dissenting.
This appeal, consisting of three consolidated actions, principally concerns two separate preemption issues: express preemption under § 514(a) of the Employee Retirement Income Security Act of 1974, 29 U.S.C. § 1144(a) (“ERISA”), and complete preemption under § 502(a) of ERISA, 29 U.S.C. § 1132. While I join Part III of the majority opinion because I agree that § 502(a) complete preemption exists, thereby establishing federal subject-matter jurisdiction,13 I must respectfully dissent from Parts IV and V of the majority opinion because, in my view, the New Jersey collateral source statute, N.J.S.A. 2A:15-97,14 is saved from express preemption under § 514(a) of ERISA as a state regulation of insurance.
I.
Three provisions of ERISA § 514 speak directly to the question of express preemption,15 the mechanics of which have been neatly summarized by the Supreme Court:
If a state law “relatefs] to ... employee benefit plants],” it is pre-empted. § 514(a). The saving clause excepts from the pre-emption clause laws that “regulat[e] insurance.” § 514(b)(2)(A). The deemer clause makes clear that a state law that “purport[s] to regulate insurance” cannot deem an employee benefit plan to be an insurance company. § 514(b)(2)(B).
Pilot Life Ins. Co. v. Dedeaux, 481 U.S. 41, 45, 107 S.Ct. 1549, 95 L.Ed.2d 39 (1987). *168Although the express preemption analysis normally requires the application of the three relevant provisions of § 514, the sole issue in this appeal is whether the District Court erred as to the second step in the analysis in finding that N.J.S.A. 2A: 15-97 is “saved” from preemption as a state statute that regulates insurance under § 514(b)(2)(A).16 Accordingly, I confine my discussion to that narrow issue.
II.
In Kentucky Ass’n of Health Plans, Inc. v. Miller, 538 U.S. 329, 341-42, 123 S.Ct. 1471, 155 L.Ed.2d 468 (2003), the Supreme Court rejected the previous use of the McCarran-Ferguson factors, and instead enunciated two requirements for a state law to be deemed a “law ... which regulates insurance” under § 514(b)(2)(A). First, the state law must “be specifically directed toward entities engaged in insurance.” Id. at 342, 123 S.Ct. 1471 (citing Pilot Life, 481 U.S. at 50, 107 S.Ct. 1549; UNUM Life Ins. Co. of Am. v. Ward, 526 U.S. 358, 368, 119 S.Ct. 1380, 143 L.Ed.2d 462 (1999); Rush Prudential HMO, Inc. v. Moran, 536 U.S. 355, 366, 122 S.Ct. 2151, 153 L.Ed.2d 375 (2002)). Second, the state law must “substantially affect the risk pooling arrangement between the insurer and the insured.” Id.
Here, there is no serious dispute that state antisubrogation laws spread policyholder risk and therefore satisfy the second Miller requirement.17 See Singh, 335 F.3d at 286 (noting that “it is difficult to imagine an antisubrogation law ... as anything other than an insurance regulation, as it addresses who pays in a given set of circumstances and is therefore directed at spreading policyholder risk”); Med. Mut. of Ohio v. deSoto, 245 F.3d 561, 574 (6th Cir.2001) (“The logical effect of [antisubro-gation laws] ... is to decrease the premiums of health care providers’ insurance and increase the premiums of health insurance — i.e., spread risks.”). What concerns us in this appeal, then, is whether the New Jersey collateral source statute satisfies the first Miller requirement, i.e., that it is specifically directed towards the insurance industry.
Focusing solely upon the statutory language, the majority concludes that the New Jersey collateral source statute is not specifically directed to the insurance industry because its definitions sweep too broadly and thereby encompass organizations or entities that do not provide insurance. To be sure, the collateral source statute does not specifically refer to health insurance or to subrogation and reimbursement clauses. It is contained in Title 2A of the New Jersey Statutes, which regulates the administration of civil and criminal justice. Moreover, the statute applies to plaintiffs “in any civil action” who receive benefits for their injuries from “any other source other than a joint tortfeasor.” N.J.S.A. 2A:15-97. As the majority opinion notes, the term “benefits received” thus encompasses more than just insurance proceeds.
On the surface, therefore, this case would appear to present a paradigmatic example of a law of general application that has some bearing on insurers. Such laws do not qualify under the “saving” clause jurisprudence. See Miller, 538 U.S. at 334, 123 S.Ct. 1471. However, the inquiry does not end here, for the New Jersey Supreme Court has spoken in a rather definitive way as to the legislative *169purpose of the collateral source statute. See Perreira v. Rediger, 169 N.J. 399, 778 A.2d 429 (2001).
While recognizing that “[o]n its face, N.J.S.A. 2A:15-97 ... is silent regarding any right to subrogation or reimbursement on the part of health insurers,” id. at 409, 778 A.2d 429, the Supreme Court in Per-reira determined that the statute had more than one purpose: “To be sure, its primary purpose was to disallow double recovery to plaintiffs, but a secondary goal was clearly the containment of spiraling insurance costs. ” Id. at 410, 778 A.2d 429 (emphasis added). In enacting N.J.S.A. 2A15-97, the Court noted that the legislature made a “separate legislative decision” as to which “segment of the insurance industry” would be the beneficiary of the double recovery disallowance. Id. As the legislative history reveals, the Court noted, the choice was made to favor liability carriers.18 Id. at 411, 778 A.2d 429 (citing cases agreeing that purpose of statute was to shift burden from casualty and liability insurance industry).
III.
I am persuaded that the foregoing statutory interpretation, coming, as it does, from the State’s highest tribunal, compels the conclusion that the New Jersey collateral source statute is “specifically directed” towards the insurance industry. In my view, the majority opinion accords too little weight to such statements from the New Jersey Supreme Court, focusing instead on the admittedly broad statutory language. Our difference, then, is mostly an hermeneutical one, centering on the interpretive import of Perreira in ascertaining the aim of the statute.
In assigning minimal value to Perreira, the majority opinion states that the mere fact that the New Jersey statute has an impact on insurance, as settled in Perreira, is not enough to satisfy the “specifically directed” requirement of the saving clause. Even if, the majority argues, the New Jersey Supreme Court has identified N.J.S.A. 2A15-97 with the insurance industry, that does not change the actual terms of the statute. See Pilot Life, 481 U.S. at 50, 107 S.Ct. 1549 (holding common law of bad faith not saved from preemption “[e]ven though the Mississippi Supreme Court has identified its law of bad faith with the insurance industry”). According to the majority opinion, the collateral source statute thus resembles the Mississippi law at issue in Pilot Life. This analogy, however, misses the critical distinction between the two provisions.
Under the relevant state law in Pilot Life, punitive damages could be sought for “bad faith” in denying claims without any reasonably arguable basis for the refusal to pay. 481 U.S. at 50, 107 S.Ct. 1549. The Supreme Court determined that although Mississippi had “identified its law of bad faith with the insurance industry, the roots of this law are firmly planted in the general principles of Mississippi tort and contract law.” Id. “Any breach of contract,” the Court observed, “and not merely breach of an insurance contract, may lead to liability for punitive damages under [the Mississippi common law of bad faith].” Id. Accordingly, the Court con-*170eluded that the Mississippi law did not “regulat[e] insurance” within the meaning of ERISA’s saving clause. Id.
The holding in Pilot Life was premised upon the finding that “the roots of [the common law of bad faith were] firmly planted in the general principles of ... tort and contract law.” Id. at 50. Pilot Life, contrary to the majority’s reading, did not involve a situation where “a state supreme court described the law as one intended to affect the insurance industry.” Majority Op. at 165. The Mississippi Supreme Court never stated that its common law of bad faith was specifically directed towards the insurance industry; it merely applied that longstanding law to the insurance context. Pilot Life, 481 U.S. at 49-50, 107 S.Ct. 1549. Under these circumstances, the Supreme Court quite properly held that “a common-sense understanding of the phrase ‘regulates insurance’ does not support the argument that the Mississippi law of bad faith falls under the saving clause.” Id. at 50, 107 S.Ct. 1549.
Common sense dictates otherwise here. This case involves a statutory enactment, which, according to the New Jersey Supreme Court, was clearly rooted in legislative concerns about spiraling insurance costs. The New Jersey Supreme Court was emphatic in emphasizing insurance in its opinion:
The effectuation of no-double-recovery [by N.J.S.A. 2A:15-97] therefore required a separate legislative decision regarding which segment of the insurance industry would be the beneficiary of that disallowance. The Legislature had two choices: to benefit health insurers by allowing repayment of costs expended on a tort plaintiff, or to benefit liability carriers by reducing the tort judgment by the amount of health care benefits received. As the legislative history reveals, the choice was made to favor liability carriers. See Kiss v. Jacob, 138 N.J. 278, 282 [650 A.2d 336] (1994) (stating that intent of the legislature was to control spiraling automobile-insurance costs); Fayer v. Keene Corp., 311 N.J.Super. 200, 208, 709 A.2d 808 (App.Div.1998) (agreeing that purpose of statute is to shift burden to health industry); Parker v. Esposito, 291 N.J.Super. 560, 565, 677 A.2d 1159 (1996) (stating that purpose of collateral source statute is to prevent double recovery thereby giving relief from increasing costs of liability insurance); Lusby v. Hitchner, 273 N.J.Super. 578, 591, 642 A.2d 1055 (App.Div.1994) (stating that legislative determination “was apparently not only to prevent plaintiffs from obtaining a double recovery but also, except where PIP payments are involved, to shift the burden, at least to some extent, from the liability and casualty insurance industry to health and disability third-party payers”).
Perreira, 169 N.J. at 410-11, 778 A.2d 429.
While the Supreme Court has held that “laws of general application that have some bearing on insurers do not qualify,” Miller, 538 U.S. at 334, 123 S.Ct. 1471, the New Jersey collateral source statute presents the inverse proposition-it is a law specifically directed towards the insurance industry that has some bearing on non-insurers. As such, it “homes [sic] in on the insurance industry and does ‘not just have an impact on [that] industry’.” Ward, 526 U.S. at 368, 119 S.Ct. 1380 (quoting Pilot Life, 481 U.S. at 50, 107 S.Ct. 1549). Because the New Jersey statute had its genesis in specific legislative action, as opposed to general principles of tort or contract law, the majority opinion’s reliance on Pilot Life is entirely misplaced.
For these reasons, I believe that this case more closely resembles FMC Corp. v. Holliday, 498 U.S. 52, 111 S.Ct. 403, 112 *171L.Ed.2d 356 (1990), where the Supreme Court dealt precisely with the question of whether a state antisubrogation law19 was saved from preemption under § 514(b)(2)(A). There, the Court held that:
There is no dispute that the Pennsylvania [antisubrogation] law falls within ERISA’s insurance saving clause .:. [The antisubrogation law] directly controls the terms of insurance contracts by invalidating any subrogation provisions that they contain. It does not merely have an impact on the insurance industry; it is aimed at it. This returns the matter of subrogation to state law. Unless the statute is excluded from the reach of the saving clause by virtue of the deemer clause, therefore, it is not pre-empted.
498 U.S. at 60-61, 111 S.Ct. 403 (citations omitted).
Likewise, the Sixth and Fourth Circuits reached the same conclusion in considering whether similar state antisubrogation laws regulated insurance. See Singh, 335 F.3d at 286 (holding that subrogation prohibition of the Maryland HMO Act is a state-law regulation of insurance); deSoto, 245 F.3d at 573 (holding that California’s anti-subrogation statute regulated insurance); see also Hampton Indus., Inc. v. Sparrow, 981 F.2d 726, 729-30 (4th Cir.1992) (noting that “limits on subrogation recoveries appear to be aimed at the insurance industry, and therefore would also appear to come within the scope of the saving clause”).
Contrary to the majority, I conclude that our understanding of the New Jersey collateral source statute must be informed by the New Jersey Supreme Court’s interpretation. In the interpretive light cast by the Supreme Court in Perreira, the New Jersey collateral source statute is, in all essential respects, similar to those statutes already held to regulate insurance by the United States Supreme Court (FMC Corp.) and our two sister courts of appeals 0Singh and de Soto).
Accordingly, I conclude that the New Jersey collateral source statute is saved from ERISA preemption.20
Because I would affirm the District Court’s holding that § 514 of ERISA does not preempt N.J.S.A. 2A:15-97, I respectfully dissent from Part IV of the majority opinion, and from Part V of the majority *172opinion, which directs the District Court to dismiss the Insureds’ claims.
. See generally Arana v. Ochsner Health Plan, 338 F.3d 433 (5th Cir.2003); Singh v. Prudential Health Care Plan, Inc., 335 F.3d 278 (4th Cir.2003).
. The statute provides:
In any civil action brought for personal injury or death, except actions brought pursuant to the provisions of P.L.1972, c. 70 (C. 39:6A-1 et seq.), if a plaintiff receives or is entitled to receive benefits for the injuries allegedly incurred from any other source other than a joint tortfeasor, the benefits, other than workers’ compensation benefits or the proceeds from a life insurance policy, shall be disclosed to the court and the amount thereof which duplicates any benefit contained in the award shall be deducted from any award recovered by the plaintiff, less any premium paid to an insurer directly by the plaintiff or by any member of the plaintiff’s family on behalf of the plaintiff for the policy period during which the benefits are payable. Any party to the action shall be permitted to introduce evidence regarding any of the matters described in this act.
N.J.S.A. 2A:15-97.
. Of the three provisions of § 514, which are set forth below, the first and third are not involved in this appeal. It is the second provision, § 514(b)(2)(A), which concerns the regulation of insurance, that this appeal focuses upon.
“Except as provided in subsection (b) of this section [the saving clause], the provisions of this subchapter and subchapter III of this chapter shall supersede any and all State laws insofar as they may now or hereafter relate to any employee benefit plan ....” § 514(a), as set forth in 29 U.S.C. § 1144(a) (pre-emption clause).
"Except as provided in subparagraph (B) [the deemer clause], nothing in this sub-chapter shall be construed to exempt or relieve any person from any law of any State which regulates insurance, banking, or securities.” § 514(b)(2)(A), as set forth in 29 U.S.C. § 1144(b)(2)(A) (saving clause). "Neither an employee benefit plan ... nor any trust established under such a plan, shall be deemed to be an insurance company or other insurer, bank, trust company, or investment company or to be engaged in the business of insurance or banking for purposes of any law of any State purporting to regulate insurance companies, insurance contracts, banks, trust companies, or investment companies.” § 514(b)(2)(B), 29 U.S.C. § 1144(b)(2)(B) (deemer clause).
. The two other clauses of § 514-the "preemption clause” and the "deemer clause” are not the subject of the certified questions in this appeal. See supra note 3.
. Indeed, United and Horizon do not attempt such an argument.
. In reviewing the legislative history, the Court placed particular emphasis on the Passed Bill Memo prepared by Governor's counsel:
This bill attempts to reduce the cost of liability insurance by reducing the likelihood of a 'double recovery’ in a liability award for items which were already compensated by insurance or by other 'collateral' sources other than a tortfeasor.
Id. at 410, 778 A.2d 429 (quoting Passed Bill Memo to Governor Thomas H. Kean (Dec. 7, 1987)).
. The relevant statute — Section 1720 of the Pennsylvania Motor Vehicle Financial Responsibility Law — "prohibits insurance providers from obtaining reimbursement payments from recoveries an insured receives from third parties in a motor vehicle accident.” Bill Gray Enters., Inc. Employee Health & Welfare Plan v. Gourley, 248 F.3d 206, 213 n. 4 (3d Cir.2001).
. Because I would hold that the New Jersey collateral source statute is saved from ERISA preemption, I would be obliged to reach the third certified question for interlocutory appeal. That question concerns whether Fer-reira, which held that the statutory collateral source rule prohibits health insurers from filing reimbursement or subrogation liens against individual settlements or recoveries from third-party tortfeasors, applies retroactively to the health insurance plans at issue in this appeal.
This is a difficult issue, and more than that, the resolution of it could be outcome determinative in this appeal. As a result, I believe that the proper course for this Court to take would be to certify the issue of retroactivity to the New Jersey Supreme Court. Under New Jersey Court Rule 2:12A-1., the New Jersey Supreme Court may answer such a question if "there is no controlling appellate decision, constitutional provision or statute in this State.” N.J. Ct. R. 2:12A-1. The use of certification "rests in the sound discretion of the federal courts.” Lehman Bros. v. Schein, 416 U.S. 386, 391, 94 S.Ct. 1741, 40 L.Ed.2d 215 (1974). Such discretion, in my judgment, would be warranted here.