Kenneth P. Bidlack v. Wheelabrator Corporation, 993 F.2d 603

Kenneth P. Bidlack v. Wheelabrator Corporation

993 F.2d 603 | 7th Cir. | 1993

Lead Opinion

POSNER, Circuit Judge.

This is a class action on behalf of retired employees of the Wheelabrator Corporation. They claim that collective bargaining agreements between Wheelabrator and the union that represented employees at Wheelabrator’s plant in Mishawaka, Indiana conferred upon the plant’s retired employees lifetime rights to certain health benefits — hence rights that survived the expiration in 1988 of the last collective bargaining agreement between the company and the union applicable to the Mishawaka plant, which Wheelabrator closed that year. The plaintiffs base the suit primarily on section 301 of the Tafb-Hartley Act, 29 U.S.C. § 185, which creates a private right of action to enforce collective bargaining agreements. The briefs devote much space to the federal pension law (Employee Retirement Income Security Act, 29 U.S.C. §§ 1001 et seq.) as well, which does in fact furnish an alternative jurisdictional basis to the Tafb-Hartley Act for maintaining this suit in federal court, 29 U.S.C. § 1132(e)(1), but which has little bearing on the issues in contention. ERISA does not require the vesting of health or other “welfare” benefits, *605as it does pension benefits, Shaw v. Delta Air Lines, Inc., 463 U.S. 86, 91, 103 S.Ct. 2890, 2897, 77 L.Ed.2d 490 (1983); Metropolitan Life Ins. Co. v. Massachusetts, 471 U.S. 724, 732, 105 S.Ct. 2380, 2385, 85 L.Ed.2d 728 (1985); Ryan v. Chromalloy American Corp., 877 F.2d 598, 603 (7th Cir.1989), but equally it does not forbid their vesting by a written contract, id., which could be part of a collective bargaining agreement. International Union, United Auto Workers v. Yard-Man, Inc., 716 F.2d 1476, 1479-80 (6th Cir.1983); United Steelworkers v. Fort Pitt Steel Casting, 598 F.2d 1273 (3d Cir.1979); cf. In re White Farm Equipment Co., 788 F.2d 1186, 1193 (6th Cir.1986).

The district court granted summary judgment for the company; and dismissed the suit, because “an examination of the formal plan documents and the collective bargaining agreements reveals no intention on Wheelab-rator’s part to provide pre-1986 retirees the same level of benefits at no cost for life” and he believed that the requisite intention could not be proved by extrinsic evidence, that is, evidence outside the written contracts themselves. 763 F.Supp. 396 (N.D.Ind.1991). The plaintiffs appealed and the appeal was argued, as is customary, before a three-judge panel. Before the panel issued its decision, the full court, in accordance with Circuit Rule 40(f), decided to hear the case en banc in order to reexamine an earlier decision of the court, Senn v. United Dominion Industries, Inc., 951 F.2d 806 (7th Cir.1992). Judge Cummings, a member of the original panel, has recused himself from further consideration of the case because of a transaction by Wheelabrator that occurred after the en banc vote.

The main question briefed and argued by the parties is whether the absence from the collective bargaining agreements of any provision that explicitly vests the health benefits of retired employees defeats those employees’ claims even though some contractual language and a great deal of “extrinsic” evidence — evidence apart from the language of the agreements — suggest that the parties may have intended to confer vested rights on the retired employees, that is, rights that would outlast the expiration of the last collective bargaining agreement. If the answer is “yes” — the inexplicitness of the agreements is an insuperable obstacle to the suit — the district judge was right to grant summary judgment for the company. A second question, presented by Wheelabrator, is whether, even if the retired employees had vested rights, it" is plain from the terms of the collective bargaining agreements that their rights were no greater than those of the active employees. For Wheelabrator did not cut off the retired employees of the Mishawa-ka plant when the last collective bargaining agreement applicable to employees at that plant expired and the plant was closed. It is continuing to pay their health benefits at the same level and cost as the health benefits that it pays the active employees at its other plants.

Beginning with the collective bargaining agreement signed in 1965, every agreement between the company and the union in this ease up to and including the last one applicable to the Mishawaka plant, which was signed in 1985, contained a section which provided both that “those employees who have retired since September 22, 1959, will have the 'full cost of their Blue Cross-Blue Shield coverage paid by the Company after they attain sixty-five (65) years of age” and that they would be entitled to certain supplemental health benefits, also paid for by the company, “which, combined with Medicare, will provide a level of benefits equal to the Wheelabrator Blue Cross-Blue Shield plan for active and retired employees sixty-five (65) years of age or over” and which “shall be continued for the spouse after the death of the retiree.” We have quoted from the final collective bargaining agreement but the corresponding language of the earlier agreements is materially identical. There are, however, other differences between the agreements. In particular, the final agreement contains a provision not found in the earlier ones. It is the section following the one from which we quoted and it provides explicitly for the “vesting” of certain health benefits for employees who retire after the new agreement takes effect, that is, after January 1, 1986. The immediately preceding section, the one at issue in this case, governs the rights of employees who retired earlier; *606it is only employees in this group, that is, employees who retired on or before January 1, 1986, who are members of the plaintiff class.

Another contractual provision to which the parties have referred us is found in the contract between Wheelabrator and Blue Cross-Blue Shield for the purchase by the former of health insurance sold by the latter. The provision in question authorizes the company to change the level of benefits insured by Blue Cross-Blue Shield. We do not think it has much relevance to the main issue in this case, that of the duration of benefits. The union was not a party to the contract between Wheelabrator and its insurer, and anyway the issue is not the obligations of Blue Cross-Blue Shield to Wheelabrator but Wheelabrator’s obligations to its retired employees. Those obligations are defined by the collective bargaining agreements, not by the insurance contract. The latter however may be relevant to the dispute over the level of benefits to which the retired employees are entitled, assuming that they are entitled to any benefits at all after the last collective bargaining agreement in which these employees were mentioned expired.

There is evidence that, if believed (an important qualification, for there has been no trial), suggests that the language we quoted from the collective bargaining agreements was intended to confer lifetime benefits on employees who retired before 1986. For twenty years — from 1968 to 1988 — Wheelabrator issued a form to each newly retired employee which under the subheading “Medical Insurance (Blue Cross-Blue Shield of Indiana)” stated without qualification that “both you and your spouse will be covered for the remainder of your lives” at no cost. And a Wheelabrator executive who participated in the negotiation of several of the collective bargaining contracts testified that he believed that they were intended to provide retired employees with vested rights to health benefits. We need not discuss the other extrinsic evidence in support of, or for that matter against, the plaintiffs interpretation. The issue is not what the contract in fact means but whether the plaintiffs are entitled to a trial on what it means.

Wheelabrator points out that ordinarily when a contract expires, it — expires. It is at an end. The parties have no more rights or duties under it. Sometimes, however, a contract creates entitlements that outlast it. At argument the plaintiffs’ counsel gave the example of wages due under a contract of employment at will, a contract terminable at the whim of either party. Suppose the employer’s practice is to pay employees at the end of each week for the work they have done during the week. Jones, an employee at will, is fired at noon on Wednesday, having worked 20 hours that week. The contract is at an end as of noon that day, and yet, quite apart from any statutory entitlement that employees may have to be paid at the agreed rate for work actually done (National Metalcrafters v. McNeil, 784 F.2d 817 (7th Cir.1986)), the employee would have a compelling argument that the employer’s promise to pay for work actually done had survived the expiration of the contract. This is not the best example for the plaintiffs’ point, however, because an alternative conceptualization of employment at will treats it as a unilateral contract that is accepted by the employee’s working at the agreed wage. 1A Arthur Linton Corbin, Corbin on Contracts § 152 at pp. 13-14 (1963). So understood, a contract of employment at will does not end until the employee is paid. But there are plenty of better examples — examples of bilateral contracts that create obligations that outlive the term of the contract because the parties wanted them to do so. An employment contract that contains a post-employment restrictive covenant is one, Tower Oil & Technology Co. v. Buckley, 99 Ill.App.3d 637, 54 Ill.Dec. 843, 425 N.E.2d 1060 (1981); J.D. Marshall International, Inc. v. Fradkin, 87 Ill.App.3d 118, 42 Ill.Dec. 509, 409 N.E.2d 4 (1980), and there are others. Litton Financial Printing Division v. NLRB, — U.S. -, -, 111 S.Ct. 2215, 2226, 115 L.Ed.2d 177 (1991); Ryan v. Chromalloy American Corp., supra; In re White Farm Equipment Co., supra, 788 F.2d at 1193. No doubt a court should cast a cold eye on contentions that a contract with a fixed term actually created a perpetual obligation, William B. Tanner Co. v. Sparta-Tomah Broadcasting Co., 716 F.2d 1155, 1159 (7th Cir.1983), and *607should, therefore, as Senn and many other cases hold (notably Litton), presume that a collective bargaining agreement ceases to obligate the employer when the agreement’s term (invariably three years) is up. But it is not an irrebuttable presumption. “Rights which accrued or vested under the [collective bargaining] agreement will, as a general rule, survive termination of the agreement.” Litton Financial Printing Division v. NLRB, supra, — U.S. at -, 111 S.Ct. at 2226. The question is what it takes to rebut the presumption. We add that the obligation for which the plaintiffs contend in this suit is not perpetual, because retired people and their widows (or widowers) do not live forever.

We reject the extreme positions on what it takes to rebut the presumption that all contractual obligations cease on the expiration date stated in the contract. The first is that the contract must either use the word “vest” or must state unequivocally that it is creating rights that will not expire when the contract expires. The second is that testimony about the parties’ intentions can substitute for contractual language indicative of intent and therefore create rights and duties that last beyond the expiration of the contract. The first approach could be defended only as a means of making life simpler for courts by creating a form that parties must use to create enforceable rights and obligations; and while we naturally are sympathetic to the institutional concerns behind such a suggestion, we do not think that a court should refuse to enforce a contract merely because the parties have failed to use a prescribed formula. It is one thing for a court to lay down default rules to solve contractual disputes when the parties’ intentions cannot be determined. That is a wholly appropriate judicial function. It is another thing for us to say to the contract parties, we can see what you’re driving at but as you have not used our preferred form of words we shall not enforce your contract. Cf. Johnson v. United States, 163 Fed. 30, 32 (1st Cir.1908) (Holmes, J.). Of course, it was the courts that, long ago, invented the requirement of consideration, which in one aspect is a form with which parties are required to comply as a condition of seeking the enforcement of their “contract.” But courts should be cautious about adding formal hoops that contract parties must jump through, and anyway we do not understand Wheelabrator to be urging us to do that here.

The second extreme, that of allowing parties to substitute oral testimony for contractual language, errs by depriving parties of the protection of a written contract. For we are supposing a case in which the written contract is silent on a subject, and a party asks the court to interpolate a clause on the basis of extrinsic evidence showing that the parties would have included the clause if they had thought about the matter. Courts do sometimes interpolate contract clauses. Restatement (Second) of Contracts § 204 (1979). A famous example is Judge Cardozo’s interpolation of a best-efforts clause in a contract providing for an exclusive dealership. Wood v. Duff-Gordon, 222 N.Y. 88, 118 N.E. 214 (1917). But he based this move on the structure of the contract rather than on self-serving testimony by a party. By granting an exclusive license to market her dress designs, Lady Duff-Gordon had delivered herself into her licensee’s power, for unless he had a contractual obligation to market her designs he could threaten to sit on the license and do nothing unless she increased his compensation. It was unlikely that the parties had intended so one-sided a deal. The Supreme Court in Litton referred approvingly to a parallel interpolation — that the duty to arbitrate a dispute under a contract providing for arbitration survives the expiration of the contract. -U.S. at - n. 3, 111 S.Ct. at 2226 n. 3. (And it thought this precept applicable to collective bargaining agreements. Id. at -, 111 S.Ct. at 2226.) The utility of an arbitration clause would be greatly impaired if the duty to arbitrate disputes arising under the contract expired with the contract, since such disputes might easily remain pending on the date of expiration. If the contract were terminable by either side on short notice, as is common, a party who saw a dispute looming could escape having to arbitrate it simply by terminating the contract.

Since there is no similar structural necessity for a collective bargaining agreement to include an undertaking .by the em*608ployer to pay lifetime medical benefits to retired employees, the use of extrinsic evidence to create such obligations nowhere alluded to in the contract would unjustifiably deprive the parties of the limitation of liabilities that is implicit in the negotiation of a written contract having a definite expiration date. Subject only to the limited protection against unforeseeable contractual obligations that is conferred by the doctrine of impossibility, a party might find itself saddled with obligations for the next twenty or thirty years (or even more, in the case of a surviving spouse’s benefits) even though it had reasonably believed that all its obligations would end in three years, when the contract expired by its own terms. Although extrinsic evidence is admissible to show that a written contract which looks clear is actually ambiguous, perhaps because the parties were using words in a special sense, FDIC v. W.R. Grace & Co., 877 F.2d 614, 620-21 (7th Cir.1989), there must be either contractual language on which to hang the label of ambiguous or some yawning void, as in the Duff-Gordon case, that cries out for an implied term. Extrinsic evidence should not be used to add terms to a contract that is plausibly complete without them. Calder v. Camp Grove State Bank, 892 F.2d 629, 632 (7th Cir.1990).

If therefore the collective bargaining agreements in this case were completely silent on the duration of health benefits for retired employees, then since (unlike Duff-Gordon and Litton) nothing in the structure of the agreements required that the duration be perpetual, we would not allow extrinsic evidence to show that those employees have a perpetual entitlement. But the agreements are not silent on the issue; they are merely vague. They say that once retired employees reach the age of 65 the company will pick up the full tab for their health insurance and that when they die their spouses will continue to receive supplemental health benefits, again at the company’s cost. This could be thought a promise to retired employees that they and their spouses will be covered for the rest of their lives. For the provision does not say “when they die or the collective bargaining agreement expires, whichever occurs first,” but simply when they die. The corresponding language in Senn was weaker: it was that the company “will continue for retired employees” a specified premium contribution. And the final agreement between the company and the union in that case, which could be thought declaratory of earlier understandings, said that coverage “shall be continued during the term of this Agreement.” 951 F.2d at 810, 815 (emphasis added). In addition, each agreement in the series contained an integration clause, declaring the agreement to be the exclusive source of the parties’ rights and duties, and although the parol evidence rule, which enforces integration clauses by barring evidence of side agreements, does not bar the use of extrinsic evidence to clarify the meaning of an ambiguous text, FDIC v. W.R. Grace & Co., supra, 877 F.2d at 620, the presence of such a clause is some indication of the parties’ desire to limit a free-ranging judicial discretion to interpolate terms.

We do not suggest that the interpretation of the relevant provision of the collective bargaining agreement in the present case as creating lifetime entitlements to company-paid health benefits is inevitable. The provision in question could mean just that retired employees are covered for the three-year term of the collective bargaining agreement. This inference is strengthened by the explicit use of the term “vesting” in the next section of the 1985 collective bargaining agreement but is weakened by the fact that the preceding section, the one in issue here, was carried over essentially verbatim from the earlier collective bargaining agreements. The vesting provision has a separate provenance, and we are given no reason to suppose that the two sections were actually compared in the drafting process, though a trial might reveal that they were. And if the guarantee of company-paid health insurance is limited to the three years in which the collective bargaining agreement is in force, the ringing promise of spousal benefits has rather a hollow sound, as it is easy to imagine a case in which an employee who retired years later dies in 1987 — and at the stroke of midnight on December 31 of that year his wife’s supplemental benefits would terminate, if the company’s interpretation is adopted. Yet, *609continuing the argumentative ping-pong game, we mustn’t leap to. the conclusion that the company’s position makes the benefits to retired employees “illusory.” Three years, or even three days, are better than none; and a company is naturally reluctant to saddle itself with indefinite future obligations— though we shouldn’t rush to the opposite extreme, and suppose that Wheelabrator will be crushed if it must pay these benefits for life. Their sum is not trivial, but most of the health costs of the retired are picked up by Medicare rather than by Blue Cross-Blue Shield.

The cynic might' argue that the union would never have pressed for lifetime guarantees for retired employees — they do not pay union dues, vote in union elections, or vote in representation elections, for they are no longer members of the bargaining unit. Allied Chemical & Alkali Workers v. Pittsburgh Plate Glass Co., 404 U.S. 157, 92 S.Ct. 383, 30 L.Ed.2d 341 (1971). The counterargument is that, knowing this, current employees who may retire in the next three years want their collective bargaining agreement to vest health benefits. United Auto Workers v. Yard-Man, Inc., supra, 716 F.2d at 1482. Most, perhaps all, the class members were active employees when a collective bargaining agreement containing the provision for retirees’ health benefits was signed.

We could argue back and forth till we were blue in the face over the proper interpretation of the language relating to the retirees’ health benefits without reaching a confident conclusion, if all we had to go on was the written contract itself. The contract, even when its logic and its other provisions as well as just the provision in issue are considered, is inconclusive on the question whether it confers an entitlement to health benefits that outlasts the contract’s expiration date. A completely intractable issue of contract interpretation can bé resolved only by the application of some default rule — a burden of persuasion, a clear-meaning rule, a presumption based on the authorship of the contract. But the time to throw up one’s hands and apply such a rule is after extrinsic evidence has been considered. For until then, we do not know whether we have an intractable interpretive issue or merely an issue that cannot be resolved without testimony or other evidence besides the language and logic (as in Wood v. Duff-Gordon) of the contract.

Only a posture, not easy to reconcile with the Seventh Amendment, of extreme mistrust of juries would entitle us to pretermit a factual inquiry and apply an interpretive canon or other tie-breaker before we know that the sides are actually tied. The contract in this case is ambiguous and both sides are poised to present testimony and documents that they claim will disambiguate it. We think they should be allowed to do so despite the alarm bells rung in the amicus brief submitted to us by an employers’ association which argues that without a strong presumption against vesting, companies will be visited with crushing liabilities (but no numbers are provided), unions will be discouraged from negotiating over health benefits, the containment of health costs will be impeded, and already retired employees will receive windfalls at the expense of the newly and the not-yet retired. That is to assign an awful lot of weight to a rule of interpretation that parties are free to change by contract. Employers adamant against assuming perpetual obligations can eliminate all doubt by insisting on a clause that makes any entitlement to health benefits granted by the agreement expire on the date the agreement expires. Employers don’t even have to bargain over health benefits of retired employees. Allied Chemical & Alkali Workers v. Pittsburgh Plate Glass Co., supra. They certainly don’t have to grant such benefits in perpetuo. If they did so in the past, not anticipating the recent rise in health costs, they should not expect the courts to bail them out by undoing the contractually determined allocation of risk on the question. Courts do not sit to relieve contract parties of their improvident commitments, except within the limited dispensation conferred by the doctrine of impossibility, not here invoked. Contracting parties who want to be spared the uncertainties of trial by jury have only themselves to blame if by failing to specify the limits of their undertakings they open the door to extrinsic evidence of contractual meaning.

We are not done, because there is the company’s further argument that even if the *610retired employees have vested rights, the rights are limited to those enjoyed by active employees. Even if this is true, we disagree that it provides an alternative ground for affirming the district court’s grant of summary judgment. The plaintiffs claim a lifetime right to benefits — ignore for a moment what level of benefits they might be entitled to. The fact that the company, without acknowledging their entitlement, is continuing to pay them the level of benefits to which (if the company’s subsidiary argument is correct) they would be entitled if they had an entitlement cannot be considered a full satisfaction of their claim. Unless their right has vested, the company can cut them off tomorrow. They are entitled to a declaration of their rights. The dispute over those rights is a real one. If they have no contractual guaranty of benefits, they had better start shopping around for other medical insurance.

The language that we quoted earlier from the last collective bargaining agreement seemed to key retired employees’ benefits to those of active employees, as the company argues. However, the 1980 agreement had provided for the first time that the health benefits to which employees were entitled would include “Dental, Vision and Hearing for all retirees” and that “the Drug Program co-payment for retirees will be reduced to $2.00.” These two benefits were removed in the last agreement for active as well as retired employees, permissibly so for the latter if the only right that had vested was a right to equality in health benefits with the active employees. The fact that the agreement specified retirees’ rights separately from those of active employees, extending to them benefits which previously had been extended to the active employees alone, suggests, though hardly demonstrates, that the collective bargaining agreements did not make the two groups march in lock step. The argument contra is that the company was hardly likely to agree to a ratchet, whereby anytime it expanded the benefits for active employees it established a plateau under the retired employees. A vested right to benefits equal to and hence varying with those of the active employees would give the retirees a good deal of protection, since health benefits are an important fringe benefit.

There is additional evidence that the collective bargaining agreements may not have fixed the same level of benefits for active and retired employees. The agreements define “employees” as employees of the Mishawaka plant, but, that plant having been closed, the company is paying the retirees the level of benefits of its active employees in other plants. Suppose it provided different levels of benefit to active employees in different plants. Which level would provide the benchmark for the retired employees? And in 1985 active employees were required to share the cost of the premium for certain benefits while the retired employees continued to receive those benefits entirely at Wheelabrator’s expense. So we cannot be certain that the agreements were intended to give the retirees no greater (net) benefits than active employees, and even if we were certain, there would still be the unresolved question whether the retired employees were receiving those benefits as a matter of grace or of right.

REVERSED AND REMANDED.

Concurrence Opinion

CUDAHY, Circuit Judge,

with whom RIPPLE and ROVNER, Circuit Judges, join, concurring in the judgment.

The issue here is the appropriate default rule for determining whether collective!y bargained retiree health benefits have vested. The default rule imposed in Senn v. United Dominion Indus., Inc., 951 F.2d 806 (7th Cir.1992), 'provides, as I understand it, that retiree benefits vest only if the language of the collective bargaining agreement states explicitly that the benefits have vested; resort to extrinsic evidence is improper even if the agreement is ambiguous with respect to vesting. One of the stated reasons for taking the present case en banc was to consider whether Senn should be overruled. The lead opinion has elected to distinguish Senn (on somewhat unconvincing grounds) rather than to overrule it. Although both the lead opinion and dissent pay it passing respects, Senn and its default rule now do seem to me to be a dead letter. So far as it goes, this is a constructive outcome.

I agree in large part with the lead opinion. I write separately only to suggest that, while *611its approach takes us in the right direction, it has stopped short of what could be a better solution. As an initial matter, the lead opinion, in reviewing the possible default rules for determining whether retiree benefits have vested, fails to note the full spectrum of available rules. It finds, ante at 607-08, only three possible default rules for determining when retiree benefits vest:

1. “Strong" No-Vest Rule. This rule, specifically adopted in Senn, provides that retiree health benefits in collective bargaining agreements do not vest unless the language in the agreement explicitly provides for vesting. If the agreement does not employ the words “vest” or “lifetime” or other clear vesting language, then the presumption against vesting governs, the document is deemed unambiguous and resort to extrinsic evidence is barred.

2. “Weak” No-Vest Rule. This rule holds that if the collective bargaining agreement is ambiguous with respect to vesting, resort to extrinsic evidence is proper. However, if a preponderance of the evidence does not show that the parties intended the benefits to vest, the presumption is that they did not.

3. Parol Substitution Rule. This rule would allow use of extrinsic evidence to determine the intent of the parties as to vesting even when there is no ambiguity in the collective bargaining agreement, as when the agreement is silent. Presumably, in the absence of extrinsic proof of vesting, the court would infer that the parties did not intend to have the benefits vest.

The lead opinion specifically adopts option 2, the weak no-vest rule. First, it rejects the strong no-vest rule (the Senn rule) because its rigidity may frustrate the actual intent of the parties, and it is in tension with the Seventh Amendment right to a jury trial. Ante at 607, 609. The lead opinion likewise rejects option 3 because that rule would resort to extrinsic evidence even when the agreement is silent about retiree benefits, thus “depriving parties of the protection of a written contract.” Id. at 607. Having rejected both of these “extreme” options, the lead opinion falls back on the one remaining option, the weak no-vest rule, and finds that it comports with settled principles of contract law. What the lead opinion does not consider, however, are the corresponding default rules presuming that the benefits do vest. The three parallel rules are:

4.“Strong” Vest Rule. This rule would presume that the parties intended the benefits to vest unless they explicitly stated otherwise. If there is an ambiguity, the presumption takes effect and there can be no recourse to extrinsic evidence.

6. “Weak” Vest Rule. This rule would presume that benefits vest unless there is evidence of an agreement to the contrary. If the collective bargaining agreement is ambiguous, then consideration of extrinsic evidence is permitted to determine whether the parties did not intend the benefits to vest.

6. Parol Substitution Rule. Like its counterpart, it would resort to extrinsic evidence even when the collective bargaining agreement is silent (or otherwise unambiguous) about the vesting of benefits. But in the absence of proof that the parties did not intend the benefits to vest, it would presume the benefits vested.

I agree with the lead opinion that we can dispose summarily of the parol substitution rules, options 3 and 6. Both rules would denigrate the function of a written contract, in derogation of the traditional parol evidence rule. Choosing among the four remaining default rules, however, is more difficult. As both the lead opinion and dissent recognize, options 1 and 4 (the strong rules) have their advantages. The requirement of clarity serves a worthy purpose. The strong rules would motivate parties to reach an explicit agreement and save courts from the difficult task of interpreting ex post what the parties intended. Options 2 and 6 (the weak rules), on the other hand, would likewise tell the parties what the law would presume in the absence of an agreement, but these rules would not foreclose judicial inquiry into extrinsic evidence to determine the actual intent of the parties. Of course, depending on *612which view — the lead’s or the dissent’s — one supports, access to extrinsic evidence is the greatest strength or weakness of options 2 and 5.

We should recognize initially that, when those affected by a chosen default rule can easily bargain around it to agree to a mutually beneficial course, the rule choice will generally make little difference to the parties’ actual agreement. Ante at 609; Post at 619; Ronald Coase, The Problem of Social Cost, 3 J.Law & Econ. 1 (1960). When bargaining is easy, courts selecting a default rule may properly look past those parties affected by the rule to other advantages or disadvantages a particular rule may entail, such as the costliness of its administration. In theory, the strong rules have some intuitive appeal because it should be easy for companies and unions to bargain around any presumption we might impose; again theoretically, it should make little difference to the parties what rule would apply should they not address the vesting question. The parties are sophisticated and the cost of bargaining is slight. On the other hand, unions and companies might sometimes prefer ambiguity: the retirees can feel secure, while the bargainers leave some unspoken flexibility for future bargains. In any event, the strong rules mean no more ambiguities, no more extrinsic evidence and no more prolonged litigation.

But this logic ignores one crucial factor: the parties before us here, and many who have yet to come before us, cannot bargain around our default rule. Our default rule comes too late for them' — the bargain has been struck When the parties or the situation are such that those affected by the rule cannot bargain around it, the choice of default rule is different; for the rule may, in effect, supply terms to which the parties are bound but to which they have not explicitly agreed. In these circumstances, courts should try to select the default rule that best reflects the course the parties would have taken had they bargained freely. Frank H. Easterbrook & Daniel R. Fischel, The Economic Structure of Corporate Law 15 (1991); Richard Posner, Economic Analysis of Law 372 (3d ed. 1986). But because the parties did, in fact, bargain at arms-length, what they would have negotiated is precisely what they did negotiate. Thus, when applying a default rule retrospectively to transactions negotiated at arms-length, courts should select the rule that best reflects what the parties did negotiate and that enables the negotiated terms to take effect.

Under those circumstances, we encounter the major flaw of the strong rules. The strong rules presume that parties to all collective bargaining agreements were negotiating under the same assumptions, and foreclose evidence of what the parties actually negotiated. Realistically, however, certain parties to a series of collective bargaining agreements, in the absence of a judicially imposed default rule, may have their own default rule; and their default rule may not be the same as that adopted by various other parties to collective bargaining agreements. As the dissent correctly notes, the parties to a collective bargaining agreement have an ongoing relationship, in which the parties, often represented by the same persons, meet every three years to negotiate over terms previously negotiated. The understandings and practices of the parties develop a sort of common law for interpreting their agreements.1 Yet the imposition of a strong rule would bind the parties to certain default terms even when there is evidence that they were working under exactly the opposite assumptions. A strong rule, consequently, may frustrate what the parties actually intended.2

*613The weak rules, on the other hand, are more flexible in that they are better able to respect the agreement of the parties. Of course, they do provide clear presumptions around which the parties can effectively bargain. And, as noted, prospective application of a weak rule should not be too different from prospective use of a strong one. But retrospective application of a weak default rule, unlike a strong rule, enables parties to correct the imposition of default terms by using extrinsic evidence of actual intent. To be sure, with flexibility goes ambiguity and the greater likelihood of litigation. But this is not too high a price to pay when something as crucial as retiree health benefits is at stake. Getting at the real intention of the parties seems well worth the costs of making the determination. See, e.g., Air Line Stewards and Stewardesses Ass’n, Local 550, etc. v. American Airlines, Inc., 763 F.2d 875, 877-78 (7th Cir.1985) (“the primary object in construing a contract is to give effect to the intention of the parties.”), cert. denied, 474 U.S. 1059, 106 S.Ct. 802, 88 L.Ed.2d 778 (1986).

Hence, I think the only real choice is between the two weak rules, options 2 and 5, and here I must differ with the lead opinion. For I may see a different social reality within which these agreements were entered.3 Before about 1980, I seriously doubt that it occurred to many employers to grant retiree health benefits on anything less than a lifetime basis.4 The overwhelmingly prevalent trend of labor contracts was to continue or improve retiree benefits from contract to contract. It was only in the eighties, with spiraling medical costs, heightened foreign competition, epidemic corporate takeovers and the declining bargaining power of labor, that thought was first given to reducing retiree benefits from contract to contract or even (though this seems more implausible) to eliminating such benefits entirely. I think that, at least before the eighties were in full swing, prevailing conditions suggested a presumption among unions and management alike that retiree health benefits vested unless there was agreement to the contrary. Hence, I would lean toward option 5 and the application of the weak vest rule — -the position adopted by the Sixth Circuit in the seminal case, International Union, United Auto, etc. v. Yard-Man, Inc., 716 F.2d 1476 (6th Cir.1983), cert. denied, 465 U.S. 1007, 104 S.Ct. 1002, 79 L.Ed.2d 234 (1984). See also Keffer v. H.K. Porter Co., 872 F.2d 60, 64 (4th Cir.1989) (citing Yard-Man as “consistent with a more far-reaching understanding of the context in which retiree benefits arise.”); Local Union No. 150-A United Food & Commercial Workers, etc. v. Dubuque Packing Co., 756 F.2d 66, 69-70 (8th Cir.1985) (inferring that benefits vested because “[t]here is simply.no evidence that the Company and the Union did not intend to *614vest the right to benefits in the retirees.”). But, looking at things as they are rather than as they might be, the direction indicated by the lead opinion in adopting option 2 and the weak no-vest rule is a significant improvement over the apparently extreme course suggested by Senn.

. [T]he parties occasionally have understandings or expectations that were so fundamental that they did not need to negotiate about those expectations. When the court "implies a promise” or holds that “good faith” requires a party not to violate those expectations, it is recognizing that sometimes silence says more tha[n] words....

Arthur Corbin, Corbin on Contracts § 570 (Colin K. Kaufman ed., Supp.1984).

. The dissent suggests that Litton Fin. Printing v. NLRB, - U.S. -, 111 S.Ct. 2215, 115 L.Ed.2d 177 (1991), mandates the imposition of the strong no-vest rule. Litton, however, held that a collective bargaining agreement's arbitration clause did not terminate on the expiration of the agreement, but continued after the agreement's termination to apply to disputes "arising under the contract.” Id. at -, 111 S.Ct. at 2225. Further, the language of the Court in Litton clearly recognizes the existence of vested *613benefits quite independent of "explicit terms.” What the Court said bears repeating:

“Rights which accrued or vested under the agreement will, as a general rule, survive termination of the agreement. And of course, if a collective bargaining agreement provides in explicit terms that certain benefits continue after the agreement’s expiration, disputes as to such continuing benefits may be found to arise under the agreement, and so become subject to the contract's arbitration provisions.”

Id. at -, 111 S.Ct. at 2226. The Court does not state that a contract right vests only if the agreement so provides in explicit terms. Rather, vesting is a matter of contract interpretation, which seeks to determine the intent of the parties.

. [Wjhat parties do "agree to,” particularly when they are silent, must be interpreted in light of [those] assumptions about reality that they share. These assumptions are captured in the prevailing conventions in the relevant community of discourse. Randy E. Barnett, The Sound of Silence: Default Rules and Contractual Consent, 78 Va.L.Rev. 821, 902 (1992).

. Nothing in Chemical Workers v. Pittsburgh Plate Glass Co., 404 U.S. 157, 92 S.Ct. 383, 30 L.Ed.2d 341 (1971), suggests otherwise. In that case, a collective bargaining agreement specified that the company could discontinue part of its contributions for retiree health insurance if Medicare was enacted by Congress. Upon the enactment of Medicare, the company attempted to modify the retiree health plan. Instead of reducing or eliminating the negotiated retiree benefits, the company sought simply to substitute Medicare coverage for the negotiated health plan "by offering to pay the supplemental Medicare premium if the employee would withdraw from the negotiated plan.” Id. at 162, 92 S.Ct. at 389. No retiree would have his benefits taken away, and no retiree would have his negotiated benefits substituted for Medicare without his consent.

Dissenting Opinion

EASTERBROOK, Circuit Judge,

with whom BAUER, Chief Judge, and COFFEY and MANION, Circuit Judges, join, dissenting.

For more than two decades Wheelabrator and the unions representing its employees in Mishawaka, Indiana, maintained collective bargaining agreements providing retired workers with health benefits. Starting in 1980 these benefits matched those of active workers. Whatever current workers received, the retirees received. Even though Wheelabrator has closed the Mishawaka plant and the last agreement has expired, retirees continue to receive the health benefits available to active workers at the firm’s other plants, at levels to which other unions have assented. Wheelabrator introduced a copayment system under which covered persons pay 20% of the cost of their care, with a maximum copayment of $1,200 per person or $2,400 per family. The maximum benefit rose from $150,000 to $500,000. The firm eliminated dental, vision, and hearing benefits. These changes help some retirees (those facing major medical expenses) while requiring others to pay more for routine care. Copayments induce employees and retirees alike to shop for lower prices and produce savings that permit the higher maximum benefit.

Mishawaka’s retirees want more. They demand free health care for life; they welcome the higher maximum benefit but want to escape the copayments and receive reimbursement for vision, dental, and hearing expenses. Plaintiffs think it dandy that their benefits rose with active workers’ and contend that they have a contractual ratchet: their benefits cannot follow active workers’ into decline — if the restructuring of the benefit package can be called a “decline.”

Until the 1985 contract, none contained language implying that the level of benefits vests on retirement or at any other time— and that contract established a special rule for persons who retired at age 65 in 1986 or later. Wheelabrator has not instituted co-payments for these retirees. Nothing in the contracts provides that the scope of coverage for other retirees will never change. Their schedule of benefits, which the collective bargaining agreement establishes only by reference to the insurance, is all today’s case is about. Duration is not in dispute. Yet the lead opinion devotes almost all of its attention to duration, relegating the level of benefits to an afterthought, and the concurring opinion discusses duration exclusively, disdaining this case in order to address questions that might be important in other cases.

The result is a disposition that is largely advisory. Litigation began after Wheelabrator instituted a system of copayments. Retirees contended that they possess a contractual entitlement to unreduced benefits for life. That the retirees stressed “for life” rather than “unreduced” is no surprise: all of their parol evidence is about “lifetime” benefits, so they sought to divert the court’s attention from the weak link in their claim. Wheelabrator denied that the contracts promise any particular level of benefits and, adopting the common belt-and-suspenders strategy, added that any guarantees expired in 1988. Like the retirees, Wheelabrator would like victory on a rationale that produces the most favorable implications for tomorrow’s case. The parties’ understandable druthers do not authorize us to address a hypothetical dispute, however. The only controversy between these parties is whether the retirees are entitled to medical benefits exceeding those of active workers. What would happen if Wheelabrator were to cut off the retirees, which it has never threatened to do, is a question for another day. Los Angeles v. Lyons, 461 U.S. 95, 103 S.Ct. 1660, 75 L.Ed.2d 675 (1983); Foster v. Center Township, 798 F.2d 237 (7th Cir.1986). If Wheelabrator had sent its retirees letters saying: “We believe that we need not continue your benefits, but we have no plans to terminate them”, the retirees would not have a justicia-ble controversy with their old employer about the meaning of the collective bargaining agreements. The retirees have not even *615received such letters. Parties cannot require (or authorize) adjudication of an issue that could arise in the future just by demanding a declaratory judgment. The retirees’ only live grievance concerns the details of an ongoing medical plan.

What do the collective bargaining agreements say about the level of benefits to which the retirees are entitled? In a word: Nothing. The first agreement (from 1965) provides that “employees who have retired since September 22, 1959 will have the full cost of their Blue Cross-Blue Shield coverage paid by the Company after they attain 65 years of age (subject to the provisions of the supplemental insurance plan listed in Paragraph (b) below). Employees retiring prior to age 65 years will be required to pay one-half of their Blue Cross-Blue Shield plan until such time as they reach age 65.” 1 Similar provisions appear in later contracts, which add other health benefits.2 The agreements say what portion of the cost retirees will pay, but they do not specify what the “Blue Cross-Blue Shield plan” entails. That was left to the agreement Wheelabrator struck with its carrier. Only the insurance policies specify the extent of coverage, and each policy reserves Wheelabrator’s right to alter things. These contracts use language tracking the agreements in Boyer v. Douglas Components Corp., 986 F.2d 999 (6th Cir.1993), and Gill v. Moco Thermal Industries, Inc., 981 F.2d 858 (6th Cir.1992). Boyer and Gill hold that the insurance policy is the “plan” under ERISA and outranks oral promises. So the shouting should be over. Retirees receive the same “Blue Cross-Blue Shield plan” as active employees, and by virtue of express language Wheelabrator controls the scope of the plan’s coverage.3

Plaintiffs’ entire submission rests on a confusion between free “coverage” and free “care.” Medical insurance can be free without being comprehensive. To see the difference, consider a common medical plan: full payment for the costs of hospitalization, but a maximum of 365 days of continuous care. Starting on day 366 the sick or injured person pays the tab himself. An insurance policy with these benefits might cost $2,000 per year in the market. Providing such a policy to retirees at no cost gives them a free “Blue Cross-Blue Shield plan” but not free medical care. In 1965, when the contract first provided some health coverage after retirement, the duration was limited. Major medical coverage did not come until 1974, and even then there was a cap on total benefits payable for a single illness. Just as a medical plan may be limited in scope or duration, so it may be limited in the percentage of costs paid. Consider an alternative plan, which pays 80% of all hospital costs without a maximum number of days. An insurer might charge $2,000 per year for this policy, too; providing such insurance to employees would give them a free “Blue Cross-Blue Shield plan” without giving them free medical care. And this is what happened. Wheelabrator’s current medical package includes .benefits of greater duration and maximum payment than the 1965 (or, for that matter, 1985) plan but requires copayments that encourage shopping among providers and discourage demands for unnecessary medical care. The labor agreements do not promise the retirees free medical care, as opposed to medical coverage commensurate with active workers.

The only thing in any of the agreements that could be understood as a promise of particular care appears in the 1980 contract: “Effective 7/1/81 these benefits include Den*616tal, Vision and Hearing for all retirees.” This sentence appears immediately after the statement that Wheelabrator “will pay the full cost of .Blue Cross/Blue Shield” for retirees who meet age and years-of-service requirements. Active workers already enjoyed dental, vision, and hearing benefits. The 1980 contract brought the retirees to parity. Read favorably to Wheelabrator, this sentence means that as of July 1981 the Blue Cross/Blue Shield plan does include particular benefits; it does not promise continuation of these benefits. Read favorably to the retirees, the sentence means that the Blue Cross/Blue Shield plan must include particular benefits. As the dental, vision, and hearing benefits were abolished in 1989 (for active workers as well as retirees), this sentence provides the plaintiffs with their best shot.4

Curiously, the lead opinion makes little of this sentence, and the concurring opinion does not mention it. It comes up at page 610 of the lead opinion. But the court then equates all questions about level and duration, even though the sentence in question, if read most favorably to the retirees, puts dental, vision, and hearing benefits on a plane different from other categories of benefits. By my colleagues’ lights, the question for trial ought to be whether the agreement assured these three kinds of benefits to persons who remained on the payroll after July 1,1981. (Those who retired earlier furnished no consideration for any promise in the 1980 contract.) Debate about the copayment system should end forthwith. There is no textual ambiguity on that subject — at least none the majority deigns to mention. And as I have stressed, there is no case or controversy about whether “the retired employees were receiving those benefits as a matter of grace or of right.” Op. at 610. Advisory opinions rendered by juries are no more palatable than those rendered by judges.

I do not think that even the sentence “Effective 7/1/81 these benefits include Dental, Vision and Hearing for all retirees” creates a jury question. I start from the premise that when a labor agreement is not ambiguous, the court should declare its meaning without ado. Ooley v. Schwitzer Division, Household Manufacturing, Inc., 961 F,2d 1293, 1298-99 (7th Cir.1992); Senn v. United Dominion Industries, Inc., 951 F.2d 806, 814-16 (7th Cir.1992). A majority of this court (the three judges subscribing to the lead opinion plus the four joining this one) reaffirms Senn — not only its holding that extrinsic evidence cannot create an ambiguity in otherwise clear documents, but also its holding that health and welfare benefits promised in collective bargaining agreements presumptively expire with those agreements. See also Merk v. Jevuel Companies, Inc., 848 F.2d 761, 763 (7th Cir.1988). These principles should lead us to affirm. Put in context, the single sentence in a single agreement means no more than the surrounding sentences: as of 1981, the retirees at last get the same plan as the active workers.

Pensions vest by law; ERISA establishes elaborate schedules and tables. 29 U.S.C. § 1053. Health and other welfare benefits are left to contract. That leaves in place the venerable rule that parties may agree to as much or as little as they please. Wheelabrator and its unions did not agree on a perpetual schedule of health benefits. They incorporated an insurance contract by reference. See Boyer and Gill. Ever since, Wheelabrator has furnished its retirees at Mishawaka with the same benefits to which active workers at other plants have assented.

Is there any reason to read into the agreements a ratchet that their language does not express? Labor law has modified contractual freedom in light of the statutory duty to bargain. An employer may not walk away from workers the same way a supplier of oil may walk away from customers. Instead the employer must bargain collectively with its employees. Once an existing agreement has ended and the bargaining obligation has been fulfilled, however, the employer may put its proposals into force, modifying or abandoning the terms that formerly applied. NLRB *617v. Katz, 369 U.S. 736, 82 S.Ct. 1107, 8 L.Ed.2d 230 (1962); Lapham-Hickey Steel Corp. v. NLRB, 904 F.2d 1180, 1185 (7th Cir.1990). Parties may agree on immutable benefits, which continue by force of contract. But the presumption is and always has been that benefits mentioned in a collective bargaining agreement do not vest. “As with the obligation to make pension contributions in [Laborers Health and Welfare Trust Fund v.] Advanced Lightweight Concrete Co.[, 484 U.S. 539, 108 S.Ct. 830, 98 L.Ed.2d 936 (1988) ], other contractual obligations will cease, in the ordinary course, upon termination of the bargaining agreement. Exceptions are determined by contract interpretation. Rights which accrued or vested under the agreement will, as a general rule, survive termination of the agreement. And of course, if a collective bargaining agreement provides in explicit terms that certain benefits continue after the agreement’s expiration, disputes as to such continuing benefits may be found to arise under the agreement”. Litton Financial Printing Division v. NLRB, - U.S. -, -, 111 S.Ct. 2215, 2226, 115 L.Ed.2d 177 (1991) (emphasis added). In Litton the Court looked for the “explicit terms” requiring the employer to honor seniority after the terminal date of the contract and, finding none, rebuffed the union’s claim. It did not search outside the terms of the agreement.

Provisions in a collective bargaining agreement establishing benefits for retirees turn out to be a weak sort of promise. For retirees’ health benefits are not a mandatory subject of collective bargaining. As a result, the employer may change levels of benefits without dickering with the union. So much is the holding of Chemical Workers v. Pittsburgh Plate Glass Co., 404 U.S. 157, 92 S.Ct. 383, 30 L.Ed.2d 341 (1971), where the employer made a mid-term alteration without, the Court held, shirking any duty of bargaining. Health and welfare benefits after retirement turn out to be unilateral rather than bilateral contracts. An employer may bind itself by making a sufficiently strong promise, id. at 181 n. 20, 92 S.Ct. at 399 n. 20; cf. Nolde Brothers, Inc. v. Bakery & Confectionery Workers, 430 U.S. 243, 97 S.Ct. 1067, 51 L.Ed.2d 300 (1977), but the extent of this promise depends on the employer’s words and deeds, not what the union’s negotiators think or say. Many courts have concluded that in unilateral promise cases employers may reduce or even discontinue post-retirement health benefits despite language stronger than that in the Wheelabrator pacts, and without regard to extrinsic evidence of the employees’ beliefs. E.g., Boyer and Gill (both holding that the terms of the insurance policies are dispositive); Meester v. IASD Health Services Corp., 963 F.2d 194, 197 (8th Cir.1992) (“welfare benefit plans ... may be terminated at any time in the absence of a specific expression of contrary intent by the employer.”); Alday v. Container Corp. of America, 906 F.2d 660, 665 (11th Cir.1990); Howe v. Varity Corp., 896 F.2d 1107, 1110 (8th Cir.1990); Musto v. American General Corp., 861 F.2d 897, 907 (6th Cir.1988); Moore v. Metropolitan Life Insurance Co., 856 F.2d 488, 491-92 & n. 1 (2d Cir.1988). Although these opinions do not deal with collective bargaining agreements, after Pittsburgh Plate Glass the “collective” part is so much window dressing. We must treat the retirement-benefits aspect of the document as terms unilaterally announced by the employer.

Let us suppose, nonetheless, that the document is a bilateral contract. What sort of language causes duties to extend past an expiration date? The lead opinion rejects as “extreme” the view that “the contract must either use the word ‘vest’ or must state unequivocally that it is creating rights that will not expire when the contract expires.” Op. at 607. In the discussion that follows, the lead opinion does not discuss Litton, which said that it takes “explicit terms” to create benefits surviving the expiration of a collective bargaining agreement. The concurring opinion invites us to look past the text of the agreement to the “intent of the parties” but does not explain how this can be squared with Litton’s emphasis on words rather than mental states. Op. at 612 n. 2. Looking for “explicit terms,” and preferring the actual content of the parties’ agreement to the supposed content of the parties’ heads, serve vital functions.

*618One appears in William B. Tanner Co. v. SpartaToniah Broadcasting Co., 716 F.2d 1155, 1159 (7th Cir.1983): as the duration and cost of the supposed promise increase, so does the level of formality required to conclude that a promise exists. Even that staunch supporter of subjectivism, Professor Corbin, concluded that courts should not interpret ambiguous writings to create lifetime promises. Arthur Linton Corbin, 3 Contracts § 553 (1960). Instead, he believed, courts should look for terms such as “lifetime” — -language these parties used in the 1985 agreement but no earlier one. It unsettles and in the end disserves the institution of voluntary agreement to permit straws in the wind to become shackles. “People write things down in order to assign duties and allocate risks — functions vital to economic life yet defeated if courts ... use the ambiguities present in all language to frustrate the achievement of certainty.” Continental Bank, N.A. v. Everett, 964 F.2d 701, 705 (7th Cir.1992).

Another reason why we should search for “explicit terms” lies in the nature of labor relations. Unlike one-shot contracts for the sale of goods, labor agreements endure and evolve. Relational contracting over long periods requires flexibility. What worked yesterday may be counterproductive today. Labor and management need freedom to adapt their arrangement as circumstances change. E.g., Consolidated Rail Corp. v. Railway Labor Executives’ Association, 491 U.S. 299, 308-09, 109 S.Ct. 2477, 2483, 105 L.Ed.2d 250 (1989); Transportation-Communication Employees v. Union Pacific R.R., 385 U.S. 157, 160-61, 87 S.Ct. 369, 371-72, 17 L.Ed.2d 264 (1966); United Steelworkers v. Warrior & Gulf Navigation Co., 363 U.S. 574, 578-81, 80 S.Ct. 1347, 1350-52, 4 L.Ed.2d 1409 (1960); McKinney v. Missouri-Kansas-Texas R.R., 357 U.S. 265, 272-74, 78 S.Ct. 1222, 1226-27, 2 L.Ed.2d 1305 (1958). Usually parol evidence facilitates flexible arrangements. People may choose the written or oral word according to their needs. When we use extrinsic evidence to create long term promises, however, we defeat accommodation to changed circumstances. Vesting means that however much labor and management believe that a new approach is called for, they are forbidden to achieve it: the employees, as third party beneficiaries, may enforce the old agreements no matter what union and management now prefer. In using extrinsic evidence to carry out what may have been the parties’ intent yesterday, we defeat their plans today. Labor and management may tie their hands if they want, but, because eliminating the ability to bargain today is so at variance with the norm, we should be sure (by insisting on express provision) that this was indeed the agreement.

Especially when the sort of evidence that could be understood to support lifetime benefits will be present almost all of the time. Because collective bargaining arrangements last decades and govern the affairs of so many, it will be possible to come up with evidence that someone thought that arrangements under the existing agreement would last forever. Notice how distant the evidence in this case is from the usual parol evidence of meaning. Courts seeking to learn the objectives of the parties turn to the negotiations — what the negotiators said to each other, the documents they exchanged. If the negotiators used “benefits will continue after retirement” to mean “the levels of benefits will never fall,” that will show up in the signs they exchange at the table. Plaintiffs have no evidence of that kind. Instead they offer statements to show how particular persons understood the contracts after they had been signed. Readings by third parties do. not show what the negotiators were trying to achieve and do not imply that they selected phrase X to achieve objective Y. By turning to understandings rather than to the negotiations, we decrease the importance of the document the parties actually sign, making it hard for the negotiators to achieve their goals.

Extended dealings, producing multiple understandings, pose a special problem when the supposedly illuminating information is oral. Someone asked in 1990 what he “understood” about the meaning of an agreement reached in 1965 may remember and report not what he actually thought in 1965 but what occurs to him today. And what he remembers today may turn out to be whatever serves his interests today. Even for wit*619nesses who seek to report accurately, cognitive dissonance produces a remarkable consonance between events as they recall them and their contemporary interests. Juries may be good at nosing out liars, but the problem in using memories and understandings is that there will be disagreement even when no one plays false to his recollections. We will have only conflicting understandings, memories, and hopes. Divergence in reported beliefs increases as the events retreat and as the parties no longer need accommodate. These collective bargaining agreements were written for unions, managers, and arbitrators in living relations, not for juries in dead ones. Once the plant has closed, it is all too easy for jurors to award a bundle to their neighbors at the expense of a distant corporation. After a week-long trial in Senn it took the jury exactly 20 minutes to conclude that the agreements had established vested benefits, 951 F.2d at 812, although this court thought that there was no evidence at all to that effect.

Risk of a “gotcha!” at the end of a collective bargaining relation affects conduct during its course. Adaptations run from a bald declaration that nothing vests, as in Ryan v. Chromalloy American Corp., 877 F.2d 598, 600 (7th Cir.1989), to formal vesting coupled with lower levels of benefits, and everything in between. So long as the rule of law is clear, the parties can arrange their own affairs. A presumption so weak that everything is up for grabs after the fact — and this is the upshot of the majority’s approach — lacks that virtue. Exposed to irreducible risk of surprise awards or inability to adapt to changing conditions, employers respond by curtailing the categories of events they are willing to cover. Third parties, too, are affected. Many employers obtain insurance for their health and welfare plans. Sometimes the plans are defined by reference to the insurance, as Wheelabrator’s was; sometimes the insurance is defined by reference to the plan. E.g., Chicago Pacific Corp. v. Canada Life Assurance Co., 850 F.2d 334 (7th Cir.1988). Insurers do not know what managers say to workers considering retirement and cannot be bound by this extrinsic evidence. So we face the possibility of a disjunction between the benefits and their funding, or perhaps we shall end up surprising the insurers too. Either way, both employers and insurers rationally reduce their promises to take account of increased risk. Uncertainty never promotes efficient production and may hinder it.

A word about the evidence the retirees want to offer against Wheelabrator. Statements describing “lifetime” benefits sound impressive until you recognize that there are at least three distinct issues: (1) what will happen to the retirees while the current rules remain in force?; (2) must the employer continue to provide some health benefits to former workers after the current agreement has expired?; and (3) must the employer maintain an unchanged level of benefits for retirees? All of the statements concern the first or second question, while our case is about the third. Between 1965 and 1980, retirees’ benefits were less than those of active workers. Under the agreements in force from 1981 through 1988, retirees received free medical benefits tracking those enjoyed by current workers. What the plaintiffs must show in order to prevail is that Wheelabrator promised that this would never change — that the “Blue Cross-Blue Shield coverage” of which the agreements speak could expand but not shrink. None of the statements establishes such a promise. In particular, not a scrap of the extrinsic evidence concerns the meaning of: “Effective 7/1/81 these benefits include Dental, Vision and Hearing for all retirees.” Yet this sentence is the only support the lead opinion offers for its assertion that ambiguity about the level of benefits requires a trial — a trial about everything, including the copayment system. What question do my colleagues think the jury must answer? What is the ambiguous text? What are the alternative readings of that text? How should the jury be told to resolve this ambiguity? The lack of answers speaks volumes about the advisory quality of the court’s opinion. Alas, the district judge cannot escape so easily on remand.

I accept the force of the lead opinion’s observation that employers who promised health benefits to their employees without “anticipating the recent rise in health costs *620... should not expect the courts to bail them out by undoing the contractually determined allocation of risk on the question. Courts do not sit to relieve contract parties of their improvident commitments”. Op. at 609. Health care after retirement is deferred compensation. An increase in the cost of carrying out a promise does not justify abandoning the covenant. A firm that has promised too much to too many people may write down . its commitments in bankruptcy, shaving a little off each rather than tossing one group of promisees to the wolves. Still, it is necessary to ascertain the “contractually determined allocation of risk on the question.” Employers drafting in the wake of Pittsburgh Plate Glass would have supposed that unless they said otherwise explicitly, they may alter retirees’ health benefits. The understanding that at the end of a collective bargaining agreement employers may implement their last offer, unless “explicit terms” curtail this power, would have strengthened this supposition. What the majority of our court requires is that employers reiterate these rules of law in the body of the agreement or lose their protection. Although it misleads to say that rules of law are “incorporated into” contracts, see General Motors Corp. v. Romein, — U.S. -, -, 112 S.Ct. 1105, 1110-12, 117 L.Ed.2d 328 (1992), law provides the background against which the parties’ words carry meaning. Labor law supplies a benchmark: employers may change terms on the expiration of an agreement (and sometimes during its course) unless they explicitly surrender that entitlement. To say that the norm vanishes as soon as someone disputes the meaning of the agreement is to dishonor a “contractually determined allocation of risk” that depends on or incorporates that norm. Judges may say that employers “have only themselves to blame” if the employers drafted carelessly; but if courts create ambiguity by pulling the rug out from under firms that relied on the principles reflected in Pittsburgh Plate Glass and Litton, then it is we who are responsible.

Uncertainty now reigns. In at least one circuit health and welfare benefits mentioned in collective bargaining agreements presumptively vest on retirement. UAW v. Yard-Man, Inc., 716 F.2d 1476 (6th Cir.1983). In this circuit such benefits presumptively do not vest, but the presumption is weak. Other circuits have still other formulae. E.g., Anderson v. Alpha Portland Industries, Inc., 836 F.2d 1512 (8th Cir.1988). And in most circuits employers retain complete freedom to alter retirees’ health benefits in the absence of collective bargaining agreements, see eases cited at p. 617 although after Pittsburgh Plate Glass it is impossible to understand why the existence of a collective bargaining agreement augments the force of the employer’s promise. Employers with national operations are subject to multiple and inconsistent rules, compounded by the uncertainty inherent in sending big-stakes questions to the constantly-changing panels that are juries. See Michael S. Melbinger & Marianne W. Culver, The Battle of the Rust Belt: Employers’ Rights to Modify the Medical Benefits of Retirees, 5 DePaul Bus.L.J. 139, 145-60 (1992) (cataloging six distinct approaches). Uncertainty never promotes industry. Risk is an uncompensated, and substantial, cost of operations, and in this respect the fortunes of labor, capital, and consumers — that is, of us all — are linked.

. Paragraph (b) provides that Wheelabrator will establish a supplemental plan that, "when combined with Medicare, will provide a level of benefits equal to the Wheelabrator Blue Cross-Blue Shield plan for active and retired employees 65 years of age or over. The Company will pay the cost of such supplemental plan." Thus Wheelab-rator promised to give its employees a package of Medicare and private insurance equal to the private insurance provided to active employees.

. For example, a provision of the 1971 agreement reads: "Effective April, 1972, the retirees will be covered by the same drug program as the active employees.” The 1974 agreement adds: "Effective October 15, 1974, retirees and spouse [sic] will be covered by major medical coverage.” By 1980 the retirees had obtained the same dental and eye care as active employees.

.In 1985 Wheelabrator added four new medical plans. Persons who retired at age 65 and elected fully vested benefits were not entitled to choose the most generous of the five plans. Nothing turns on this complication, however.

. The sentence in the 1980 contract about the $2 copayment for drugs does not assist the retirees. The drug program is subject to this provision: "Effective April, 1972, the retirees will be covered by the same drug program as the active employees.” An entitlement to "the same drug program as the active employees” is the antithesis of a ratchet.