SUTTON, J., delivered the opinion of the court in which GIBBONS, J., joined. DONALD, J. (pp. 686-91), delivered a separate dissenting opinion.
OPINION
SUTTON, Circuit Judge.In litigation, as in film, sequels rarely satisfy. This case is no exception. Three *683years ago, we remanded this dispute to the district court for factfinding necessary to determine whether CNH America’s proposed modifications to its retiree healthcare benefits are reasonable. The district court did not reach the reasonableness question and did not create a factual record that would permit us to answer the question on our own. As a result, we reverse and remand for further proceedings.
I.
Our previous opinion makes it unnecessary to recount the protracted history of this litigation. See Reese v. CNH America LLC, 574 F.3d 315, 318-20 (6th Cir.2009). There, we considered two questions: “Did [CNH] in the 1998 CBA agree to provide health-care benefits to retirees and their spouses for life? And, if so, does the scope of this promise permit CNH to alter these benefits in the future?” Id. at 321. In answering the first question, we rejected CNH’s claim that the CBA permitted the company to terminate the benefits, holding that eligibility for lifetime healthcare benefits had “vested.” Id. at 322; see Yolton v. El Paso Term. Pipeline Co., 435 F.3d 571, 580 (6th Cir.2006); UAW v. Yard-Man, Inc., 716 F.2d 1476, 1482 (6th Cir.1983).
In answering the second question (“What does vesting mean in this setting?”), we rejected the suggestion that the scope of this commitment in the context of healthcare benefits, as opposed to pension benefits, meant that CNH could make no changes to the healthcare benefits provided to retirees. Unlike pension obligations, we explained, healthcare benefits cannot readily be monetized at retirement or for that matter practically fixed. See Reese I, 574 F.3d at 324. Doctors and medical-insurance providers come and go. Medical plans change from year to year. And fixed, unalterable medical benefits at all events are not what retirees want. Nothing, indeed, would make employers happier than to know that vesting in the healthcare-benefits context meant the same thing as vesting in the pension context. For then, a company faced with the obligation could account for what it had spent on each employee for healthcare benefits on the day of retirement, then commit to spend no less through the end of the retiree’s (and spouse’s) life. Nor would most employers be troubled if this commitment, like most defined-benefit pension plans, increased based on inflation as measured by the consumer-price index. The reality is that, even though we have relied on language tying healthcare benefits to pension benefits as a basis for determining that healthcare benefits have vested, vesting in the context of healthcare benefits provides an evolving, not a fixed, benefit. See Yolton, 435 F.3d at 583 (relying on language in the summary plan descriptions saying that “continued coverages will be the same as those that were in effect on the day preceding your retirement”)-, see also Noe v. PolyOne Corp., 520 F.3d 548, 558 (6th Cir.2008); McCoy v. Meridian Auto. Sys., Inc., 390 F.3d 417, 422 (6th Cir.2004); Golden v. Kelsey-Hayes Co., 73 F.3d 648, 656 (6th Cir.1996).
The rub for retirees and employers alike is that healthcare benefits — what is provided and what it costs — have not been remotely static in modern memory. The reason has little to do with traditional causes of inflation and more to do with the expansion of the benefit: the remarkable growth in modern life-saving and comfort-improving medical procedures, devices and drugs. New and better medical procedures arise while others become obsolete. And it is the rare medical innovation that costs less than the one it replaces. Retirees, quite understandably, do not want lifetime eligibility for the medical-insurance *684plan in place on the day of retirement, even if that means they would pay no premiums for it. They want eligibility for up-to-date medical-insurance plans, all with access to up-to-date medical procedures and drugs. Whatever else vesting in the healthcare context means, all appear to agree that it does not mean that beneficiaries receive a bundle of services fixed once and for all. Companies want the freedom to change health-insurance plans. And beneficiaries want something more than a fixed, unalterable bundle of services; they want coverage to account for new and better, yet likely more expensive, procedures and medications than the ones in existence at retirement.
All of this was borne out by the parties’ implementation of the relevant collective bargaining agreements — in at least two respects. As explained in our prior opinion, the 1998 CBA “created a Managed Health Care Network Plan for past and future retirees. In other words, it imposed managed care on all of them, which represented a reduction in the effective choices of coverage available for all retirees and the coverage actually provided to many, if not most, of them.” Reese I, 574 F.3d at 325. “Pre-1998 retirees thus saw their coverage downgraded in at least one respect: Unlike the prior plan, under which they could choose any doctor without suffering a financial penalty, they generally had to pay more for choosing an out-of-plan doctor.” Id. Other cases reach the same conclusion. Winnett v. Caterpillar, Inc., 609 F.3d 404, 412-13 (6th Cir.2010) (explaining that a change to managed care is a “significant change” representing a “reduction” in benefits); cf. UAW v. Aluminum Co. of Am., 932 F.Supp. 997, 1011 (N.D.Ohio 1996) (challenging employer’s adoption of a managed-care plan as a reduction in benefits).
Also confirming that the parties did not perceive the relevant CBAs as establishing fixed, unalterable benefits was the passage of the Medicare Prescription Drug, Improvement, and Modernization Act of 2003, Pub.L. No.. 108-173, 117 Stat.2066. No one batted an eye when the healthcare plans for which retirees were eligible were modified to account for the creation of Medicare Part D, the prescription-drug benefit for seniors.
In view of the distinction between the vesting of eligibility for a benefit and the scope of that commitment and in view of the parties’ practice under the 1998 CBA of altering healthcare benefits under CBAs with materially identical language, we concluded that CNH could make “reasonable” changes to the healthcare plan covering eligible retirees. Reese, 574 F.3d at 325-27; see also Zielinski v. Pabst Brewing Co., 463 F.3d 615, 619 (7th Cir.2006) (distinguishing the vesting of healthcare benefits from the scope of that commitment and holding that the employer was obligated to provide benefits “reasonably commensurate” with pre-retirement plans). We listed three considerations: Does the modified plan provide benefits “reasonably commensurate” with the old plan? Are the proposed changes “reasonable in light of changes in health care”? And are the benefits “roughly consistent with the kinds of benefits provided to current employees”? Reese I, 574 F.3d at 326. We remanded the case to the district court to take evidence and to decide in the first instance the legal question whether CNH’s proposed modifications were reasonable. Id. at 327.
Back in the district court, CNH moved for approval of its proposed modifications to the benefits, introducing evidence (including affidavits from its employee-benefits director and an employee-benefits consultant and attorney) that the changes were reasonable. The plaintiffs introduced little new evidence on the issue of *685reasonableness. They instead relitigated several questions our court had already decided, moving for summary judgment on the ground that CNH lacks the ability to modify any benefits, save at the approval of the union that once represented them. The district court agreed and granted their motion.
II.
The plaintiffs and the district court misread the panel opinion. In holding that “CNH ... may reasonably alter” the plaintiffs’ benefits, we recognized that CNH could alter them on its own, not as part of a new collective-bargaining process. Reese I, 574 F.3d at 327. Past changes to retiree healthcare benefits, we noted, had not been collectively bargained, which comes as no surprise since “a union does not represent retired employees when it bargains a new contract for its employees.” Id. at 324; see Allied Chem. & Alkali Workers of Am., Local Union No. 1 v. Pittsburgh Plate Glass Co., 404 U.S. 157, 172, 92 S.Ct. 383, 30 L.Ed.2d 341 (1971). The district court, in short, erred when it disregarded our holding that the company may make reasonable modifications to the plaintiffs’ healthcare benefits.
That leaves two options: remand the case to the district court yet again or resolve the reasonableness question as a matter of law based on the evidence in the record. It is tempting to resolve the case now. This long-running dispute needs to come to an end, and it is particularly unfair to prolong the dispute when the status quo — a preliminary injunction in favor of the plaintiffs — not only favors just one party but also risks mooting the economic stakes of the case for the other party. Yet this sticky reality remains. The case turns in part on facts not in the record: How much did retirees pay for their health care under the old plan? How much did CNH pay? How much will the retirees and CNH each pay under the new plan, and how quickly are each side’s costs likely to grow? How does the quality of care provided under the old plan compare to the quality of care under the new plan? Do the retirees’ benefits differ in material respects from those offered to current employees and people retiring today? How do the benefits compare to benefits offered by other companies in similar industries? In the absence of a record on these points, the answer to the reasonableness question is a shot in the dark.
To gauge whether CNH has proposed reasonable modifications to its healthcare benefits for retirees, the district court should consider whether the new plan provides benefits “reasonably commensurate” with the old plan, whether the changes are “reasonable in light of changes in health care” (including access to new medical procedures and prescriptions) and whether the benefits are “roughly consistent with the kinds of benefits provided to current employees.” Reese I, 574 F.3d at 326. In doing so, the district court should take evidence on the following questions (and others it considers relevant to the reasonableness question):
• What is the average annual total out-of-pocket cost to retirees for their healthcare under the old plan (the 1998 Group Benefit Plan)? What is the equivalent figure for the new plan (the 2005 Group Benefit Plan)?
• What is the average per-beneficiary cost to CNH under the old plan? What is the equivalent figure for the new plan?
• What premiums, deductibles and co-payments must retirees pay under the old plan? What about under the new plan?
• How fast are the retirees’ out-of-pocket costs likely to grow under the old *686plan? What about under the new plan? How fast are CNH’s per-benefíciary costs likely to grow under each?
• What difference (if any) is there between the quality of care available under the old and new plans?
• What difference (if any) is there between the new plan and the plans CNH makes available to current employees and people retiring today?
• How does the new plan compare to plans available to retirees and workers at companies similar to CNH and with demographically similar employees?
It is not lost on us that the reasonableness inquiry is a vexing one. But the difficulty of the inquiry flows at least in part from the vagueness of the commitment underlying this litigation. It is well to remember the language of the relevant commitment: “Employees who retire under the Case Corporation Pension Plan for Hourly Paid Employees after 7/1/94, or their surviving spouses eligible to receive a spouse’s pension under the provisions of that Plan, shall be eligible for the Group benefits as described in the following paragraphs.” Id. at 318. What that means in the context of ever-changing medical-care developments, and ever-changing healthcare plans, is not easy. But if the parties cannot resolve the point on their own, we (and the district court) will do our best to resolve it for them.
One other thing on this score: the district court concluded that the East Moline Shutdown Agreement created “unalterable and irreducible” healthcare benefits for those who retired under it. R. 304 at 22. The Moline Agreement, however, governs only the “economic closedown benefits and the eligibility rules established and set forth in” that agreement. R.290-2 at 19. Nothing in the agreement speaks to healthcare. It neither establishes nor sets forth healthcare benefits, which are instead found in the 1998 CBA, which are vested, and which remain subject to reasonable modification.
In view of this disposition of the appeal, we think it premature to address the parties’ attorney-fees arguments. For one, this case is not over. For another, the district court may wish to revisit its attorney-fee decision based on its resolution of the questions identified above.
The dissent proposes a different path— that we reconsider our decision in Reese I and go back to square one. Unlike Judge Gibbons and me, Judge Donald did not sit on Reese I, so it would not be fair to call this an about-face or buyer’s remorse. At the same time, the law-of-the-case doctrine presents a serious impediment to this approach. And the reality that neither party to this dispute has invoked an exception to the law-of-the-case doctrine presents a conclusive impediment to it.
III.
For these reasons, we reverse and remand for proceedings consistent with this opinion.