Opinion filed PER CURIAM.
Appellants — American Medical International, Inc. (AMI), more than thirty of its hospital subsidiaries, and a hospital it manages — challenged in the District Court a decision by the Administrator of the Health Care Financing Administration, a division of the Department of Health and Human Services, denying reimbursement of certain costs that appellants allegedly absorbed while furnishing health care to beneficiaries of the Medicare program. 1 The District Court upheld that ruling and rejected the claim. 2
For reasons amply delineated by the District Court, 3 we agree that the costs in question are not reimbursable because they were not necessarily incurred in the provision of health-care services to Medicare patients, as is required by the Medicare Act. 4 That does not entirely dispose of the case, however, for the court did not address one issue: Whether the Administrator is collaterally estopped from withholding reimbursement of so-called “stock maintenance costs,” 5 the status of which was a question raised by appellant AMI-Chanco 6 and decided against the Federal Government in earlier litigation in the Court of Claims. 7 That problem implicates some of the most difficult and troubling aspects of the law of issue-preclusion, and thus, we believe, demands full treatment.
I
The Secretary insists that an estoppel is foreclosed by the Supreme Court’s decision in Commissioner v. Sunnen 8 three decades *120 ago. That argument, at first glance, seems persuasive. 9 So long as the facts in two cases are formally “separable,” the Sunnen Court declared, it is inappropriate to use a judgment in the first to estop a party on a legal point in the second:
[I]f the relevant facts in the two cases are separable, even though they be similar or identical, collateral estoppel does not govern the legal issues which recur in the second case. Thus the second proceeding may involve an instrument or transaction identical with, but in form separable from, the one dealt with in the first proceeding. In that situation, a court is free in the second proceeding to make an independent examination of the legal matters at issue.... Before a party can invoke the collateral estoppel doctrine in these circumstances, the legal matter raised in the second proceeding must involve the same set of events or documents and the same bundle of legal principles that contributed to the rendering of the first judgment. 10
This prong of Sunnen has never been popular with commentators — who maintain that formal separability of facts, without some difference pertinent to the first decision, does not justify denial of preclusive effect to the initial adjudication 11 — and has also been less scrupulously applied by the courts. 12 If valid, however, Sunnen’s separable-facts doctrine would, as the Secretary suggests, appear to control here, for the Court of Claims’ litigation involved different years — and hence a different, though for all relevant purposes an identical, bundle of facts — from those in the case at bar. 13
There is, however, no need to wrestle with the intricacies of this problem, for the controversial aspect of Sunnen is no longer good law. In one of its most recent analyses of collateral estoppel, the Supreme Court allowed a judgment predicated upon one set of contracts to estop litigation of legal issues in another case dealing with a similar, though unrelated, set of contracts, because there had been no “changes in facts essential to [the] judgment.” 14 The first ruling had not been “predicated” 15 on facts peculiar to the contracts there in issue, so that differences in the documents were not of “controlling significance.” 16 While the Court did not expressly overrule the separable-facts doctrine, it characterized Sunnen as holding simply “that modifications in *121 ‘controlling legal principles’ . . . could render a previous determination inconsistent with prevailing doctrine;” 17 “unless there have been major changes in the law,” the Court said, Sunnen is inapplicable. 18 Since this cannot be squared with Sunnen’s declaration that “the legal matter raised in the second proceeding must involve the same set of events or documents” as that involved in the first, 19 it seems clear that “Sunnen was limited by the Court ... to cases in which there had been a significant ‘change in the legal climate....’” 20 Plainly that is not the case here.
II
Although it is thus apparent that an estoppel is not inexorably foreclosed by Supreme Court authority, the inquiry cannot end at this point. “[S]pecial circumstances,” the Court has admonished, “warrant an exception to the normal rules of preclusion.” 21 It remains to inquire whether the circumstances here are so unusual that the administration of justice would best be served by refusing an estoppel in the instant proceeding.
Were this a case involving only private litigants or only simple issues of fact, we would not hesitate to conclude that an estoppel should arise. The case would then stand like any other in which “the issues presented by [the] litigation are in substance the same as those resolved [earlier, and the] controlling facts or legal principles have [not] changed significantly since the [first] judgment....” 22 This is not, however, a typical case. A federal agency, not a private party, lost on an issue of federal law, not an issue of fact, in the first lawsuit. To allow nonparties to the Court of Claims’ ruling 23 to win simply on the basis of an estoppel would mean that we simply and uncritically bind ourselves to follow another court’s interpretation of a federal statute in virtually all cases involving that legislation. The broader and more serious implication of such a holding is that the first court to hear a case raising a public law issue litigable only with the Federal Government would — if it ruled against the Government — rigidify the law to be applied by every court in every case presenting that issue. 24 What makes that both possible and *122 alarming are the Supreme Court’s relatively recent estoppel decisions doing away with mutuality, 25 sanctioning application of offensive estoppel, 26 and generally broadening the permissible uses of issue-preclusion when questions of law are involved. 27
This court has not yet faced squarely the problem whether, given the new estoppel rules, federal relitigation of questions of national law with new parties in new circuits should be permitted. 28 Three other courts, however, have addressed that question, 29 and each has concluded that no collateral estoppel should arise in such circumstances. 30 They all bottom their decisions on one central policy argument: that “in the context of agency relitigation of legal issues with substantial public policy implications,” 31 “inquiry by other circuits should not be foreclosed; indeed, a conflict among the circuits could be a healthy matter.” 32 And whatever might be said of the technical legal arguments made by these courts 33 the policy they adopted is valid and, we believe, ultimately persuasive.
It can hardly be gainsaid that complex questions of national import can wholesomely and profitably be explored by more than a single court. And it is a truism that proposed legal rules can be improved through repeated examination by a variety of legal minds. Rigorous analysis of the law and concomitantly its healthy development would thus be hampered were an estoppel indulged in a case such as this. By the same token, independent reconsideration of legal issues by the circuits can facilitate Supreme Court review by highlighting their complications and controversial aspects, 34 and thus make for better informed *123 decisions. Indeed, the Court itself has acknowledged the value of such a shifting process in the courts of appeals. In a recent case presenting a problem dividing the circuits, the Court noted that “[t]his litigation exemplifies the wisdom of allowing difficult issues to mature through full consideration by the courts of appeals. By eliminating ... many subsidiary, but still troubling, arguments . .. these courts have vastly simplified our task, as well as having underscored the reasonableness of the [ultimately prevailing] view.” 35
This conclusion draws support from the treatment traditionally accorded decisions of sister circuits by the courts of appeals themselves. When rases presenting legal questions with national implications have arisen, this court has always considered itself obligated to subject the analyses of coordinate tribunals to close scrutiny, with the aim of producing a sound and well-reasoned decision. 36 More generally, while the courts of appeals consistently recognize the importance of uniformity, 37 they have never considered themselves hidebound by other circuits on legal questions involving federal-agency defendants, 38 perhaps because “[i]ssue preclusion has never been applied to issues of law with the same rigor as to issues of fact.” 39 Moreover, relitigation of legal questions across the circuits has been the policy of many federal agencies. 40 To an extent, of course, the vitality of these practices has been threatened by the Supreme Court’s recent decisions on collateral estoppel. But the earlier law cannot be dismissed out of hand; the policies under-girding it 41 are the very sort of countervailing factors 42 that, absent clear Supreme Court authority to the contrary, may justify a refusal to apply estoppel.
This is not to say that arguments opposing relitigation of issues lose all force in cases like the one at bar. Duplicative litigation has its own irreducible minimum of waste, 43 and there is much to be said in favor of invariable sameness and resulting certainty in the law. That it is more important to have a sure rule than to have the right rule is a dictum noted with such frequency that it has evolved from truism to cliche. 44 And there is an element of inequity in a special rule applicable only in litigation against the Government, particularly since federal agencies are in a better position than most litigants to petition the Supreme Court or Congress to modify what they believe to be erroneous interpretations of national law. 45 But all that these considerations suggest is that the courts should be cautious in declining to apply estoppel to *124 previously litigated questions of law, not that they should never do so.
Ill
The foregoing analysis intercepts only parties who did not participate in the initial suit. If the Government is litigating with the same adversary in both cases, an estoppel will petrify the law only as to that one party — obviously not an unjust result — and accordingly will work no preclusion generating national shockwaves. 46 What has been said up to this point, then, does not apply to AMI-Chanco, which was directly involved in the Court of Claims litigation. 47
That brings us to the crux of the matter here. It patently would be unfair to apply one rule for AMI and AMI-Chanco, and another for the other subsidiaries. The consensus of the legal profession is that “[rjefusal of preclusion is ordinarily justified if the effect of applying preclusion is to give one person a favored position in current administration of the law.” 48 We are in full accord. Collateral estoppel “is not meant to create vested rights in decisions . . . thereby causing inequities among [similarly situated parties].” 49 To permit an estoppel here would do exactly that. The judgment of the District Court is accordingly
Affirmed.
Notes
.
American Medical Int’l, Inc. v. Secretary of HEW,
. Id. at 609.
. Id. at 612-616.
. 42 U.S.C. §§ 1395 et seq. (1976).
. These are costs attending maintenance of a corporation’s capital stock: accounting and other expenses related to Securities and Exchange Commission filings, stock transfer fees, and expenses in connection with stockholder annual reports and meetings.
On the basis of
Memorial, Inc. v. Harris,
. AMI-Chanco, which merged with AMI in 1972, is now an AMI subsidiary.
.
AMI-Chanco v. United States,
.
. Brief of Appellee at 38. The Secretary also advances two less potent contentions against an application of estoppel. The first is that only AMI-Chanco can seek to estop, since only it was involved in the initial suit, and assertedly it is not in privity with either AMI or its sister subsidiaries.
Id.
at 36-37. But in light of the Supreme Court’s recent decisions in
Montana v. United States,
.
Commissioner v. Sunnen, supra
note 8,
. See,
e.g.,
IB J. Moore, Federal Practice ¶¶ 0.442[4], 0.448 at 3415, 4237-4239 (1980). On how different the two causes of action can be, see
Carr v. District of Columbia, supra
note 9,
. Compare
Starker v. United States,
. See note 7 supra.
.
Montana v. United States, supra
note 9,
.
Id.,
.
Id.
at 160,
.
Id.
at 161,
. Id.
.
Commissioner v. Sunnen, supra
note 8,
.
Starker v. United States, supra
note 12,
.
Montana v. United States, supra
note 9,
.
Montana v. United States, supra
note 9,
. Most of the entities attacking the Administrator’s decision here were not parties to the proceeding in the Court of Claims. See note 9 supra and text supra at note 7.
. The type of situation we refer to has a particular makeup. The legal issue must be such — like the one now before us — that a federal agency or officer necessarily will be a party to every suit in which it arises. If private parties can litigate the issue between themselves, the law cannot be frozen by a single ruling, for they will not be bound by prior adjudications with which they were not associated. Furthermore, the governmental unit must have lost the first case presenting the question; for if it won the first but loses subsequently, it is sheltered by
Parklane’s
caveat on inconsistent prior decisions. See
Parklane Hosiery Co. v. Shore, supra
note 9,
.
Parklane Hosiery Co. v. Shore, supra
note 9,
.
Id.;
see also
Blonder-Tongue Laboratories, Inc. v. University of Illinois Foundation,
. Compare IB J. Moore,
supra
note 11, H 0.448 at 4234, Restatement of Judgments § 70 (1942), and
Moser v. United States,
. We have held the Government estopped when legal issues of only local import — those unlikely to be raised in other circuits — were resolved.
E.g. Carr v. District of Columbia, supra
note 9,
.
Divine v. Commissioner,
. But see
Starker v. United States, supra
note 23,
.
Western Oil & Gas Ass’n v. EPA, supra
note 27,
.
United States v. Anaconda Co., supra
note 29,
.
Western Oil & Gas Ass’n
relied in part on the continuing vitality of the
Sunnen
separable-facts doctrine, see
.
Cf. Train v. Natural Resources Defense Council, Inc.,
.
E. I. du Pont de Nemours & Co. v. Train,
. See,
e.g., United States v. Washington Post Co.,
. See,
e.g., Federal Life Ins. Co. v. United States,
. See
Western Oil & Gas Ass’n v. EPA, supra
note 20,
.
Segal v. American Tel. & Tel. Co.,
. See Vestal, supra note 38, 55 N.C.L.Rev. at 123.
. See Restatement (Second) of Judgments § 68.1 at 30-31 (Tent. Draft No. 4, 1977).
. See cases cited supra note 21.
. See Vestal, supra note 38, 55 N.C.L.Rev. at 127, 169-170; Kelly & Rothenberg, supra note 33, 13 J.L.Ref. at 323-324.
. See
Burnet v. Coronado Oil & Gas Co.,
. See Vestal, supra note 38, 55 N.C.L.Rev. at 178.
. See note 24 supra. It is true, of course, that legal analysis might well benefit were all the circuits free to consider independently identical questions of law arising between a single party and the Government — or, for that matter, two private parties. It is also true that such treatment might facilitate Supreme Court review in ways already described. See text supra at notes 34-35. Estoppels in such cases would not entirely block those avenues, however, for those not parties to the first suit could still litigate the issue. In such situations, then, polides favoring preclusion would appear to outweigh the disadvantages discussed above.
. See note 9 supra and notes 6-7 supra and accompanying text.
. Restatement (Second) of Judgments § 68.1 at 33 (Tent. Draft No. 4, 1977).
.
Commissioner v. Sunnen, supra
note 8,
