Outboard Marine Corporation is in Chapter 7 bankruptcy, and among its holdings are the assets, currently worth some $14 million, in what is known as a "rabbi trust." Bank of America, as the agent of Outboard's secured creditors, claims a security interest in these assets, while the trustee in bankruptcy claims them for the unsecured creditors. The security agreement on which Bank of America relies covers all Outboard's "general intangibles," a term of great breadth in commercial law, see UCC § 9-102(a)(42) and official comment 5(d), and broadly defined in the agreement as well to include, besides a number of irrelevant enumerated items, "all other intangible personal property of every kind and nature." The term describes the assets of the rabbi trust, but the bankruptcy court, seconded by the district court, held that they nevertheless were not subject to the security agreement, and so ruled for the trustee. The ruling was a final, appealable order because it resolved a discrete dispute that, were it not for the continuing bankruptcy proceedings, would have been a stand-alone dispute between Bank of America and the trustee as the representative of the general creditors. In re Golant, 239 F.3d 931, 934 (7th Cir.2001); In re Rimsat, Ltd., 212 F.3d 1039, 1044 (7th Cir.2000). "A judgment does not lose its finality merely because there is uncertainty about its collectibility, corresponding to uncertainty about how many cents on the dollar the creditor will actually receive on his claim once all the bankrupt's assets are marshaled and compared with the total of allowed claims, and the priorities among those claims are determined. Thus the fact that the bankruptcy proceeding continues before the bankruptcy judge does not preclude treating an interlocutory order by him—interlocutory in the sense that it does not terminate the entire proceeding—as final for purposes of appellate review. (And if it is final for those purposes, then so is the district court's affirmance of his order.)" In re Szekely, 936 F.2d 897, 899 (7th Cir.1991).

A rabbi trust, so called because its tax treatment was first addressed in an IRS letter ruling on a trust for the benefit of a rabbi, Private Letter Ruling 8113107 (Dec. 31, 1980); see also IRS General Counsel Memorandum 39230 (Jan. 20, 1984), is a trust created by a corporation or other institution for the benefit of one or more of its executives (the rabbi, in the IRS's original ruling). See, e.g., Westport Bank & Trust Co. v. Geraghty, 90 F.3d 661, 663-64 (2d Cir.1996); Hills Stores Co. v. Bozic, 769 A.2d 88, 99 (Del.Ch.2000); Kathryn J. Kennedy, "A Primer on the Taxation of Executive Deferred Compensation Plans," 35 John Marshall L.Rev. 487, 524-27 (2002). The main reason (recited at the outset of the trust document in this case) for such a trust is that, should the control of the institution change, the new management might reduce the old executives' compensation, or even fire them; the trust, which consistent with this purpose is not funded until the change of control occurs, cushions the fall.

But as the IRS explained in the letter ruling, unless an executive's right to receive money from the trust is "subject to substantial limitations or restrictions," rather than being his to draw on at any time (making it income to him in a practical sense), the executive must include any contribution to the trust and any interest or other earnings of the trust in his gross income in the year in which the contribution was made or the interest obtained. See McAllister v. Resolution Trust Corp., 201 F.3d 570, 572-73, 575 (5th Cir.2000). The "substantial limitations or restrictions" condition was satisfied in the transaction on which the IRS ruled. The trust agreement provided that the rabbi would not receive the trust assets until he retired or otherwise ended his employment by the congregation. Until then the corpus of the trust and any interest on it would be owned by the congregation, see Maher v. Harris Trust & Savings Bank, 75 F.3d 1182, 1185 (7th Cir.1996); Goodman v. Resolution Trust Corp., 7 F.3d 1123, 1125 (4th Cir.1993), so the rabbi would have neither legal nor equitable right to the money. Cf. 26 U.S.C. § 457(f)(1)(A). And, what is key in this case, the trust instrument provided that "the assets of the trust estate shall be subject to the claims of [the congregation's] creditors as if the assets were the general assets of [the congregation]."

The word "creditors" is not defined either in the IRS's letter ruling or in the trust agreement in this case; but a "Model Rabbi Trust" agreement approved by the IRS states that the assets of the trust are subject to the claims of the settlor's "general creditors," Rev. Proc. 92-64, 1992-2 C.B. 422 (July 28, 1992), a term invariably used to refer to a debtor's unsecured creditors. See, e.g., United States v. Munsey Trust Co., 332 U.S. 234, 240, 108 Ct. Cl. 765, 67 S. Ct. 1599, 91 L. Ed. 2022 (1947); Dewsnup v. Timm, 502 U.S. 410, 431-32, 112 S. Ct. 773, 116 L. Ed. 2d 903 (1992) (dissenting opinion); In re Merchants Grain, Inc., 93 F.3d 1347, 1352 (7th Cir.1996); United States v. One-Sixth Share, 326 F.3d 36, 44 (1st Cir.2003); United States v. Watkins, 320 F.3d 1279, 1283 (11th Cir.2003); United States v. $20,193.39 U.S. Currency, 16 F.3d 344, 346 (9th Cir.1994); Douglas G. Baird, The Elements of Bankruptcy 12, 101, 154, (3d ed.2001). The cases assume rather than hold that "general creditor" means "unsecured creditor," but what else could it mean? What work does "general" do unless to distinguish unsecured from secured creditors? Bank of America has no answer to that question.

Outboard is conceded to have established a bona fide rabbi trust, so that its contributions to the trust and the income that those contributions generated were not includible in the executives' gross income. Therefore, if the validity of a rabbi trust depends on its assets' being reserved for the employer's unsecured creditors, we can stop right here and affirm; the Bank of America, as a secured creditor, would have no right to the assets—otherwise the trust's beneficiaries would not have received the favorable tax treatment accorded the beneficiaries of a rabbi trust, and they did receive it. But it is uncertain whether such a reservation actually is essential to the favorable tax treatment of a rabbi trust. All that the tax law requires is that there be substantial limitations on the beneficiaries' access to the trust assets, and a reservation of the assets in the event of bankruptcy to both the secured and the unsecured creditors of the settlor, rather than to the unsecured creditors, might well be thought substantial. For the reservation would keep those assets, most of them at any rate, out of the beneficiaries' hands—though this is provided that the limitation were coupled with a limitation on the beneficiaries' having free access to the assets of the trust before they leave their employment with the grantor. Without such a limitation, the reservation of creditors' rights would be illusory—the beneficiaries would pull the money out of the trust as soon as insolvency loomed on the horizon—and indeed the trust's assets might well be taxable as income to the beneficiaries. But we recall that, consistent with this concern, the assets of the rabbi trust were owned by the congregation until the rabbi's employment ended.

We say that a limitation to all, rather than just to the unsecured, creditors "might be" rather than "would be" substantial enough to satisfy the Internal Revenue Service because executives often are creditors of their firm; if they were secured creditors and their security interest embraced the assets of the trust, their claims to those assets would be superior to those of the firm's unsecured creditors, which would tend to make the limitation that is fundamental to the favorable tax treatment of the rabbi trust—that the creditors have a superior claim to the beneficiaries—illusory. But the trust instrument in this case took care of that concern by providing that Outboard's executives could not obtain a security interest in the trust's assets.

Even if the executives would not have sacrificed their favorable tax treatment had the trust instrument reserved the assets of the trust for all the company's creditors, secured and unsecured alike, in the event of bankruptcy, the instrument did not do this; it reserved those assets for the unsecured creditors. It states (we italicize the key terms) that the "Trust Corpus ... shall remain at all times subject to the claims of the general creditors of [Outboard]. Accordingly, [Outboard] shall not create a security interest in the Trust Corpus in favor of the Executives, the Participants [a term that apparently refers to retired executives] or any creditor." In the event of insolvency, the trustee "will deliver the entire amount of the Trust Corpus only as a court of competent jurisdiction, or duly appointed receiver or other person authorized to act by such court, may direct to make the Trust Corpus available to satisfy the claims of the Company's general creditors."

This couldn't be clearer: secured creditors have no claim to the trust assets. And judges usually interpret written contracts (the instrument creating the rabbi trust in this case was an agreement nominally between Outboard and the trustee of the trust, Northern Trust Company, but realistically between Outboard and the executives who were the beneficiaries of the trust, see Westport Bank & Trust Co. v. Geraghty, supra, 90 F.3d at 663-64) according to the conventional meaning of their terms, that is, literally. This is especially appropriate in the case of a negotiated contract involving substantial stakes between commercially sophisticated parties, as in this case, who know how to say what they mean and have an incentive to draft their agreement carefully. Such a style of interpretation protects the parties against the vagaries of the litigation process—a major reason for committing contracts to writing—without too great a risk of misinterpretation.

But literal interpretation of written contracts, even when the parties are sophisticated and the stakes substantial, is merely presumptively the right approach to take. Even sophisticated lawyers and businessmen sometimes stumble in their use of language, or use language that is specialized to their trade and departs from normal usage, or fail to anticipate contingencies that may make the language of the contract yield absurd results if it is read literally, and if these circumstances are evident to the court the contract will not be interpreted literally. Bank of America argues in this vein that of course all that Outboard intended to do in the passages of the trust agreement that we quoted was to create a rabbi trust, that is, a grantor trust that would enjoy a favorable tax status, and so if a rabbi trust does not necessarily forfeit its favorable tax status by reserving the trust assets for secured as well as unsecured creditors, neither does the trust agreement. The security agreement, which we quoted at the beginning of this opinion, contains no language to suggest that the assets of the rabbi trust would be excluded from Bank of America's security interest just because they are pledged to any creditor and not just to unsecured creditors.

This argument is not negligible but neither is it sufficiently compelling to rebut the presumption in favor of literal interpretation to which we referred. Rather the contrary. The language of the Model Rabbi Trust would make it natural for Outboard to assume that to create a valid rabbi trust it would have to reserve the trust's assets for its general creditors, which undoubtedly it would understand to mean its unsecured creditors. The assumption may have been incorrect, more precisely may have been excessively cautious; but it provides the best guide to the meaning that Outboard and the executives ascribed to the agreement. The executives in particular would tend to favor the cautious approach rather than jeopardize their tax benefits for the sake of Outboard's secured creditors. And though they might benefit indirectly, and Outboard directly, from the company's being able to pledge more of its assets to secure a loan to the company, this benefit—since the assets in a rabbi trust are likely to be only a small fraction of the company's total assets—would probably be outweighed by the risk of forfeiting favorable tax treatment by departing from the template of the Model Rabbi Trust.

The trust agreement does not merely reserve the trust's assets for the general creditors, moreover; it forbids Outboard to create a security interest in favor not only of the executives (which might make the trust illusory and forfeit the beneficiaries' favorable tax treatment) but also of any creditor. So even if Outboard thought that the term "general creditors" includes secured creditors, the agreement explicitly forbids the creation of a security interest in the trust assets. The trust instrument took as it were the extra step to make clear that the parties really intended to reserve the trust assets for Outboard's unsecured creditors. The security agreement, as we said, does not exclude the assets in the rabbi trust; but to determine what assets it does include (because they are not listed in the agreement), one must look beyond the security agreement. And when one looks one finds the trust instrument, which excludes those assets. It is important to note in this connection that the rabbi trust was funded before the security agreement between Outboard and Bank of America was executed. Had it been funded after, Outboard's contribution of assets to the trust would have been subject to the security agreement regardless of the terms of the trust. For Outboard could not be permitted to impair the bank's security interest by putting some of the assets covered by the agreement into a trust that the bank could not reach.

Bank of America has a second string to its bow: it argues that Illinois law, which the parties agree governs the interpretation of the trust agreement, will enforce a contractual antiassignment provision, such as the provision in the trust instrument that forbids assigning a security interest in the assets of the rabbi trust to creditors, against an assignee only if the provision states that the assignor has no power, and not merely no right, to assign. So, the argument continues, because the trust instrument does not say in so many words that any attempt by Outboard to create a security interest in the trust assets would be void, ineffectual, etc., the creation of such an interest is not prohibited although a party (including any third-party beneficiaries, which Outboard's general creditors may or may not be, see Exchange National Bank v. Harris, 126 Ill.App.3d 382, 81 Ill. Dec. 277, 466 N.E.2d 1079, 1084 (1984); Town & Country Bank v. James M. Canfield Contracting Co., 55 Ill.App.3d 91, 12 Ill. Dec. 826, 370 N.E.2d 630, 634-35 (1977)—we needn't decide), could sue for damages in the event of a breach of the provision.

Clauses in conveyances, or in other instruments contractual or otherwise that create property rights, that forbid the recipient of the property to sell it free and clear—or in legal jargon that create a "restraint on alienation"—are traditionally disfavored. Gale v. York Center Community Co-op., Inc., 21 Ill. 2d 86, 171 N.E.2d 30, 33 (1961); Avon-Avalon, Inc. v. Collins, 643 So. 2d 570, 574 (Ala.1994). Sometimes they are disfavored because they are thought to create monopoly, concentrate wealth, or cater to "the capricious whims of the conveyor." Gale v. York Center Community Co-op., Inc., supra, 171 N.E.2d at 33. But more often and more realistically it is because they can increase transaction costs by preventing subsequent purchasers or assignees from knowing what they are getting. Cf. Gregory S. Alexander, "The Dead Hand and the Law of Trusts in the Nineteenth Century," 37 Stan. L.Rev. 1189, 1258-60 (1985). A legal requirement that the restraint be express, recorded, or otherwise readily ascertainable by potential purchasers and assignees minimizes, and often eliminates, those additional costs, cf. Noblesville Redevelopment Comm'n v. Noblesville Limited Partnership, 674 N.E.2d 558, 562-63 (Ind. 1996); if the recipient's purchaser knows exactly what he is (not) getting, a refusal to enforce the restriction merely confers a windfall on him.

The requirement of express and readily ascertainable notice is satisfied here. When Bank of America made its credit agreement with Outboard, it knew, if it bothered to read the trust agreement along with the other documents that defined Outboard's assets, as it should have done and no doubt did do, that the security interest it was acquiring would not cover the assets (currently some $14 million) in the rabbi trust. Nothing would have been added to the trust agreement but empty verbiage had it said "and not only is Outboard forbidden to create a security interest in these assets in favor of any creditor, but if it tries to do so its action shall be null, void, and of no effect." Of course, if Illinois required those magic words, as many states still do, see Rumbin v. Utica Mutual Ins. Co., 254 Conn. 259, 757 A.2d 526, 530-33, 535 (2000), and cases cited there, to rebut the presumption of non-assignability, then Bank of America could argue persuasively that it had relied on their absence when it signed the security agreement. But Illinois does not require them. In re Nitz, 317 Ill.App.3d 119, 250 Ill. Dec. 632, 739 N.E.2d 93, 96, 101 (2000); Henderson v. Roadway Express, 308 Ill.App.3d 546, 242 Ill. Dec. 153, 720 N.E.2d 1108, 1113 (1999); see also CGU Life Ins. Co. v. Singer Asset Finance Co., 250 Ga. App. 516, 553 S.E.2d 8, 15 (2001).

Illinois's approach implements the modern view, expressed in Restatement (Second) of Contracts § 322(2) (1981), that an antiassignment provision in a contract is unenforceable against an assignee "unless a different intention is manifested." Magic words are not required: "Where there is a promise not to assign but no provision that an assignment is ineffective, the question whether breach of the promise discharges the obligor's duty depends on all the circumstances." Id., comment c. The circumstances here weigh heavily in favor of enforcing the antiassignment provision when we consider the alternative remedy that is all that a "magic words" state would allow in the absence of the magic words—a suit for damages for breach of the provision. If the credit agreement between Outboard and Bank of America violated it by creating a security interest in the trust assets, then the contract breaker, and therefore the defendant in such a suit, would be Outboard, which is to say the trustee, while the plaintiffs would be the general creditors—the trustee also. Enough said.

The Bank of America has one last argument, this one thoroughly frivolous—that the trustee under the trust agreement, who, remember, was in the event of Outboard's solvency to seek directions from a court concerning the disposition of the trust assets, was an "account debtor" of Outboard, that is, someone who owed Outboard money. UCC § 9-105(1)(a) (now superseded by UCC § 9-102(a)(3), unchanged however so far as bears on this case). An antiassignment clause is ineffective against an assignment of the debt of an account debtor. UCC § 9-318(4) (now, and again with immaterial changes, UCC § 9-406(d)(1)). Accounts and other simple written promises to pay are important collateral in modern commercial transactions, and their value as collateral is maximized by stripping them of encumbrances, such as an antiassignment clause unlikely to be noticed in the haste of transacting. The trust agreement was not that kind of instrument. And in any event the trustee owed Outboard nothing. The trustee was the debtor in a sense (an odd sense—one doesn't usually think of a trustee as the debtor of the trust's beneficiaries, though of course he holds its assets on their behalf) of the executives so long as Outboard was solvent, and after that he was the "debtor" in the same odd sense of Outboard's creditors. But he was never Outboard's "debtor."

Bank of America, a large, responsible, and well represented enterprise, should not have made the account-debtor argument. Nor should it have treated a district court decision (Lomas Mortgage U.S.A., Inc. v. W.E. O'Neil Construction Co., 812 F. Supp. 841 (N.D.Ill.1993)) as an authoritative statement of Illinois law. Not only has the Supreme Court instructed us not to give special weight to a district judge's interpretation of state law even if it is the state in which he sits, Salve Regina College v. Russell, 499 U.S. 225, 230-31, 111 S. Ct. 1217, 113 L. Ed. 2d 190 (1991); Beanstalk Group, Inc. v. AM General Corp., 283 F.3d 856, 863 (7th Cir.2002), but we have repeatedly reminded the bar that district court decisions cannot be treated as authoritative on issues of law. "The reasoning of district judges is of course entitled to respect, but the decision of a district judge cannot be a controlling precedent. E.g., Colby v. J.C. Penney Co., 811 F.2d 1119, 1124 (7th Cir.1987); Anderson v. Romero, 72 F.3d 518, 525 (7th Cir.1995). The law's coherence could not be maintained if district courts were deemed to make law for their circuit, let alone for the nation, since district courts do not have circuit-wide or nationwide jurisdiction." FutureSource LLC v. Reuters Ltd., 312 F.3d 281, 283 (7th Cir.2002).

AFFIRMED.