McLAUGHLIN, Circuit Judge:
Connecticut enacted a statute, the Uncompensated Care Pool Act (“Act I”), which required those with insurance to subsidize medical care for the poor. Conn.Pub.Acts 91-2 & 92-16, as codified by Conn.Gen.Stat. § 19a-168 et seq., and as amended by Conn. PubAets 93^4 & 93-229. Under Act I, hospitals had to impose a surcharge on the bill of any patient with private health care insurance. The hospitals sent the surcharge to the state, which, after pooling it in an “uncompensated care pool,” returned it to the hospitals to cover the costs hospitals incurred for uncompensated and undercom-pensated care. This pass-through also enabled Connecticut to qualify for federal Medicaid matching funds.
A patient, her union, and the union’s self-funded Taft-Hartley Fund, which was also a self-insured employee benefits plan, challenged the statute in the United States District Court for the District of Connecticut (José A. Cabranes, then-Chief Judge). The plaintiffs argued that the Employee Retirement Income Security Act of 1974, as amended, 29 U.S.C. § 1001 et seq. (“ERISA”) preempted the Connecticut statute. They moved for summary judgment, and the district court granted the motion, enjoined the [1027]*1027enforcement of the statute, and ordered restitution. New England Health Care Employees Union Dist. 1199 v. Mount Sinai Hosp., 846 F.Supp. 190, 195-200 (D.Conn.1994).
The district court offered three bases for its decision:
(1) Act I imposed a substantial, albeit indirect economic impact on ERISA plans; (2) Act I substantially depended on ERISA plans; and (3) Act I referred to ERISA plans.
Id. at 194-98. For support, it relied on our decision in Travelers Insurance Co. v. Cuomo, 14 F.3d 708 (2d Cir.1993) [hereinafter, Travelers /]. Thereafter, the Supreme Court reversed Travelers I. New York State Conf. of Blue Cross & Blue Shield Plans v. Travelers Ins. Co., — U.S. -, 115 S.Ct. 1671, 131 L.Ed.2d 695 (1995) [hereinafter Travelers II].
On appeal, the Connecticut official charged with enforcing Act I argues that, in the wake of Travelers II, ERISA does not preempt it. Given our recent decisions construing Travelers II, we are compelled to agree. Accordingly, we reverse and remand with instructions to enter judgment for the defendants.
BACKGROUND
American hospitals have a notable history of providing medical care for those who cannot afford it. See Erik J. Olson, note, No Room at the Inn: A Snapshot of an American Emergency Room, 46 Stan.L.Rev. 449, 468 & nn. 104-05 (1994). Because Medicaid and Medicare do not pay for all costs of healthcare provided to the poor, hospitals traditionally overcharged their paying patients — those with private insurance — to cover the cost of uncompensated and undercom-pensated care. See Travelers II, — U.S. at -, 115 S.Ct. at 1679; Olson, supra at 468-70; see also Conn.Agencies Regs. § 19a-165q-5 et seq. (allowing Connecticut hospitals to bill costs incurred from uncompensated and undercompensated care as “overhead”); Conn.Agencies Regs. § 19a-167g-13 et seq. (same). The practice has its own euphemism: “cost-shifting.”
In 1991, Connecticut passed Act I, sanctioning the cost-shifting Connecticut hospitals had been doing for decades. Conn.Pub. Acts 91-2, as codified by Conn.Gen.Stat. § 19a-168 et seq. Act I allowed hospitals to impose up to a 30.7% surcharge on the bill of paying patients. Under the original version of Act I, the hospitals remitted the entire surcharge to the state. The state pooled this revenue in an “uncompensated care pool” (the “UCP”), administered by the Connecticut Commission on Hospitals and Health Care (the “Commission”). The Commission then returned the revenues to the hospitals to subsidize the costs of uncompensated and undercompensated care. This pass-through qualified Connecticut for approximately $150 million in federal Medicaid matching funds annually, which the state used to balance its budget (rather than to provide additional Medicaid benefits).
In 1992, Nina Milner, a member of New England Health Care Employees Union, District 1199, SEIU AFL-CIO (the “Union”), twice checked into Mount Sinai Hospital (the “Hospital”). The Union’s self-insured ERISA plan, the New England Health Care Employees Welfare Fund (the “Fund”), which also happened to be a self-funded Taft-Hartley fund, had to pay her hospital bills. Accordingly, Milner signed some forms authorizing the Hospital to bill the Fund directly. She remained liable, however, for whatever charges the Fund did not cover.
The Hospital sent a bill to the Fund, designating the surcharge as “uncompensated care assessments.” The Fund refused to pay the surcharge. Instead, it escrowed enough money for the surcharge, and paid the Hospital only for that portion of Milner’s bill representing the care she actually received. The Hospital, as required by Act I, applied the Fund’s payment to the surcharge first. It then tried to collect the balance of the bill from Milner herself.
In response, Milner (as the named plaintiff in a class action), the Union, and the Fund brought 42 U.S.C. § 1983 and ERISA claims in the United States District Court for the District of Connecticut against the Commission, the Chairman of the Commission, Ste[1028]*1028phen J. Bongard,1 and the Hospital. The plaintiffs argued that ERISA preempted Act I. They sought an injunction barring enforcement of Act I, and demanded restitution for the surcharges already paid to the Hospital.
The district court dismissed the suit against the Commission on Eleventh Amendment grounds, and allowed the Connecticut Hospital Association (“CHA”), a non-profit association of health care facilities (including the Hospital), to intervene as a defendant. Meanwhile, Connecticut amended Act I. Conn.Pub.Act 92-16. Patients with private health insurance (including ERISA plans) still paid roughly 30.5% in surcharges. But, a hospital now had to remit to the state only an 8.4% surcharge (a “sales tax”); the hospital kept the additional 22.1% surcharge (by adding it into the rates it charged for each service — a classic cost-shift). (Thereafter, Act I was amended to break up the sales tax into two components, both of which were remitted to the state. Conn.Pub.Acts 93-44 & 93-229.)
The parties, together with Connecticut’s acute care hospitals, entered into a consent order, charting the course of the litigation. The order provided that the disputed surcharge was “deemed to be equal to 30.5% of all hospital charges.” Under the order, the Fund was to escrow 8.4% of each hospital’s bill. The hospitals continued to receive directly the remaining 22.1%, as authorized by the amended Act I. If the plaintiffs won, the Fund was to get the escrowed money back, and the Hospital had to reimburse the plaintiffs for the 22.1% cost-shift surcharge.
The parties then moved for summary judgment. The district court granted the plaintiffs’ motion, and denied the defendants’. Relying on and extending our opinion in Travelers I, the court held that ERISA preempted Act I. New England Health Care, 846 F.Supp. at 199. The court reasoned that Act I “depended on” the existence of ERISA plans because: (1) roughly 70% of the UCP’s revenue came from ERISA plans; and (2) Act I expressly referred to ERISA. Id. at 195. In addition, it held that Act I had a substantial economic effect on ERISA plans by forcing them either to increase plan costs or reduce benefits. Id. at 196-97. Significantly, however, the court pointed to no proof that any ERISA plan’s costs actually increased because of Act I, rather than because of the upwardly spiralling cost of medical care generally.
Accordingly, the district court enjoined the Chairman from enforcing Act I. It also enjoined the Hospital (and other hospitals that were parties to the consent order) from including the combined 30.5% surcharge when billing patients (i.e., the 8.4% sales tax and the 22.1% cost-shift). The court also ordered the escrow agent to return to the Fund the moneys escrowed for the 8.4% sales tax. Finally, the court ordered the Hospital (and the other hospitals that were parties to the consent order) to reimburse the Fund and the class plaintiffs any money collected through the 22.1% cost-shift.
In the wake of the district court’s ruling, Connecticut repealed Act I, replacing it with a new version (“Act II”) containing many of the same features. See Conn.Pub.Acts 94-9. CHA, a defendant in this action, has since successfully challenged Act II on ERISA preemption grounds. See Connecticut Hosp. Ass’n v. Pogue, 870 F.Supp. 444 (D.Conn.1994), appeal pending, No. 94-9237 (2d Cir.1995).
Despite Act I’s repeal, the defendants appealed the district court’s decision. The Hospital and CHA did not contest the preemption issue, but contended that the injunction was too drastic. The Chairman, on the other hand, argued that, as a matter of law, ERISA did not preempt Act I and, alternatively, that material disputes concerning Act I’s impact on ERISA plans precluded summary judgment. Not surprisingly, the plaintiffs moved to dismiss the Chairman’s appeal [1029]*1029as moot, because Act I was no longer in effect. We reserved decision on the motion.
After oral argument in this Court, the Supreme Court issued its decision in Travelers II. Because it reconfigured the landscape for ERISA preemption, we requested letter briefs from the parties on the impact of Travelers II.
DISCUSSION
In his letter brief, the Chairman argues that he still has an interest in this appeal, even though Act I has been repealed, and that Travelers II leaves us no choice but to reverse the judgment of the district court. On the other hand, Milner, the Fund, and the Union continue to argue that, notwithstanding Travelers II, ERISA totally preempts Act I, and, if not, it preempts at least with respect to self-insured plans like the Fund.
I. Mootness
We cannot hear an appeal in the absence of a live case or controversy. See U.S. Const, art. Ill; Burke v. Barnes, 479 U.S. 361, 363, 107 S.Ct. 734, 736, 93 L.Ed.2d 732 (1987) (“Article III of the Constitution requires that there be a live case or controversy at the time that a federal court decides the case.”). If the appellant through its own voluntary action renders its appeal moot, we dismiss the appeal and leave the decision below intact. See Karcher v. May, 484 U.S. 72, 83, 108 S.Ct. 388, 396, 98 L.Ed.2d 327 (1987); Manufacturers Hanover Trust Co. v. Yanakas, 11 F.3d 381, 383 (2d Cir.1993). That said, so long as the appellant retains some interest in the case, so that a decision in its favor will inure to its benefit, its appeal is not moot. See Richmond, v. J.A. Croson Co., 488 U.S. 469, 478 n. 1, 109 S.Ct. 706, 713 n. 1, 102 L.Ed.2d 864 (1989) (expiration of set-aside law did not moot case where plaintiff argued that prior application of ordinance entitled it to damages); Firefighters Local Union No. 1784 v. Stotts, 467 U.S. 561, 568-72, 104 S.Ct. 2576, 2582-84, 81 L.Ed.2d 483 (1984).
Here, the Connecticut legislature repealed Act I, arguably mooting the Chairman’s appeal. The Chairman, however, contends that he still has a stake in the appeal. We agree. Because Connecticut hospitals were enjoined from collecting the surcharge, less money was fed into the uncompensated care pool, and Connecticut qualified for less Medicaid matching funds. Thus, Connecticut would be entitled to recoup over $1 million if we were to hold that Act I was not preempted and the surcharges had to be paid to the state. These collateral consequences give the Chairman “a concrete interest in the outcome of the litigation,” and ensure that his appeal is not moot. Firefighters, 467 U.S. at 571, 104 S.Ct. at 2584; see also Klein v. Califano, 586 F.2d 250, 253 (3d Cir.1978) (in banc) (though government’s appeal of prospective application of injunction was mooted by agency ruling, the appeal was not moot “as concerns [the government’s] intention to seek recoupment ... for past expenditures in compliance with what it asserts to be a wrongfully granted injunction”).
The plaintiffs question whether the Chairman has in fact suffered any collateral consequences. Their argument is that Act I required hospitals to apply to the surcharge the first dollars it received from a patient and, accordingly, that the Chairman never suffered a loss. But, this ignores the terms of the district court’s injunction. The court enjoined the Chairman from “enforcing [Act I] against [the] fund and/or members of the certified plaintiff class.” 846 F.Supp. at 199. The court also enjoined the Hospital from “including in hospital bills incurred by patients who are participants or beneficiaries of the Fund charges which exceed those authorized by the Commission acting in accordance with this ruling.” Id. at 199-200. Thus, under the injunction, which was not stayed pending appeal, the Hospital could not impose the surcharge on patients covered by ERISA plans, nor could the Chairman collect it.
Accordingly, we may still hear the Chairman’s appeal. The plaintiffs’ motion to dismiss it as moot is denied.
II. ERISA Preemption
Turning to the merits, we review a grant of summary judgment de novo. See Mays v. Mahoney, 23 F.3d 660, 662 (2d [1030]*1030Cir.1994) (Timbers, J.), cert. denied, 115 S.Ct. 2608 (1995). Because a motion for summary judgment searches the record, see Fed.R.Civ.P. 56(c), we may reverse the grant of summary judgment and order judgment for the non-moving party if we find undisputed support in the record entitling the non-moving party to judgment as a matter of law. See, e.g., Procter & Gamble Indep. Union v. Procter & Gamble Mfg. Co., 312 F.2d 181, 190 (2d Cir.1962) (where district court erroneously granted plaintiff summary judgment and “there was no triable issue of fact, ... summary judgment will be entered for defendant”), cert. denied, 374 U.S. 830, 83 S.Ct. 1872, 10 L.Ed.2d 1053 (1963); see also 28 U.S.C. § 2106 (reviewing court may “reverse any judgment ... of a court lawfully brought before it for review, and may remand the cause and direct the entry of such appropriate judgment ... as may be just”).
ERISA generally preempts any state law that “relate[s] to” an ERISA plan. 29 U.S.C. § 1144(a). A law “relates to” an ERISA plan (and thus triggers ERISA preemption) “‘if it has a connection with or reference to sueh a plan.’ ” District of Columbia v. Greater Wash. Bd. of Trade, — U.S. -, -, 113 S.Ct. 580, 583, 121 L.Ed.2d 513 (1992) (quoting Shaw v. Delta Air Lines, Inc., 463 U.S. 85, 97, 103 S.Ct. 2890, 2900, 77 L.Ed.2d 490 (1983)). Before Travelers II, we applied the “connection with” prong of ERISA preemption differently, depending on whether the law under attack “directly” or “indirectly” affected ERISA plans. See Travelers v. Pataki, 63 F.3d 89, 91-92, 94 (2d Cir.1995) (per curiam) [hereinafter, Travelers III].
As we explained in Travelers III, 63 F.3d at 94, when the law targeted the healthcare industry and directly affected a plan, our opinion in NYSA-ILA Medical & Clinical Services Fund v. Axelrod, 27 F.3d 823 (2d Cir.1994), vacated sub nom. Chassin v. NYSA-ILA Medical & Clinical Services Fund, — U.S. -, 115 S.Ct. 1819, 131 L.Ed.2d 742 (1995), controlled. Thus, if an ERISA plan ran a hospital, and a State taxed payments received by hospitals, that was a direct impact, and ERISA preempted the tax. This was true even if the tax affected the ERISA plan only minimally. Id. at 826-28. We reasoned that the tax directly depleted the plan’s assets, and thus “related to” the plan “in a very direct way.” Id. at 827. (In addition, dicta in NYSA-ILA suggested that ERISA preempted any law that targeted the healthcare industry, since ERISA plans operate in the healthcare industry. Id.)
On the other hand, when a state law affected an ERISA plan only indirectly, our opinion in Travelers I, 14 F.3d at 717-23, controlled, see Travelers III, 63 F.3d at 91-92, and ERISA preempted any state surcharge law that “purposely interfere[d] with the choices that ERISA plans make for health care coverage.” Travelers I, 14 F.3d at 719. Furthermore, we held that ERISA preempted any state law that “substantially increase[d] the cost to ERISA plans of providing beneficiaries with a given level of health care benefits.” Id. at 720 (emphasis added). We also held that ERISA preempted any state law that imposed surcharges that forced ERISA plans to either increase “plan costs or reduce plan benefits.” Id.
The Supreme Court, however, has reversed our decision in Travelers I, see Travelers II, — U.S. at-, 115 S.Ct. at 1673-83, and vacated our decision in NYSA-ILA, see Chassin v. NYSA-ILA Med. & Clinical Servs. Fund, — U.S. -, 115 S.Ct. 1819, 131 L.Ed.2d 742 (1995). We have discussed elsewhere the Court’s recasting of the law of ERISA preemption. See NYS Health Maint. Org. Conf. v. Curiale, 64 F.3d 794 (2d Cir.1995); Travelers III, 63 F.3d at 91-95. Suffice it to say that a law’s indirect economic effect on ERISA plans, in and of itself, generally will not trigger ERISA preemption. See Travelers II, — U.S. at -, 115 S.Ct. at 1679; NYS Health Maint. Org. Conf. 64 F.3d at 803-04. The Court, though, left open or did not directly address a number of issues now raised by the plaintiffs here. We address each in turn.
A. Self-Insured Plans
Travelers II did not address the New York “surcharge statute insofar as it applie[d] to self-insured plans.” Travelers II, — U.S. at [1031]*1031-n. 4, 115 S.Ct. at 1675 n. 4. The Fund argues that, because it is a self-insured ERISA plan, Travelers II is wholly inappo-site. We disagree.
We have already rejected the notion that self-insured plans deserve special treatment when determining if a state law “relates to,” and thus is preempted by, ERISA generally. See Travelers III, 63 F.3d at 93 (“[A] plan’s self-insured status matters only if a state law that should generally be preempted under [the “relates to” test] qualifies as a law regulating insurance. ... If a state law regulates insurance, it is generally saved from preemption, but this ‘saving’ cannot apply to a self-insured plan.”) (citations omitted). Thus, if a law does not “relate to” insured ERISA plans, it does not relate to a self-insured plan either, unless the cost of complying with it forces the plan “to adopt a certain scheme of substantive coverage,” Travelers II, — U.S. at -, 115 S.Ct. at 1683, or to buy insurance rather than self-insure, Travelers III, 63 F.3d at 93.
B. Substantial Economic Impact
Even if ERISA did generally accord self-insured plans special protection under its preemption clause, we would reverse the decision below. We are well aware that self-insured plans feel the bite of a surcharge statute more directly and more deeply than do plans that purchase insurance. But, there was no proof that Act I actually increased self-insured plans’ costs at all.
The defendants offered uncontested allegations that hospitals’ fees remained largely the same after Act I’s enactment (and, despite the surcharge, even decreased at some hospitals). And, the plaintiffs conceded that “[h]ospital rates ... have always included ‘cost-shifting’ to reimburse hospitals for uncompensated care ... and undercompensat-ed care.” (J.A at 256.) Before Act I, hospitals used to pass through their losses from nonpaying and underpaying patients to paying patients. Act I merely formalized this practice.
True, the district court noted that suburban hospitals treated fewer poor patients than did urban hospitals. It speculated that, without Act I, the Fund might “have reduced costs by negotiating group discounts for uncompensated care or by encouraging participants to choose hospitals with low uncompensated care costs [i.e., suburban hospitals].” New England Health Care, 846 F.Supp. at 196. But, these hypothetical costs cannot carry the day for self-insured plans.
In the first place, a Connecticut law unrelated to Act I forbade discounts. See Conn. Gen.Stat. § 19a-166. In addition, there was no evidence that patients like Milner would go to suburban hospitals. Furthermore, the district court cited no evidence that suburban hospital costs were in fact lower before Act I. And, even if suburban hospital costs were lower, there was no evidence that Act I substantially increased those costs by placing urban and suburban hospitals on the same footing.
As evidence of Act I’s economic impact, the district court also noted that the Fund had “reduce[d] benefits in part as a result of’ Act I. 846 F.Supp. at 196. We cannot agree. The Fund did adduce proof that it closed a looming budget shortfall by reducing benefits and making administrative changes, saving an estimated $3.5 million. But, there was no proof that Act I was the cause of the budget strain — indeed, the Fund believed that “the increased cost of actual hospital services undoubtedly contributed to [its] financial woes.” Fund’s letter brief at 9 n. 6. In any event, the benefits reductions were ordered on December 6, 1991, before Act I even went into effect. See 846 F.Supp. at 192 (Act I “was adopted on December 17, 1991”); Trustees of the New England Health Care Employees Welfare Fund (December 6, 1991) (Schein-man, Arb.) (ordering benefits reductions).
Finally, the plaintiffs argue that Act I’s purported economic impact, even if minimal, triggers ERISA preemption because it is direct. See NYSA-ILA 27 F.3d at 826-28. We disagree. As in Travelers III,
[t]he surcharge statute ... does not “directly” deplete assets, at least not as directly as did the statute in NYSA-ILA. That statute taxed payments made by patients to an ERISA plan’s hospital. NYSA-ILA 27 F.3d at 825-26. There was [1032]*1032no third party between the plan and the tax. Here, the statute raises costs to a third party — the patient — who is liable for the balance of any bill which an ERISA plan or insurance carrier refuses to pay. Granted, self-insured plans are one step closer to the surcharge than are insured plans — the latter will not see increased costs until their insurers raise premiums. But, self-insured plans are still not directly liable for the surcharge. In any case, in the wake of Travelers (and the Court’s decision to vacate NYSA-ILA), we question whether we should continue .to distinguish between laws that directly deplete ERISA plans’ assets and laws that only indirectly do so.
Travelers III, 63 F.3d at 94.
C. “Reference to” Preemption
Act I contains a “self-destruct” clause that expressly mentions ERISA:
In the event that the final result of litigation brought pursuant to the federal Employee Retirement Income Security Act of 1974 (ERISA) ... is that any class of charges for patient care services ... is exempted from the assessment, the operation of the pool shall be terminated....
Conn.Gen.Stat. § 19a-168m, as amended by Conn.Pub.Aets 93-44, § 17. Seizing upon this, the plaintiffs contend that the self-destruct clause is a sufficient “reference” to ERISA to invoke preemption. Again, we disagree.
True, the Supreme Court has repeatedly said that state laws referring to ERISA plans “relate to” ERISA and are thus preempted. See, e.g., Travelers II, — U.S. at -, 115 S.Ct. at 1677. Nevertheless, “the Supreme Court has never found a statute to be preempted simply because the word ERISA (or its equivalent) appears in the text.” NYS Health Maint. Org. Conf., 64 F.3d at 800. Thus, “a state law has a reference to ERISA if its text mentions or alludes to ERISA plans, and if the law affects ERISA plans in some manner.” Id. (emphasis added).
The self-destruct clause here does not single out ERISA plans for special treatment. Quite the contrary, it guarantees that Act I will apply to all private health insurance providers, or none at all. Contrast Ingersoll-Rand Co. v. McClendon, 498 U.S. 133, 140, 111 S.Ct. 478, 483, 112 L.Ed.2d 474 (1990) (ERISA preempts statute that creates cause of action specifically for employees covered by ERISA plans); Mackey v. Lanier Collection Agency & Serv., Inc., 486 U.S. 825, 829-30, 108 S.Ct. 2182, 2185, 100 L.Ed.2d 836 (1988) (ERISA preempts statute that singles out ERISA plans for special treatment). And, Act I does not impose structural requirements on ERISA plans. Contrast FMC Corp. v. Holliday, 498 U.S. 52, 60, 111 S.Ct. 403, 408, 112 L.Ed.2d 356 (1990) (ERISA preempts statute that, in effect, prohibits ERISA plans “from being structured in a manner requiring reimbursement in the event of recovery from a third party”).
The Fund argues that, because the surcharge ultimately subsidizes the medical costs of the poor, Act I imposes a substantive coverage requirement on the Fund and thus “affects” it. This argument has a certain surface appeal. See Greater Wash. Bd. of Trade, — U.S. at -, 113 S.Ct. at 582-85 (ERISA preempts statute that requires employers to, in effect, extend their ERISA plans to employees eligible for workers’ compensation benefits).
The statute in Greater Washington, however, required employers “to evaluate employee benefits under their existing ERISA plan and then use those figures to calculate the benefits due to disabled employees.” NYS Health Maint. Org. Conf. 64 F.3d at 800. In contrast, Act I simply imposes a flat surcharge. To pay it, the Fund need not calculate the cost of benefits for hypothetical members, nor must it extend the terms of its plan to those who ultimately benefit from the surcharge. Accordingly, Act I’s glancing reference to ERISA plans cannot, of its own weight, support preemption.
D. The Exclusive Benefit Rule
The Fund must comply with both ERISA and the Labor Management Relations Act, 29 U.S.C. § 141 et seq. (the “LMRA”). Each statute, in turn, mandates that the Fund be used exclusively for the benefit of participants and beneficiaries — the so-called “exclusive benefit rule.” See 29 U.S.C. § 1104(a) (ERISA); 29 U.S.C. [1033]*1033§ 186(c) (LMRA). The Fund argues that, by forcing it to subsidize the medical costs of the poor, Act I conflicts with both the LMRA and ERISA. We are not persuaded.
Like the Third Circuit, we believe an ERISA plan (and, by analogy, a Taft-Hartley fund) discharges its fiduciary duties “when it pays whatever portion of the price charged by the health care provider the plan has assumed.” United Wire, Metal & Mach. Health & Welfare Fund v. Morristown Mem. Hosp., 995 F.2d 1179, 1196 (3d Cir.), cert. denied, — U.S. -, 114 S.Ct. 382, 126 L.Ed.2d 332 (1993); see NYS Health Maint. Org. Conf. 64 F.3d at 799 n. 13 (“the plan administrator’s fiduciary duty is realized when she submits the required payments to the insurer pursuant to the benefits contract”). Thus, the Fund will not breach its fiduciary duty to its participants and beneficiaries if it pays hospital bills that include surcharges for uncompensated or undercom-pensated care.
To hold otherwise would ignore that “Congress envisioned state experiments with comprehensive hospital reimbursement regulation.” Travelers II, — U.S. at - n. 6, 115 S.Ct. at 1683 n. 6. More importantly, as the Travelers II Court held regarding New York’s surcharge statute, Congress “seems to [have] envision[ed] a system very much like the one [Connecticut] put in place.” Id. at -, 115 S.Ct. at 1682. To obtain federal matching funds under the Medicaid Act, a state must “take into account the situation of hospitals which serve a disproportionate number of low income patients with special needs.” 42 U.S.C. § 1396a(a)(13)(A). The state can pay for these patients (and thus obtain federal matching funds) by using either taxes of general applicability, 42 U.S.C. § 1396a(t), or healthcare-related taxes, 42 U.S.C. § 1396b(w)(3). What Congress gives with one hand, it usually does not take away with the other. See Rebaldo v. Cuomo, 749 F.2d 133, 139-40 (2d Cir.1984) (Van Graafeiland, J., concurring), cert. denied, 472 U.S. 1008, 105 S.Ct. 2702, 86 L.Ed.2d 718 (1985).
E. “Substantial Dependence” Preemption
Finally, the district court held that Act I depended on ERISA plans because they provided some 70% of the revenue for the UCP; without ERISA plans’ participation, Act I simply would not succeed. 846 F.Supp. at 195. The plaintiffs argue that this market share/dependence test for ERISA preemption is an obvious application of existing caselaw. We cannot agree.
Until the district court’s decision, no court had ever held that ERISA preempts a statute simply because the law’s success “depended” on funds derived from ERISA plans. Indeed, we think this approach stands ERISA preemption principles on their head. Under ERISA preemption’s “connection with” prong, courts have always looked to the impact a law has on ERISA plans, not vice versa. See, e.g., Metropolitan Life Ins. Co. v. Massachusetts, 471 U.S. 724, 739, 105 S.Ct. 2380, 2389, 85 L.Ed.2d 728 (1985).
Moreover, we believe Travelers II implicitly rejected a market share/dependence approach to ERISA preemption. After all, we noted in Travelers I that over 80% of New Yorkers relied on ERISA plans for their health care coverage. Travelers I, 14 F.3d at 711. And, New York’s surcharge statute depended on the participation of ERISA plans even more directly than did Connecticut’s, since New York enacted its statute so as to encourage ERISA plans to purchase insurance from the Blues. Id. at 712, 722. If ERISA plans were exempt from the New York surcharge statute, that statute could not achieve its aim of “level[ing] [the] playing field for the Blues in their competition with commercial insurers.” Id. at 712 (internal quotation marks omitted). Nonetheless, the Supreme Court held that ERISA did not preempt New York’s surcharge statute.
Finally, the market share/dependence rationale sounds uncomfortably like the dicta we proffered in NYSA-ILA: any regulation that targets the healthcare industry will necessarily affect healthcare plans. See NYSA-ILA, 27 F.3d at 827. Under such an approach, ERISA would preempt all state regulation of hospitals, a conclusion firmly and flatly rejected by the Travelers II Court. See Travelers II, — U.S. at-, 115 S.Ct. at 1680 (nothing in ERISA’s text or legislative history “indicates that Congress chose to displace general health care regulation, which historically has been a matter of local [1034]*1034concern”); id. at -, 115 S.Ct. at 1681-82. Since NYSA-ILA has been vacated in light of Travelers II, see NYSA-ILA, — U.S. at -, 115 S.Ct. at 1819, we do not give the market share/dependence argument much weight.
To sum up: the plaintiffs failed to show that Act I, which sanctioned the cost-shifting that Connecticut hospitals practiced for decades, so increased costs to ERISA plans that the plans had to restructure themselves, buy insurance rather than self-insure, or reduce benefits. In addition, Act I’s self-destruct clause does not trigger ERISA preemption. Finally, the plaintiffs’ “exclusive benefit rule” and market share/dependence arguments cannot be the basis for ERISA preemption.
CONCLUSION
The repeal of Act I has not mooted the Chairman’s appeal; we therefore deny the plaintiffs’ motion to dismiss.
Furthermore, we see no basis for holding that ERISA preempts Act I, even with respect to self-insured plans. Accordingly, we reverse the judgment of the district court and remand with instructions to enter judgment for the defendants, and to enter an appropriate order regarding any payments that should have been made under Act I.