the Constitution may compel the adjustment. See also Richard J. Pierce, The Regulatory Treatment of Mistakes in Retrospect: Cancelled Plants and Excess Capacity, 132 U.Pa.L. Rev. 497, 525-32, 541-44 (1984).
Indiana allows its for-profit utilities to charge rates reflecting the greater risk created by the used-and-useful rule. Citizens Action Coalition v. Public Service Company of Indiana, Inc., 552 N.E.2d 834 (3d Dist.1990). For non-profit utilities the equation between greater risk (because of disallowing return on failed plants) with higher yield and greater assurance of recovery (by allowing a return on all investments prudent when made) with lower yield does not work. Because non-profit utilities are not supposed to make “profits” and do not have “equity investments” in the legal sense, the state does not allow investors any return on “investment”. It asks instead only what the firm’s “costs” are — wages, materials, and debt service. [450]*450Excluding a failed project from the rate base means that debt investors get nothing.
When asking for a rate increase, Wabash contended that the used-and-useful rule should be applied only to for-profit utilities, which alone are entitled to return on investment. Moreover, it contended (on instructions from the REA), in a co-op the members are risk-bearers in the same sense as the investors in a for-profit firm. Consequences of mistaken decisions come home to roost in for-profit firms in lower returns on investment; for non-profits, the penalty should be visited through higher rates. The United States intervened in the administrative proceedings to drive home this contention that non-profit utilities should be treated differently. The regulatory agency disagreed and applied the used-and-useful rule across the board. In re Wabash Valley Power Association, Inc., No. 37472 (Jan. 14, 1987). Wabash did not seek judicial review, but the United States did, joined by the National Rural Utilities Cooperative Finance Corporation (CFC). Both the appellate court, CFC v. Public Service Commission of Indiana, 528 N.E.2d 95 (Ind.App. 3d Dist.1988), and the Supreme Court of Indiana, 552 N.E.2d 23 (1990), agreed, the latter in a 3-2 decision. The majority reasoned that because co-ops were established to “provide low cost electricity to rural areas”, rates should not be allowed to rise even when the owner-members are responsible for the costly blunder, id. at 27. Because “[a] utility may impose a charge on its ratepayers only for service”, and Marble Hill provides none, “any charge purportedly imposed for service would have been inappropriate here.” Id. at 28.
Shortly after the court of appeals rendered its decision, the REA took matters into its own hands. On November 23, 1988, the acting administrator of the REA sent Wabash a letter (copies to regulatory officials in Indiana and Michigan) asserting federal control over Wabash's rates:
By this letter, we notify you that the REA hereby exercises jurisdiction over the rates charged by Wabash. Thus, the Indiana Commission and the Michigan Commission are no longer entitled to serve as regulatory authorities whose approval must be obtained for changes in Wabash’s rates. The Commissions may retain the right to exercise jurisdiction over matters unrelated to Wabash’s rates so long as Federal interests are not compromised.
Wabash is hereby instructed to raise its rates 9% across the board immediately upon occurrence of any of the following events [all having to do with approval of the bankruptcy court].
Wabash shall, in addition, raise its rates 9% across the board on the first and second anniversary of the first occurrence of any event listed above.
We believe that Wabash is capable of repaying its debt, and calculate that the required rate increases are the minimum necessary. We are requiring phased-in rate increases rather than an immediate one-time increase to enable Wabash and its members to adjust to the higher rate level.... We will notify you of any different or additional rate increases to be implemented, either as part of a plan of reorganization [in bankruptcy] or otherwise.
Wabash did not put the rate increases into effect or ask the bankruptcy court for permission to do so. Instead it filed this suit, seeking a declaratory judgment that the REA lacks authority to preempt state law and exercise direct control over its rates. Michigan intervened to support Wabash; Indiana has filed papers as amicus curiae backing Wabash’s views.
The district court granted Wabash’s motion for summary judgment. 713 F.Supp. 1260 (S.D.Ind.1989). It concluded that because formal published policies of the REA defer to state regulation, its letter to Wabash is ineffectual. Id. at 1266-67. Although Arkansas held open the possibility that state laws preventing the REA from recovering its loans might be preempted, the district court concluded that the REA’s failure to argue preemption in the state litigation forecloses it from raising the subject in a separate suit. Id. at 1267-69.
[451]*451II
Although none of the parties has raised the question, our initial inquiry concerns federal jurisdiction. Wabash, a debt- or in bankruptcy, has been told to raise its rates. The REA believes that the additional revenues will enable it to pay all of its debts and emerge prosperous. By contrast, the plan of reorganization proposed by Wabash’s members, based on the assumption that Wabash will not raise its rates, proposes to pay only about 50% of the debt to the REA, the senior secured lender. Junior creditors would get nothing, given the absolute priority rule of 11 U.S.C. § 1129(b)(2)(A). Wabash has filed this suit for the purpose of opposing the receipt of additional income. We can see why Wabash’s members oppose a rate increase, but they did not sue. Wabash itself appears to be better off as a result of the REA’s directive. Where is the injury in fact sufficient to create a case or controversy under Article III of the Constitution? See Love Church v. Evanston, 896 F.2d 1082, 1084-85 (7th Cir.1990).
Wabash is operating as a debtor in possession. A debtor in bankruptcy is supposed to maximize the value of the estate, which opposition to a rate increase does not. This suit serves the interests of Wabash’s members at the expense of the interests of Wabash’s creditors. We are surprised that the bankruptcy judge has allowed Wabash to operate as debtor in possession, when the clash of interests between creditors and its current management is so obvious. Nonetheless, the REA has not asked for the appointment of a trustee, see 11 U.S.C. § 1104(a). Because the bankruptcy court has allowed Wabash to prosecute this action (no one asked the court to block the suit), we ask only whether Wabash can establish injury in fact.
Wabash argues that it may suffer two injuries. First, if it charges the rates the REA demands, state regulatory officials may require it to disgorge the proceeds and pay penalties to boot. Second, rates exceeding those authorized by state law may put Wabash in violation of contracts with its members requiring them to purchase their requirements of electricity from Wabash. Freed of their commitment, these members may turn elsewhere for wholesale power (if cheaper power is to be had); the loss of business may make the price increase unprofitable. Both of these sequences are speculative. We do not know the penalties that would apply if the REA’s attempt to preempt state law should fail; more, unless the states require rebates at the current interest rate, it could be profitable to collect revenues now and return them later with penalties. As for the members’ willingness to turn elsewhere for power: Wabash does not represent to us that its members have wholesale power available at less than Wabash’s current rates plus the increase the REA demands. Still, it is possible that such power may be available in the future, and that higher rates now may lead to losses later.
Although this is thin gruel, it is no thinner than the claim in Duke Power Co. v. Carolina Environmental Study Group, Inc., 438 U.S. 59, 98 S.Ct. 2620, 57 L.Ed.2d 595 (1978), that but for a law limiting the damages that could be assessed against the operator of a nuclear plant, the plant would not be built and the risks of radiation would be reduced. The Court thought this sufficient to establish the standing of persons who live near the proposed plant, id. at 72-81, 98 S.Ct. at 2629-35, even though the effect of the law on construction was far from clear. (The law put a cap on liability, but the corporate form does that automatically. Firms could have established even tighter caps on liability by holding their plants as subsidiaries.) Wabash has a logical causal chain between the federal decision and potential injury. This gives it standing. There is injury in fact, and the injury is redressable by a favorable decision.
We say this cognizant that if injury befalls Wabash, the REA bears the cost. The REA is the residual claimant of the firm. Unless Wabash raises its rates, the REA stands to lose about half of its billion-dollar investment. If Wabash raises the rates only to find, because customers switch or states impose penalties, that the firm is [452]*452even worse off, the REA bears the entire loss. Obviously the REA does not perceive much risk of these losses; it has no reason to compel Wabash to take steps that will reduce the realization on the debt. Still, just as we do not pierce the corporate veil to find standing in the interests of Wabash’s members, cf. In re Kubly, 818 F.2d 643 (7th Cir.1987), so we respect the corporate form rather than look straight to the interests of the REA. Wabash, and not the REA or its members, is the plaintiff. Wabash’s exposure to loss as a result of a price increase is enough, if barely, to allow a decision by an Article III court.
Prudential concerns do not counsel dismissal. Behind the haze created by the corporate form and the bankruptcy proceedings, there is a substantial and justicia-ble dispute: who bears the loss of Marble Hill, the federal government or Wabash’s members? Once we pass from the constitutional test to prudential concerns, it becomes significant that Wabash is acting as a tool of its members. Although a corporation technically does not “represent” its investors, many cases allow organizations to represent the interests of their members. See Auto Workers v. Brock, 477 U.S. 274, 281-88, 106 S.Ct. 2523, 2528-32, 91 L.Ed.2d 228 (1986). Moreover, if we were to throw the case out the sides would switch, and it would come back immediately. Wabash would refuse to raise its rates; the REA would file suit to compel it to do so, Wabash would stand on state law, and the legal questions would recur. These questions must eventually be settled between these parties, and now is as good a time as any.
A second jurisdictional question is whether Congress has authorized a suit in the district court. Wabash contends that an agency’s action is contrary to law. The Administrative Procedure Act authorizes declaratory judgment actions in such circumstances, see Gardner v. Toilet Goods Ass’n, 387 U.S. 167, 87 S.Ct. 1526, 18 L.Ed.2d 704 (1967), and an amendment to the APA in 1976, see 5 U.S.C. § 702, largely waives the sovereign immunity of the United States in equitable cases. Bowen v. Massachusetts, 487 U.S. 879, 108 S.Ct. 2722, 101 L.Ed.2d 749 (1988). The difficulty is the Tucker Act, 28 U.S.C. §§ 1346(a), 1491, which assigns to the Claims Court all actions seeking equitable relief based on a contract with the United States. Wabash contends that when it borrowed the money, the REA agreed to the application of state law. An effort to obtain specific performance of that promise (if the REA made it, a matter we do not discuss) belongs in the Claims Court. See Sharp v. Weinberger, 798 F.2d 1521, 1523-24 (D.C.Cir.1986) (Scalia, J.).
In the district court the REA argued that Wabash’s invocation of contractual rights required the district court to transfer the entire ease to the Claims Court. The district court disagreed, holding that it may issue declaratory relief so long as that relief is not grounded in contract. 713 F.Supp. at 1263-64. We agree with that conclusion for the reasons the district judge gave. See also Sharp, 798 F.2d at 1524; North Side Lumber Co. v. Block, 753 F.2d 1482, 1486 (9th Cir.1985). Notwithstanding its recognition of jurisdictional limits, the district court dwelt on the REA’s promises to Wabash. 713 F.Supp. at 1269-70. Because any effort to obtain specific performance of a contract with the United States belongs in the Claims Court, the district judge should not have considered Wabash’s contractual claims. We put these to one side and consider only whether the arguments based on the APA support the judgment of the district court.
Ill
Arkansas holds that as a rule states may regulate the prices charged by co-ops to which the REA has extended credit. The Court mentioned two possible exceptions:
There may come a time when the REA changes its present policy, and announces that state rate regulation of rural power cooperatives is inconsistent with federal policy. If that were to happen, and if such a rule was valid under the Rural Electrification Act, it would of course preempt any further exercise of jurisdiction by the Arkansas PSC.... [453]*453Moreover, even without an explicit statement by the REA, a particular rate set by the Arkansas PSC may so seriously compromise important federal interests, including the ability of the [debtor] to repay its loans, as to be implicitly preempted by the Rural Electrification Act.
461 U.S. at 388-89, 103 S.Ct. at 1914-15. The REA contends that its letter of November 1988 is a change of policy and expressly preempts state law. It also contends that the rates set by Indiana and Michigan “so seriously compromise important federal interests, including the ability of the [debt- or] to repay” that they are preempted implicitly. We discuss these in turn.
A
The Supreme Court did not say that the REA may preempt state law at will. It said that if the REA were to change its policy, and “if such a rule was valid under the Rural Electrification Act”, then state law must yield. There are three questions: has the REA changed its policy?; does it have substantive authority to replace state with federal regulation?; has it exercised this authority in the proper way?
At the time the Court decided Arkansas, language in the REA’s Bulletin 111-4 recognized that states may regulate the rates of its borrowers. The language of Bull. 111-4 on which the Court relied, 461 U.S. at 387-88,103 S.Ct. at 1914, is still in force. The REA has not yet changed its “policy”, although it has attempted to force Wabash to change its rates. The REA has not established its own ratemaking apparatus or said that rates must be left to the market. Instead it has left the general rule in place and has prescribed for Wabash a rate that has nothing to do with anything except repaying the loan. This is not the sort of express preemption Arkansas contemplated.
There is, moreover, substantial doubt about whether a rule preempting state regulation would be “valid under the Rural Electrification Act”. Neither REA’s letter to Wabash nor its brief in this court cites any provision of the statute allowing it to regulate the rates charged by its borrowers. Unless the REA has this authority, it is hard to see how it can preempt state law, given the holding of Arkansas that federal law does not occupy the field. REA needs a source of authority to set up a system making demands inconsistent with those of state law, so that under the Supremacy Clause the federal obligation would prevail.
Congress’ assumption when it created the Rural Electrification Act in 1936 seems to have been that the REA could protect itself by setting conditions on the extension of credit. (The opinions in Arkansas collect the legislative history, which we do not rehearse.) The REA may refuse to make loans unless state agencies guarantee rates high enough to allow repayment. It may refuse to approve a loan unless the customers (whose promises to buy power are the real security for the debt) guarantee the loan or authorize the REA to dictate rates. If either state or customer balks, the REA may deploy its funds elsewhere. Since Indiana’s adoption of the used-and-useful standard, the REA has refused to make or guarantee new loans to co-ops in that state. The conditioning power is a big club, which can achieve (effective) preemption and more. See Trans-Alaska Pipeline Rate Cases, 436 U.S. 631, 98 S.Ct. 2053, 56 L.Ed.2d 591 (1978). But REA did not use this power when extending credit. It approved contracts with Wabash’s customers that oblige them to pay no more than the state rate; it did not secure guarantees.
We need not decide whether the REA has statutory authority to abandon Bull. 111-4 and adopt a general system of regulation. It has issued a notice of proposed rulemaking, 55 Fed.Reg. 12194 (Apr. 2, 1990), to consider thoroughgoing substitution of federal for state regulation. It would be inappropriate to interrupt that process by deciding in advance whether the REA has statutory authority. Its effort to preempt state regulation of Wabash’s rates alone founders on a mundane obstacle: it neglected to use the procedures required by the APA.
In order to preempt state authority, the REA must establish rules with the force of [454]*454law. Regulations adopted after notice and comment rulemaking have this effect. Fidelity Federal Savings & Loan Ass’n v. de la Cuesta, 458 U.S. 141, 153, 102 S.Ct. 3014, 3022, 73 L.Ed.2d 664 (1982) (“[federal regulations have no less pre-emptive effect than federal statutes.”). Yet although ratemaking is a “rule” under the APA, 5 U.S.C. § 551(4), the REA did not follow the procedures the APA prescribed for rule-making. It sent Wabash a letter. There was no notice, no opportunity for comment, no statement of basis, no administrative record, no publication in the Federal Register—none of the elements of rulemaking under the APA. 5 U.S.C. § 553. See Thomas v. New York, 802 F.2d 1443 (D.C.Cir.1986) (Scalia, J.) (holding that a letter from the Administrator of the EPA does not have the force of law because it was not issued after notice and comment and was not published in the Federal Register). See also Brock v. Cathedral Bluffs Shale Oil Co., 796 F.2d 533, 538 (D.C.Cir.1986) (Scalia, J.) (rule not published in the Federal Register is not a regulation). Cf. Cosby v. Ward, 843 F.2d 967, 979-81 (7th Cir.1988).
In support of its submission that a letter may preempt state law, the REA relies exclusively on Blum v. Bacon, 457 U.S. 132, 140-41, 102 S.Ct. 2355, 2360-61, 72 L.Ed.2d 728 (1982). This is a baffling citation, because the case involved several regulations, published in the Federal Register after full notice and opportunity for comment. The case did not involve an informal procedure remotely similar to the one the REA followed. We have not found any case holding that a federal agency may preempt state law without either rulemak-ing or adjudication. Perhaps closest to the mark is a case saying that a consent decree negotiated between an auto manufacturer and the Federal Trade Commission may preempt state regulation of the manufacturer’s warranty policies. General Motors Corp. v. Abrams, 897 F.2d 34 (2d Cir.1990) (holding that the decree is not inconsistent with the state's rule). The Second Circuit did not discuss Martin v. Wilks, — U.S. -, 109 S.Ct. 2180, 104 L.Ed.2d 835 (1989), which holds that a consent decree may not foreclose the rights of non-parties. Because neither the state nor the consumers were parties to the FTC’s case, it is hard to understand how the decree could blot out their claims based on state law. Whether the decree has such an effect should depend on whether it was adopted by the agency as its own policy following the procedures the APA requires; then the preemption would come from substantive rules rather than the parties’ assent. See United States v. AT & T, 552 F.Supp. 131 (D.D.C.1982), affirmed without opinion under the name Maryland v. United States, 460 U.S. 1001, 103 S.Ct. 1240, 75 L.Ed.2d 472 (1983), holding that an antitrust consent decree may preempt state law when the states have had an opportunity to intervene and their contentions have been resolved on the merits. It is unnecessary to pursue the question. There has been no consent and no rulemaking. Procedural shortcomings prevent giving this letter the force of law.
B
The other string to the REA’s bow is its contention that when a state sets a rate so low that the federal advance cannot be repaid, that rate is preempted. To the extent the REA believes that states must adopt rules of law that always allow federal lenders to recoup their investments, it is mistaken. See United States v. Kimbell Foods, Inc., 440 U.S. 715, 99 S.Ct. 1448, 59 L.Ed.2d 711 (1979). Although obligations to federal lenders are determined under federal law, see Kimbell and, e.g., D’Oench, Duhme & Co. v. FDIC, 315 U.S. 447, 456, 62 S.Ct. 676, 679, 86 L.Ed. 956 (1942), the obligation here runs from Wabash to the REA; the REA seeks to affect relations between Wabash and its customers, not relations between Wabash and the federal government. What is more, Kim-bell holds that the federal rule governing the relation between the United States and its borrowers will be taken from state law, provided that law is not hostile to federal interests. The REA does not suggest that Indiana has depressed Wabash’s rates because the federal government is its princi[455]*455pal lender. Wabash received the same treatment as Public Service Co. of Indiana (which does not have federal loans), and under state law the REA receives the same treatment as any other lender to Wabash. The government seeks not neutrality but preference.
Its best line of argument is based on D’Oench, which holds that when the federal government takes over a failed bank, the relations between the bank and its customers are governed by federal law. See also Public Service Co. of Indiana v. Hamil, 416 F.2d 648 (7th Cir.1969). The REA has not taken over Wabash, so the state is not trying to regulate the federal government. Yet once the debt to the REA becomes the residual claim—the one that will be paid from any new receipts—it might make sense to say that Wabash and the REA are fused. When a borrower from the federal government is far under water, regulation of it is regulation of the United States.
Even this analogy may not be sufficient, however, because the argument that states may not regulate the federal government comes from intergovernmental tax immunity cases. In recent years the Court has held that states may levy taxes that as a practical matter fall on the federal treasury, provided the taxation is neutral between governmental and private payors. Washington v. United States, 460 U.S. 536, 103 S.Ct. 1344, 75 L.Ed.2d 264 (1983); United States v. New Mexico, 455 U.S. 720, 102 S.Ct. 1373, 71 L.Ed.2d 580 (1982). Neutral regulation that affects the REA’s coffers may be treated similarly. After all, the REA made its loans and guarantees on the assumption that state regulation would apply. Lack of guarantees from Wabash’s customers—coupled with the obvious incentives of a customer-owned borrower—may have led to an adjustment in the interest rate, which, in turn, increases the rates Wabash charges. (If there was no adjustment, people at the REA were asleep on the job.) When the risk materialized, the REA has demanded shelter of federal law plus the interest rate that reflected the risk created by state law. Nice cushion, if you can get it.
We need not decide whether the REA is entitled to a federal rule that protects agencies that did not protect themselves. (More realistically, a rule that protects the Treasury from federal employees who neglected to take precautions and private parties who take advantage of their omissions.) Once again, we stop short of a decision on the merits because of a procedural bar. The REA was a party to the administrative proceedings and obtained review from the state courts. It did not argue that Wabash’s rates should be increased because federal law preempts the used-and-useful rule or otherwise requires the state to set rates high enough to repay the loans. Its argument was based on state law. Under 28 U.S.C. § 1738 the final decision in the REA’s case blocks further litigation in federal court if a state court would deem the judgment preclusive. Principles of claim preclusion apply to the United States as to other litigants. United States v. Stauffer Chemical Co., 464 U.S. 165, 104 S.Ct. 575, 78 L.Ed.2d 388 (1984); Montana v. United States, 440 U.S. 147, 99 S.Ct. 970, 59 L.Ed.2d 210 (1979). So the question is whether Indiana would allow the REA, having challenged an administrative decision in state court under state law and lost, to urge in a different suit that state law is inconsistent with federal law. We think the answer is no.
Alleghany Corp. v. Haase, 896 F.2d 1046, 1049 (7th Cir.1990), assumes a negative answer to this question under Indiana law. See also Button v. Harden, 814 F.2d 382 (7th Cir.1987) (Illinois law). Indiana, like most other states, does not allow a party to split a single claim for relief. Multiple legal theories supporting relief on account of one transaction must be litigated at one go. Fischli v. Fischli, 1 Blackford 360, 361 (Ind.1825); Comparet v. Hanna, 34 Ind. 74, 77-80 (1870); Hammond Pure Ice & Coal Co. v. Heitman, 221 Ind. 352, 47 N.E.2d 309 (1943). We summarized Indiana’s law of preclusion in Leal v. Krajewski, 803 F.2d 332, 334 (7th Cir.1986), observing that claim preclusion follows from a combination of (1) a judgment rendered by a court of competent jurisdiction, [456]*456(2) on the merits, (3) in a suit between the same parties, where (4) the matter now at issue was, or could have been, decided. All four elements are present here. The REA could have raised its preemption argument in state court but withheld it.
The agency reminds us that many Indiana decisions say that only matters “within the issues” of the first case — that is, subjects on which “the same evidence will support the issues involved in both actions” — are deemed precluded. E.g., M.R. v. Meltzer, 487 N.E.2d 836, 841 (Ind.App. 4th Dist.1986); Biggs v. Marsh, 446 N.E.2d 977, 982 (Ind.App. 3d Dist.1983). As we understood it in Leal, this is another way of asking whether the two suits deal with a common occurrence. Two different legal theories in support of relief (state law in the state court, preemption in federal court) do not arise out of separate occurrences, and a party who neglects to put in the evidence that would support the withheld legal theory cannot later point to its own omission as justification for filing another suit. See Ulrich v. Drischell, 88 Ind. 354, 359 (1882); Jones v. American Family Mutual Insurance Co., 489 N.E.2d 160, 166 (Ind.App. 2d Dist.1986); Moxley v. Indiana National Bank, 443 N.E.2d 374 (Ind.App. 4th Dist.1983). Such an approach would gut the rules of preclusion. Like today’s case, Leal deals with a suit in state court to challenge an administrative decision on state grounds, followed by a suit in federal court contesting the decision on federal grounds. We held that Indiana would deem the federal grounds foreclosed because they had been withheld in state court, when the federal grounds supplied only a different legal rationale for relief arising out of a single occurrence. The same conclusion is appropriate here.
Anticipating that we would agree with the district court’s understanding of Indiana’s law, the REA tries to sidestep § 1738 on the ground that it expressly told the state court that it was reserving all federal questions for federal court. It refers us to a footnote in Haring v. Prosise, 462 U.S. 306, 313 n. 7, 103 S.Ct. 2368, 2373 n. 7, 76 L.Ed.2d 595 (1983), reserving the question whether preclusion would be appropriate in a case under 42 U.S.C. § 1983 in which a litigant has withheld a federal question from state court. When Haring was decided, a debate was under way about whether § 1738 applies in full measure to cases under § 1983. Our case does not present that question. More to the point, Migra v. Warren City School District Board of Education, 465 U.S. 75, 104 S.Ct. 892, 79 L.Ed.2d 56 (1984), a case the REA does not cite, resolves adversely to it even the question left open in Haring. See also Village of Bolingbrook v. Citizens Utilities Co., 864 F.2d 481, 485 (7th Cir.1988). Section 1738 requires the application of state law. Indiana does not allow a party to split its claim just because it has announced distinctly to the tribunal that it is going to do so.
Echoing United States v. Michaud, 897 F.2d 264, 267 (7th Cir.1990), we express concern over what seems to be an emerging pattern at the Department of Justice of citing cases while not informing the court that they are no longer authoritative or do not support the proposition. The Department’s cites United States v. Mendoza, 464 U.S. 154, 104 S.Ct. 568, 78 L.Ed.2d 379 (1984), which holds that offensive non-mutual issue preclusion does not apply to the United States, while omitting Stauffer and Montana, which hold that claim preclusion applies in full force. The REA contends that because preemption is a question of federal law it is particularly appropriate for federal court, so that ordinary principles of preclusion do not apply. Marrese v. American Academy of Orthopaedic Surgeons, 470 U.S. 373, 105 S.Ct. 1327, 84 L.Ed.2d 274 (1985), which the REA does not mention in this part of its brief, is to the contrary. It contends that, despite § 1738, claim preclusion “is an equitable doctrine subject to equitable principles” (Br. 36), citing Premier Electrical Construction Co. v. National Electrical Contractors Association, Inc., 814 F.2d 358, 364 (7th Cir.1987), which does not support the proposition (it did not involve either § 1738 or claim preclusion), while ig[457]*457noring Federated Department Stores, Inc. v. Moitie, 452 U.S. 394, 101 S.Ct. 2424, 69 L.Ed.2d 103 (1981), which rejects it. And as we have observed the government’s brief relies (exclusively!) on Blum v. Bacon for the proposition that letters from federal administrators may preempt state law, although Blum involved formal regulations rather than letters. This is not a pattern to be welcomed.
One final potential limitation on preclusion. Indiana’s courts rejected the REA’s challenge to the administrative decision. Michigan has not had an opportunity to address the matter, and it is at least conceivable that Michigan would allow the United States to try again. Michigan would be bound to allow litigation on the question whether it, like Indiana, applies a used-and-useful rule in the first place to non-profit utilities. That subject was not and could not have been decided in the Indiana case. Would Michigan also let the REA argue that state law has been preempted? Perhaps. Section 1738 provides, however, that in federal court the preclusive force of a state judgment depends on its effect within the rendering state, rather than within some other state that has an interest in the controversy. At all events, none of the parties contends that Michigan would allow litigation of preemption in its courts following the omission in Indiana. We limit our reach to the subjects that have been presented.
Everything we can see about the incentives of Wabash’s members and the operation of Indiana’s law suggests that federal taxpayers have been taken to the cleaners. States that show creditors the door may expect loans to be few and costly. The REA has stopped lending there. Other creditors will charge more. Indiana’s economy will suffer. Indiana seeks to limit the damage through a 1987 law that abrogates the used-and-useful rule for new plants, and allows rates to be based on all prudent investment, provided the utility obtains the state commission’s approval in advance. Ind.Code § 8-1-8.5-1 et seq. If this works, Indiana will have transferred to the federal government and to other investors (most from out of state) the cost of its adventure in nuclear power, while keeping its industry healthy for the future. Creditors may be skeptical of Indiana’s promise; at oral argument counsel for Michigan said that his state would feel free to repudiate such a promise if new legislators or regulators had a different view. Cf. Chemical Bank v. Washington Public Power Supply Co., 102 Wash.2d 874, 691 P.2d 524 (1984). None of this matters today, for the REA lacks a sound legal basis on which to preempt Indiana’s law. Appointment of a trustee in the bankruptcy court might have been the best recourse, but that is water under the bridge.
Affirmed.