June 16, 2000
APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF MASSACHUSETTS [Hon. Reginald C. Lindsay, U.S. District Judge] —
Christopher N. Souris, with whom Krakow Souris, LLC was on brief, for appellants.
Richard D. Wayne, with whom Charles E. Schaub, Jr., Willard Krasnow, and Hinckley, Allen Snyder LLP were on brief, for appellee.
Before Selya, Boudin and Lynch, Circuit Judges.
In this matter, the district court, understandably deeming itself bound by our holding in Williams v. Ashland Eng'g Co., 45 F.3d 588 (1st Cir. 1995), ruled that the plaintiffs' action — brought to enforce claims under a labor and materials bond for wages and fringe benefit contributions allegedly due in respect to construction of a public works project in Peabody, Massachusetts — was preempted by the Employee Retirement Income Security Act of 1974 (ERISA), 29 U.S.C. § 1001-1461, and in particular, 29 U.S.C. § 1144(a). Accordingly, the court terminated the action by entering judgment on the pleadings.
We have considerably greater freedom than the district courts to evaluate the impact of recent Supreme Court precedent on our previous decisions. Having reexamined Williams against the changed legal landscape that now confronts us, we are persuaded that subsequent developments have overtaken our decision. Consequently, we abrogate the central holding of Williams, reverse the judgment below, and remand for further proceedings consistent with this opinion.
Following the procedure described in cases such as Trailer Marine Transport Corp. v. Rivera Vazquez, 977 F.2d 1, 9 n. 5 (1st Cir. 1992), and Gallagher v. Wilton Enterprises, Inc., 962 F.2d 120, 124 n. 4 (1st Cir. 1992), the panel opinion in this case was circulated to all active judges of the court prior to publication. None interposed an objection to our proposed course of action. We caution, however, that the use of this informal procedure does not convert this opinion into an opinion en banc, nor does it preclude a suggestion of rehearing en banc on any issue in the case.
The individual plaintiffs performed carpentry work on a public works project in Peabody, Massachusetts. All of them belonged to Carpenters Local No. 26 (the union). At the times relevant hereto, their employer, Henry Construction, Inc. (Henry), was a party to a collective bargaining agreement with the union that required contributions to various fringe benefit funds on the individual plaintiffs' behalf. Henry defaulted on this obligation before completing the Peabody job.
In Massachusetts, a so-called bond statute, Mass. Gen. Laws ch. 149, § 29, the pertinent text of which is set forth in the margin, requires the general contractor on a public works project to post a bond covering labor and materials (including indebtedness incurred by subcontractors and suppliers for wages and fringe benefits). The general contractor on the Peabody project obtained such a bond from defendant-appellee United States Fidelity Guaranty Company (USF G). When Henry, a subcontractor, failed to make contributions in respect to fringe benefits, the union's collection agent, Massachusetts Carpenters Central Collection Agency (MCCCA), acting for the aggrieved employees, staked a claim on the bond.
The statute provides:
Officers or agents contracting in behalf of the commonwealth or in behalf of any county, city, town, district or other political subdivision of the commonwealth . . . for the construction, reconstruction, alteration, remodeling, repair or demolition of public buildings or other public works . . . shall obtain security by bond in an amount not less than one half of the total contract price, for payment by the contractor and subcontractors for labor performed or furnished and materials used or employed therein [subject to certain restrictions]; for payment of transportation charges for materials used or employed therein . . .; for payment by such contractor and subcontractors of any sums due for the rental or hire of vehicles . . . and other appliances and equipment . . .; for payment of transportation charges directly related to such rental or hire; and for payment by such contractor and subcontractors of any sums due trustees or other persons authorized to collect such payments from the contractor or subcontractors, based upon the labor performed or furnished as aforesaid, for health and welfare plans, supplementary unemployment benefit plans and other fringe benefits which are payable in cash and provided for in collective bargaining agreements between organized labor and the contractor or subcontractors. . . .
Mass. Gen. Laws ch. 149, § 29.
Resolution of the controversy eluded the parties. The individual plaintiffs, the union, and MCCCA (hereinafter collectively the appellants) then sued USF G in the state superior court. The surety removed the action, see 28 U.S.C. § 1441, and immediately sought judgment on the pleadings. The appellants opposed this initiative and moved to remand the action to the state court. On November 3, 1998, the federal district court denied the motion to remand. Some seven months later, it granted USF G's motion for judgment on the pleadings. Relying on Williams, the court anchored both orders in ERISA preemption. This appeal followed.
ERISA is a comprehensive statutory scheme that governs employee benefit plans. It was enacted in response to growing concerns about "the mismanagement of funds accumulated to finance employee benefits and the failure to pay employees benefits from accumulated funds." Massachusetts v. Morash, 490 U.S. 107, 115 (1989). The statute brooks no interference; it contains an express preemption clause providing that it shall "supersede any and all State laws insofar as they may now or hereafter relate to any [covered] employee benefit plan." 29 U.S.C. § 1144(a). Thus, when state-law claims "relate to" ERISA plans, those claims are transmuted into ERISA claims. See Whitt v. Sherman Int'l Corp., 147 F.3d 1325, 1329 (11th Cir. 1998); Parrino v. FHP, Inc., 146 F.3d 699, 703 (9th Cir.), cert. denied, 525 U.S. 1001 (1998). In that situation, "any civil complaint raising [such] a state law claim . . . is of necessity so federal in character that it arises under federal law for purposes of 28 U.S.C. § 1331 and permits removal to federal court." Plumbing Indus. Bd. v. E.W. Howell Co., 126 F.3d 61, 66 (2d Cir. 1997).
Despite this prophylaxis, ERISA preemption is not inexorable. As the language of section 1144(a) makes plain, the incidence of ERISA preemption turns on the parameters of the phrase "relate to." See California Div. of Labor Standards Enforcement v.Dillingham Constr., 519 U.S. 316, 324 (1997). That locution is not self-defining, and the Justices have been at least mildly schizophrenic in mapping its contours. The Court initially glossed the phrase by portraying the scope of ERISA preemption as "deliberately expansive." Pilot Life Ins. Co. v. Derdeaux, 481 U.S. 41, 46 (1987). As time passed, it grew more guarded, emphasizing the "starting presumption that Congress does not intend to supplant state law," New York State Conf. of Blue Cross Blue Shield Plans v. Travelers Ins. Co., 514 U.S. 645, 654 (1995); accord De Buono v. NYSA-ILA Med. Clin. Servs. Fund, 520 U.S. 806, 813 (1997), and warning that, unless congressional intent to preempt clearly appears, ERISA will not be deemed to supplant state law in areas traditionally regulated by the states,see Dillingham, 519 U.S. at 325; Travelers, 514 U.S. at 655.
Importantly, these variations in emphasis have led the Court to conclude in recent years that the phrase "relate to," as used in ERISA's preemption provision, cannot be read literally. "If `relate to' were taken to extend to the furthest stretch of its indeterminacy, then for all practical purposes preemption would never run its course. . . ." Travelers, 514 U.S. at 655. To scale the phrase down to size, the Court has devised a disjunctive test: "A law relate[s] to a covered employee benefit plan for purposes of § 514(a) if it  has a connection with or  a reference to such a plan." Dillingham, 519 U.S. at 324 (citations and internal quotation marks omitted) (alterations in original). We apply this test to the Massachusetts bond statute, mindful that a state law which comes within the compass of either branch of the test is subject to preemption.
A. Connection .
This aspect of the case does not require us to consider USF G'sstare decisis argument. Our decision in Williams focused solely on the "reference to" furculum of ERISA preemption analysis, see 45 F.3d at 591, and we will discuss it under that rubric. See infra Part II(B).
For the same reasons that infinite relations cannot be the measure of pre-emption, neither can infinite connections. We simply must go beyond the unhelpful text and the frustrating difficulty of defining its key term, and look instead to the objectives of the ERISA statute as a guide to the scope of the state law that Congress understood would survive.
Cataloguing the objectives of the ERISA statute is a fairly straightforward exercise. When Congress conceived the ERISA scheme, it made manifest its intention to "protect . . . the interests of participants in employee benefit plans and their beneficiaries . . . by establishing standards of conduct, responsibility, and obligation for fiduciaries of employee benefit plans, and by providing for appropriate remedies." 29 U.S.C. § 1001(b). Achieving this end requires the avoidance of "a multiplicity of regulation" and, concomitantly, the creation of a climate that "permit[s] the nationally uniform administration of employee benefit plans." Travelers, 514 U.S. at 657. Using this template, the Massachusetts bond statute, on its face, in no way inhibits the accomplishment of ERISA's overall goals.
It is well accepted, however, even under the new regime, that state laws which furnish alternative enforcement mechanisms threaten the uniformity that Congress labored to achieve and thus are preempted by ERISA. See id.; see also Ingersoll-Rand Co. v.McClendon, 498 U.S. 133, 142-45 (1990). This category comes to mind because the most obvious tie between the Massachusetts bond statute and ERISA plans as a class concerns the former's role as a vehicle of enforcing funding obligations. The question, then, is whether the bond statute impermissibly supplies an alternative enforcement mechanism for ERISA plan benefits (and thereby triggers preemption). See Travelers, 514 U.S. at 658; Parrino, 146 F.3d at 705; see also Turner v. Fallon Community Health Plan, Inc., 127 F.3d 196, 199 (1st Cir. 1997).
In Ingersoll-Rand, the Supreme Court recognized that a state law can be preempted as an alternative enforcement mechanism to ERISA § 502(a). See Ingersoll-Rand v. McClendon, 498 U.S. 133, 142-45 (1990). In Travelers and its progeny, the Court has yet to state unequivocally where § 502(a) preemption should be placed within the reshaped ERISA doctrine. The Second Circuit, however, has treated the alternative enforcement mechanism inquiry as a separate branch of ERISA preemption analysis. See Howell, 126 F.3d at 68. This is by no means inevitable. Conceivably, alternative enforcement mechanism preemption can be considered as part and parcel of a "connection" or "reference" analysis, or fit under § 514(a) as a category like "connection" and "reference," or (asHowell suggests) constitute a preemption prong independent of § 514(a). In this case, it makes no difference under which heading we place the inquiry, and so, while we discuss this subject under the "connection" rubric for the sake of convenience, we leave the question open.
We answer that question in the negative. ERISA preemption proscribes the type of alternative enforcement mechanism that purposes to provide a remedy for the violation of a right expressly guaranteed and exclusively enforced by the ERISA statute. See Ingersoll-Rand, 498 U.S. at 145. Those state laws which touch upon enforcement but have no real bearing on the intricate web of relationships among the principal players in the ERISA scenario (e.g., the plan, the administrators, the fiduciaries, the beneficiaries, and the employer) are not subject to preemption on this basis. See Woodworker's Supply, Inc. v.Principal Mut. Life Ins. Co., 170 F.3d 985, 990 (10th Cir. 1999). It follows that a state statute which only creates claims against a surety does not constitute an impermissible alternative enforcement mechanism as that term is used in ERISA jurisprudence.See Trustees for Mich. Laborers' Health Care Fund v. Seaboard Sur. Co., 137 F.3d 427, 429 (6th Cir. 1998); Bleiler v. Cristwood Constr., Inc., 72 F.3d 13, 15 (2d Cir. 1995).
That ends this aspect of the matter. The Massachusetts bond statute does not constitute a proscribed alternate enforcement mechanism. By the same token, it has no other meaningful nexus with ERISA; it does not, for example, interfere with the administration of covered employee benefit plans, purport to regulate plan benefits, or impose additional reporting requirements. Last — but far from least — it regulates an area of the law traditionally thought to be the states' preserve: enforcing contracts under state law for the citizenry's protection. See Operating Eng'rs Health Welfare Trust Fund v.JWJ Contracting Co., 135 F.3d 671, 678 (9th Cir. 1998); Romney v.Lin, 105 F.3d 806, 811 (2d Cir. 1997). Consequently, we conclude that the Massachusetts bond statute does not have a sufficient "connection with" covered employee benefit plans to warrant ERISA preemption.
B. Reference .
Discerning no impermissible connection, we turn to the second branch of the ERISA preemption analysis and ask whether the Massachusetts bond statute refers to covered employee benefit plans so directly as to justify preemption. Prior to the Supreme Court's decision in Travelers, this court answered that query affirmatively, holding that the bond statute singled out ERISA plans and was therefore preempted. See Williams, 45 F.3d at 591. In USF G's view, that ought to be the end of the matter. Because special circumstances obtain here, we disagree.
We do not gainsay that the principle of stare decisis forms an integral part of our system of justice. Withal, that system is not only precedent-based but also hierarchical. When emergent Supreme Court case law calls into question a prior opinion of another court, that court should pause to consider its likely significance before giving effect to an earlier decision. See, e.g., Odum v.Boone, 62 F.3d 327, 332 n. 2 (10th Cir. 1995); Pritzker v. Merrill Lynch, Pierce, Fenner Smith, Inc., 7 F.3d 1110, 1115 (3d Cir. 1993). Indeed, Williams itself acknowledged that there would be occasions, albeit "relatively rare," on which a newly constituted panel should eschew prior circuit precedent in deference to intervening authority. 45 F.3d at 592. Our task, then, is to determine whether Williams, though not directly overruled or superseded, fairly can be said to have fallen by the wayside. We conclude that it has.
Let us be perfectly clear. We value finality, stability, and certainty in the law, particularly in the field of statutory construction. See Hubbard v. United States, 514 U.S. 695, 711 (1995). But stare decisis is neither a straightjacket nor an immutable rule; it leaves room for courts to balance their respect for precedent against insights gleaned from new developments, and to make informed judgments as to whether earlier decisions retain preclusive force. See United States v. Connor, 926 F.2d 81, 83 (1st Cir. 1991). We explain below why we believe that this case represents one of those uncommon instances in which intervening developments in the law make reconsideration appropriate.
The place to begin such an odyssey normally would be with Williams itself. Here, however, we retreat further into the past, cognizant that Williams relied heavily on an earlier precedent, McCoy v.Massachusetts Institute of Technology, 950 F.2d 13 (1st Cir. 1991). Given that symbiosis, gaining perspective on Williams necessitates an appreciation of McCoy.
In McCoy, we ruled that ERISA preempted the operation of a Massachusetts mechanics' lien statute, Mass. Gen. Laws ch. 254, because the language of the statute "expressly single[d] out ERISA plans for special treatment." McCoy, 950 F.2d at 19. We noted that the mechanics' lien statute, by its terms, inured to the advantage of "the trustee or trustees of any fund or funds, established pursuant to section 302 of the Taft Hartley Law ( 29 U.S.C. § 186), providing coverage or benefits to [an employee]." Id. (quoting statute). On that basis, we concluded that:
[A]ny plan that grants benefits under 29 U.S.C. § 186, which is another way of describing any plan that grants benefits under section 302 of the Taft-Hartley Act, is by definition an ERISA plan. Put bluntly, by singling out "section 302" plans for special treatment, the Massachusetts mechanics' lien law, in the same stroke, singles out ERISA plans for special treatment. It is, therefore, preempted as it applies to ERISA-regulated plans.
Id. at 19-20. We attributed this conclusion in large part to what we termed "considered dictum," id. at 19, appearing in Mackey v.Lanier Collection Agency, 486 U.S. 825, 838 n. 12 (1988) (declaring that "any state law which singles out ERISA plans, by express reference, for special treatment is pre-empted"). McCoy set the stage for Williams. There, we harkened back to McCoy and depicted the mechanics' lien statute and the bond statute as "sisters under the skin." Williams, 45 F.3d at 592. Because we "perceive[d] no rational basis on which to distinguish between [them] for the purpose of gauging ERISA's preemptive reach," we concluded that ERISA preempted the latter statute. Id. at 591. It is this holding that we reexamine today.
Footnote 12 of Mackey (and, thus, the McCoy rationale) likely endures. See Dillingham, 519 U.S. at 324. The mechanics' lien statute does not; the Massachusetts legislature amended it in 1996 to eliminate any mention of "trustees" and "section 302 of the Taft Hartley Act." See Mass. Gen. Laws ch. 254, § 1 (Supp. 2000).
The key precedents are the Supreme Court's subsequent opinions inTravelers and Dillingham. The extent to which the analytical shift chronicled in these opinions applies to the "reference to" aspect of the ERISA preemption inquiry is less than obvious. TheTravelers Court quickly ruled out any possibility that the state law it was called upon to consider made reference to an ERISA plan and refined the ERISA preemption analysis in the context of the "connection with" inquiry. See Travelers, 514 U.S. at 656. TheDillingham Court recounted the reasoning of Travelers without explicitly limiting that reasoning to the "connection with" inquiry, but its actual treatment of the "reference to" question relied entirely on pre-Travelers precedent and made only passing mention of ERISA's objectives. See Dillingham, 519 U.S. at 325-28. Thus, both opinions stop short of explicitly endorsing a new analytic modality for the "reference to" inquiry. See Prudential Ins. Co. v. National Park Med. Ctr., 154 F.3d 812, 820-22 (8th Cir. 1998).
USF G contends that, absent an outright endorsement, we should disregard the reasoning of Travelers and Dillingham in pursuing the "reference to" question. We think not. Although the Travelers Court had no occasion to link its newly conceived "objectives" analysis to the "reference to" inquiry, the two building blocks on which that analysis rests — the starting presumption that Congress did not intend to supplant state law and the requirement that no preemption be deemed to occur in areas of traditional state regulation except in accord with the clear and manifest purpose of Congress — logically undergird both inquiries. See Travelers, 514 U.S. at 654-56 (dealing with baseline presumptions before beginning its bifurcated "connection with" and "reference to" analyses). We thus proceed to apply the teachings of Travelers and Dillingham as we understand them.
To begin with, Dillingham makes clear that two types of state laws — those that impose requirements by reference to ERISA plans and those that specifically exempt ERISA plans from otherwise generally applicable provisions — as well as state causes of action that are predicated on the existence of ERISA plans all refer to, and thus relate to, ERISA plans for purposes of 29 U.S.C. § 1144(a). See Dillingham, 519 U.S. at 324-25. Put another way, in the post-Travelers era the "reference to" inquiry will result in preemption "[w]here a State's law acts immediately and exclusively upon ERISA plans . . . or where the existence of ERISA plans is essential to the law's operation." Id. at 325 (citations omitted).
Examining Williams through the prism of Travelers and Dillingham, a rational distinction between the mechanics' lien statute and the bond statute, not previously thought to be important, bubbles to the surface: unlike the mechanics' lien statute, the bond statute makes no direct reference to section 302. Instead, it states that bonds for public works projects shall cover, in addition to labor and materials, any sums due trustees or other persons authorized to collect such payments from the contractor or subcontractors, based upon the labor performed or furnished as aforesaid, for health and welfare plans, supplementary unemployment benefit plans and other fringe benefits which are payable in cash and provided for in collective bargaining agreements between organized labor and the contractor or subcontractors. . . .
Mass. Gen. Laws ch. 149, § 29. The language concerning trustees is not ERISA-specific, but stands at the end of a long list of items that contractors are required to secure (e.g., amounts due for labor performed, materials furnished, transportation, equipment rental, and the like). The diversity of this list is telling (especially since most of these items have nothing whatever to do with ERISA). Furthermore, the bond statute treats contributions to fringe benefit plans in exactly the same manner as it treats the other (non-ERISA-related) elements that fall within the statutory sweep. Given the Dillingham screen, this combination of factors stretches any inference that the bond statute singles out ERISA plans for special treatment past the breaking point. See Seaboard Sur., 137 F.3d at 429. Dillingham confirms in another way that a reviewing court should differentiate between the mechanics' lien statute and the bond statute for purposes of the "reference to" inquiry. The Supreme Court's approach there reflects a conservative view of the inquiry, suggesting that a reference must be patent before ERISA preemption looms. See Dillingham, 519 U.S. at 324-25. The precedents that Dillingham cites, see id., illustrate this point. In each of those cases, an ERISA plan lay at the center of the inquiry. In contrast, the bond statute functions irrespective of the existence or non-existence of an ERISA plan.
Temporal considerations bolster this conclusion. Massachusetts adopted the bond statute in 1957 — a quarter-century before Congress enacted ERISA. This chronology undercuts any inference that the drafters of the statute intended to target ERISA plans.See JWJ Contracting, 135 F.3d at 679.
The Court listed three examples. (1) The reference in Mackey was too explicit to be construed any other way: the statute at issue there targeted "[f]unds or benefits of a pension, retirement, or employee benefit plan or program subject to the provisions of the federal Employee Retirement Income Security Act. . . ."Mackey, 486 U.S. at 828 n. 2 (quoting Ga. Code Ann. § 18-4-22.1). (2) In Ingersoll-Rand, the plaintiff's cause of action was based on an allegation that his employer discharged him to avoid making contributions to his pension fund, and, thus, "in order to prevail, [the] plaintiff must plead, and the court must find, that an ERISA plan exists and the employer had a pension-defeating motive in terminating the employment."Ingersoll-Rand, 498 U.S. at 140. (3) So too District of Columbia v. Greater Washington Board of Trade, 506 U.S. 125 (1992), in which the Court held preempted a state statute that required employers to provide health insurance coverage for eligible employees "at the same benefit level" as that already provided by existing ERISA plans. Id. at 130.
The sockdolager is that emergent Supreme Court precedent, by disavowing a strictly textual approach to the interpretation of ERISA's preemption provision, encourages us for the first time to conduct the "reference to" inquiry in light of the actual operation of the challenged state statute. See De Buono, 520 U.S. at 815; Dillingham, 519 U.S. at 324-25. Here, that glimpse is revealing. The bond statute neither imposes requirements on ERISA plans nor exempts such plans from otherwise applicable statutory provisions. In operation, therefore, the statute comports fully with ERISA's objectives. Furthermore, it does not dictate the form that a covered plan may take, specify the mode or manner of plan administration, or jeopardize the sort of uniformity that Congress aspired to achieve. Given these facts, the statutory reference to the term "trustees" seems too tenuous to trigger ERISA preemption.See Howell, 126 F.3d at 68. Indeed, the case at hand is, in this respect, reminiscent of Dillingham, in which the Supreme Court held that a statutory mention of an apprenticeship program was not sufficient to cause preemption under the "reference to" prong. See 519 U.S. at 325.
USF G's argument that the bond statute singles out ERISA plans because it is limited to public (as opposed to private) construction contracts lacks force. It is common ground that state laws of general application are safe from ERISA preemption even if they impose some incidental burdens on the administration of covered plans. See Washington Physicians Serv. Ass'n v. Gregoire, 147 F.3d 1039, 1043 (9th Cir. 1998), cert. denied, 525 U.S. 1141 (1999); Howell, 126 F.3d at 67; cf. Travelers, 514 U.S. at 659-60 (explaining that the fact that a law has an indirect economic influence on ERISA plans does not, in and of itself, justify preemption). USF G's argument amounts to a claim that the bond statute is something other than a law of general application.
In our view, the concept of "general application" cannot be parsed that closely. A state law that applies to a wide variety of situations, including an appreciable number that have no specific linkage to ERISA plans, constitutes a law of general application for purposes of 29 U.S.C. § 1144(a). See, e.g., Mackey, 486 U.S. at 838 n. 12; Shea v. Esensten, 208 F.3d 712, 717 (8th Cir. 2000);Arizona State Carpenters Pension Trust Fund v. Citibank (Ariz.), 125 F.3d 715, 724 (9th Cir. 1997). Under this definition, the bond statute ranks as a law of general application for ERISA preemption purposes because it applies to a sufficiently broad, sufficiently generalized universe of situations — all Massachusetts public works projects — without mentioning ERISA and without regard to whether any affected person is (or is not) involved with a covered plan. See De Buono, 520 U.S. at 815; Howell, 126 F.3d at 67-68.
To sum up, the bond statute, gauged by the principles embodied in recent Supreme Court case law, neither singles out ERISA plans for special treatment nor depends on their existence as an essential part of its operation. Rather, the statute is "indifferent to . . . ERISA coverage." Dillingham, 519 U.S. at 328. It is properly classified, therefore, as "one of `myriad state laws' of general applicability that impose some burdens on the administration of ERISA plans but nevertheless do not `relate to' them within the meaning of the governing statute." De Buono, 520 U.S. at 815. Thus, it does not trigger preemption. See Travelers, 514 U.S. at 656; JWJ Contracting, 135 F.3d at 679; Howell, 126 F.3d at 68.
We need go no further. The issue here is whether ERISA preempts the appellants' state-law cause of action. Believing, as we do, that Williams no longer aids us in our consideration of this issue, we abrogate its central holding. See supra note 1 accompanying text.
Taking a fresh look at the Massachusetts bond statute and giving due weight to Travelers and its progeny, we conclude that USF G has not overcome the starting presumption against preemption. Accordingly, the bond statute does not "relate to" any covered employee benefit plan within the meaning of 29 U.S.C. § 1144(a). It follows that the entry of judgment on the pleadings must be reversed and the case remanded for further proceedings consistent with this opinion.
On remand, the lower court should consider whether any other basis for federal jurisdiction exists (and if it discerns none, should restore the case to a state forum).