concurring.
I join the judgment and opinion of the court in full and without reservations, but I would go even further than my colleagues and hold that promissory estoppel can never be used to alter the terms of a defined-benefit plan, especially when it is a multiemployer plan.
It is an unsettled question to what extent or even whether estoppel, either equitable or promissory, is a permissible basis for claiming benefits from an ERISA plan. Coker v. Trans World Airlines, Inc., 165 F.3d 579, 585 (7th Cir.1999); City of Hope National Medical Center v. Healthplus, Inc., 156 F.3d 223, 230 n. 9 (1st Cir.1998); Jensen v. SIPCO, Inc., 38 F.3d 945, 953 *904(8th Cir.1994). ERISA does not mention promissory estoppel, and its strict injunction that benefits may be paid only in accordance with the written plan (if there is one), 29 U.S.C. § 1102(a)(1); Clair v. Harris Trust & Savings Bank, No. 98-2334, 1999 WL 600403, at *2 (7th Cir. Aug. 10, 1999), and its specification of procedures for amending plans, § 1102(b)(3), cut against any inference that benefits can be predicated on a promise not made in the plan itself or (for reasons I explain later) the summary plan description. Coker v. Trans World Airlines, Inc., supra, 165 F.3d at 585; Coleman v. Nationwide Life Ins. Co., 969 F.2d 54, 58-60 (4th Cir.1992); see Health Cost Controls of Illinois, Inc. v. Washington, 187 F.3d 703, 711 (7th Cir.1999). However, courts in ERISA cases retain their normal common law powers to fill gaps in the statute. Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 110, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989); Miller v. Taylor Insulation Co., 39 F.3d 755, 758 (7th Cir.1994); Kane v. Aetna Life Ins., 893 F.2d 1283, 1285 (11th Cir.1990). For statutes are rarely comprehensive; they are enacted against a rich background of common law principles that can be drawn on, as necessary, to put flesh on the legislative skeleton. But any common law elaborations of an incomplete statute must be consistent with the statute’s language and animating policies. Mertens v. Hewitt Associates, 508 U.S. 248, 259, 113 S.Ct. 2063, 124 L.Ed.2d 161 (1993); D’Oench, Duhme & Co. v. FDIC, 315 U.S. 447, 470-75, 62 S.Ct. 676, 86 L.Ed. 956 (1942) (Jackson, J., concurring); Mathews v. Sears Pension Plan, 144 F.3d 461, 465-66 (7th Cir.1998); Henry J. Friendly, “In Praise of Erie — and of the New Federal Common Law,” 39 N.Y.U. L.Rev. 383, 405-22 (1964).
Consistent with these principles, this court “has recognized a form of estoppel as a matter of federal common law in at least some ERISA cases,” Coker v. Trans World Airlines, Inc., supra, 165 F.3d at 585 (7th Cir.1999); Principal Mutual Life Ins. Co. v. Charter Barclay Hospital, Inc., 81 F.3d 53, 57 (7th Cir.1996); Miller v. Taylor Insulation Co., supra, 39 F.3d at 758; Thomason v. Aetna Life Ins. Co., 9 F.3d 645 (7th Cir.1993); Black v. TIC Investment Corp., 900 F.2d 112, 115 (7th Cir.1990), but only a very narrow form that can rarely be invoked successfully, being “expressly limited ... to ‘claims for benefits under unfunded single-employer welfare benefit plans’ out of concern that a broader application of estoppel principles might pose a risk to plans that rely on actuarial soundness.” Coker v. Trans World Airlines, Inc., supra, 165 F.3d at 585, quoting Black v. TIC Investment Corp., supra, 900 F.2d at 115 (emphasis added). See also Krawczyk v. Harnischfeger Corp., 41 F.3d 276, 280 (7th Cir.1994). Anything beyond that would be inconsistent with the principles set forth in the preceding paragraph concerning the appropriate scope of judicial completion of ERISA. The cases that I have just cited imply quite rightly that a claim for ERISA benefits that is based on the doctrine of promissory estoppel can never succeed in a case involving a defined-benefit plan. As does, though a little less emphatically, Russo v. Health, Welfare & Pension Fund, 984 F.2d 762, 767 n. 4 (7th Cir.1993), where we said that “our concerns about actuarial soundness expressed in Black would argue against applying estoppel to a multi-employer funded pension plan.” As does the case law of the other circuits. See Jensen v. SIPCO, Inc., supra, 38 F.3d at 953 n. 4; Watkins v. Westinghouse Hanford Co., 12 F.3d 1517, 1527-28 (9th Cir.1994); Miller v. Coastal Corp., 978 F.2d 622, 624-25 (10th Cir.1992); Armistead v. Vernitron Corp., 944 F.2d 1287, 1299-1300 (6th Cir.1991).
A defined-benefit plan guarantees a specified pension benefit to the plan participants when they retire, in contrast to a defined-eontribution plan, which gives each participant when he retires whatever amount happens to be in his defined-contribution plan then. See, e.g., Hughes Air*905craft Co. v. Jacobson, — U.S. -, -, 119 S.Ct. 755, 761, 142 L.Ed.2d 881 (1999); Clair v. Harris Trust & Savings Bank, supra, 1999 WL 600403 at *1. To determine how much the employer and the employees must contribute to the plan in order to fund the benefits (that is, in order to make sure that the plan will have sufficient assets to pay benefits as they come due to retiring participants) requires complex actuarial calculations. The starting point for calculation is the terms of the plan. See Central States, Southeast & Southwest Areas Pension Fund v. Gerber Truck Service, Inc., 870 F.2d 1148, 1151 (7th Cir.1989) (en banc). If terms are added by the operation of promissory es-toppel that appear nowhere in the plan documents upon which the actuarial calculations are based, that the actuaries who designed the funding mechanism in the plan did not know about, and that, unlike terms appearing in a summary plan description (which is a statutorily required plan document, 29 U.S.C. § 1022(a), on which the participant is entitled to rely, Health Cost Controls of Illinois, Inc. v. Washington, supra, 187 F.3d at 711), do not purport merely to summarize the plan, the plan may turn out to be seriously underfunded. These considerations are decisive against allowing the invocation of promissory estoppel in any case involving a defined-benefit plan. I thus would not either require the defendant to prove, or permit the plaintiff to disprove, that honoring the promise would actually impair the actuarial soundness of the plan. I would not subject defined-benefit plans to the expense and uncertainty of litigating actuarial issues.
This conclusion is even clearer in a case such as this in which the promise is made to a number of workers, rather than just one, since the actuarial implications are graver the more workers are involved. And it is particularly clear in a case— which is also this case — in which the defined-benefit plan is a multiemployer plan. For the consequence of the underfunding may then be that an employer who had absolutely nothing to do with the promise that is the basis for working an estoppel may have to dip into his pocket to make up the shortfall. This consideration led us in Gerber to hold that multiemployer plans are not subject to equitable defenses that would ordinarily excuse an employer from having to contribute to a plan. 870 F.2d at 1153-55. Gerber nixes efforts by employers to appeal to equity to pay less than the plan requires; it stands to reason that employees should not be allowed to appeal to equity to get more than the plan allows. They should not be allowed to claim benefits that, being off-plan, may impose a financial burden on doubly innocent employers: employers who never employed these employees and who may have had equitable grounds that they were forbidden to invoke for paying less than the law required them to pay.
So Shields’s claim fails at the threshold and it is therefore unnecessary to consider whether the promise on which he bases his suit was sufficiently definite, and induced sufficient detrimental reliance, to satisfy the elements of the doctrine of promissory estoppel, although the court is correct that it was neither. It is enough that this was a defined-benefit plan, and if that is not enough, that it was a promise made to a number of workers, and if more is needed (it is not) that it was a multiemployer plan.