dissenting.
I agree that the majority’s outcome is a reasonable application of New York choice of law principles. My concern is that the majority’s conclusion is not the only reasonable application of New York law. Because I think the courts of New York have not clearly stated whether the Neumeier rules override, in the loss-allocation context, the interest-balancing approach that is generally conducted in tort conflicts, or whether the Neumeier rules instead incorporate this general interest-balancing approach, I would certify the choice of law question to the New York Court of Appeals. I thus respectfully dissent from the majority’s opinion.
Neumeier Rule Three states that in a case in which the parties are domiciled in different states and the alleged tort occurred in a third state, the applicable law “is that of the jurisdiction where the accident happened unless it appears that ‘displacing [the] normally applicable rule will advance the relevant substantive law purposes’ of the jurisdictions involved.” Neumeier v. Kuehner, 31 N.Y.2d 121, 335 N.Y.S.2d 64, 286 N.E.2d 454, 458 (1972) (emphasis added) (alteration in original) (quoting Tooker v. Lopez, 24 N.Y.2d 569, 301 N.Y.S.2d 519, 249 N.E.2d 394, 404 (1969) (Fuld, C.J., concurring)). As New York is the situs of the alleged tort, I start with the presumption that New York charitable immunity law applies, and then consider whether displacing the normally applicable rule will advance the substantive law purposes of the jurisdictions involved.
*113As the majority notes, the leading case applying Neumeier Rule Three is Schultz v. Boy Scouts of America, Inc., 65 N.Y.2d 189, 491 N.Y.S.2d 90, 480 N.E.2d 679 (1985). Schultz involved tort claims against two defendants, the Boy Scouts of America and Franciscan Brothers. The plaintiffs and the Boy Scouts were domiciled in New Jersey, Franciscan Brothers was domiciled in Ohio, and the tort occurred primarily in New York. In evaluating the plaintiffs’ claim against Franciscan Brothers, the court decided that Franciscan Brothers had “met its burden of demonstrating that the law of New Jersey, rather than the law of New York, should govern plaintiffs’ action against it.” Id. at 687. The court noted that application of New Jersey charitable immunity law would further New Jersey’s interests in “enforcing the decision of its domiciliaries to accept the burdens as well as the benefits of that State’s loss-distribution tort rules,” and in “promoting the continuation and expansion of [Franciscan Brothers’] charitable activities in that State.” Id. As for New York’s interests, the court held that while applying New Jersey law “may not affirmatively advance the substantive law purposes of New York, it [would] not frustrate those interests because New York has no significant interest in applying its own law to this dispute.” Id.
Applying the analysis of Schultz to the instant case, I agree with the majority that because the locus jurisdiction in loss-allocation cases has only a minimal interest in determining the right of recovery, id. at 685, 687, New York’s substantive law purposes would at the least not be undermined by the application of New Jersey law. It is also obvious that displacing New York law in favor of New Jersey law would advance New Jersey’s substantive law purposes. However, it is unclear from Neu-meier and Schultz whether, before we displace New York law, we must also find that Connecticut’s interests would be advanced by the application of New Jersey law.
The lack of clarity regarding the role of Connecticut’s interests stems from the Schultz court’s silence concerning the interests of the third state involved in that case — Schultz simply considers the interests of New York and New Jersey. Id. at 687. The third state, Ohio (the domicile of Franciscan Brothers), had a limited charitable immunity statute that denied immunity in actions, such as the one at issue in that case, based on negligent hiring and supervision. Id. at 682. But it is far from clear whether Schultz declined to discuss Ohio’s interests because the court found those interests to be advanced (or at least not hindered) by the application of New Jersey law, or because it found them to be hindered but of insufficient weight to overcome New Jersey’s greater interest. The outcome in Schultz could be the result of either a direct, literal application of Neu-meier Rule Three, or of an indirect application of the rule through the incorporation of a more flexible interest-balancing approach. That is, the Schultz opinion could be read as concluding that Ohio’s substantive law purposes would be advanced (or at least not hindered) by the application of New Jersey charitable immunity law; or it could be read as concluding that on balance, New Jersey had the greater interest in seeing its law applied. I am not alone in finding New York law unclear on this point: “There is confusion in New York state and federal courts about the proper construction of ... Schultz and the ... Neumeier rules. Some courts have applied the Neumeier rules rigidly. Others have ignored them in favor of a pure governmental interest analysis.” Lee S. Kreindler et al., 16 N.Y. Practice Series, N.Y. Law of Torts § 18:36 (2002) (footnotes omitted) (citing cases).
*114By its terms, Neumeier Rule Three seems to require something other than (and different from) general interest balancing; I read Neumeier to require that the substantive law purposes of all relevant states must be advanced before we can displace the lex loci delicti. See Cook v. Goodhue, 842 F.Supp. 1509, 1511 (N.D.N.Y.1994) (“The exception to [Neu-meier'] Rule Three simply states that applying the law of a state other than that of the place of the injury will advance the relevant substantive law purposes. This means the purposes of all of the substantive laws relevant to the conflict, not just the laws of the state whose laws are [sought] to be applied. There is no ‘either’ in the rule. Otherwise, it would be very easy for a party to demonstrate that the application of a particular state’s law will advance the purposes of that law.”). The test under Neumeier is thus not which state among the three has the greatest interest in applying its law; rather, the test is whether each state’s interest would be advanced (or not hindered) by application of a law other than the lex loci de-licti.1 So long as Connecticut’s interests would be hindered by the application of New Jersey law, I read Neumeier to require that we not displace the lex loci delicti.
The majority’s analysis, with its overriding emphasis on the strength of New Jersey’s interests and the extent of the parties’ contacts in that state, comes much closer to a general interest balancing than to literal application of Neumeier Rule Three. The majority could be right in its interpretation of New York choice of law principles; the New York Court of Appeals might incorporate an interest-balancing approach into Neumeier Rule Three and decide that New Jersey has the greater interest in seeing its law applied. Cf. Padula v. Lilarn Props. Corp., 84 N.Y.2d 519, 620 N.Y.S.2d 310, 644 N.E.2d 1001, 1002 (1994) (“In the context of tort law, New York utilizes interest analysis to determine which of two competing jurisdictions has the greater interest in having its law applied in the litigation.”). As I have explained, however, the majority’s application of Neumeier Rule Three is not the only possible interpretation of New York law (nor, as I note below, is it even the most reasonable), and in these circumstances I think that federal-state comity and the exercise of good judgment call for certification of this question to the New York Court of Appeals.
Certification of a determinative question of state law is appropriate where, as here, existing state precedents do not enable us to predict how that state’s highest court would decide the question. See McCarthy v. Olin Corp., 119 F.3d 148, 154 (2d Cir.1997) (“Because it is our job to predict how the forum state’s highest court would decide the issues before us, we will not certify questions of law where sufficient precedents exist for us to make [that] determination.”). For example, in Liriano v. Hobart Corp., 132 F.3d 124 (2d Cir.1998), we certified a question of state law to the New York Court of Appeals where our review of the applicable precedents concluded that “at least four possible views of New York law present themselves.” Id. at 131. Here, certification is warranted because there are at least two possible and conflicting views of New York law that we can ascertain from the relevant precedents — Neumeier Rule Three is either to *115be directly and literally applied according to its language, or is to be indirectly applied by combining it with a general interest-balancing approach.
Liviano is admittedly distinguishable from the instant case because Liviano involved a question of law that the New York Court of Appeals had never addressed and on which there existed divergent views in the intermediate New York appellate courts, see id. at 132, while the instant case instead involves two decisions of the New York Court of Appeals that have left unclear which of two possible analytical approaches should be applied. The same rationale, however — that certification is warranted where existing state-court precedents do not enable us to predict with confidence the outcome that the state’s highest court would reach — applies equally to the state-law question at issue in Liviano and the state-law question we confront here.
Certification is especially warranted in the instant case because the majority’s reading of New York law represents a step away from the formulaic approach that the Neumeiev rules embody, and a return to the kind of general interest balancing that the Neumeiev rules were intended, in my view, to replace. If the Court of Appeals intended to move away from Neumeiev with Schultz, we ought to give that court the opportunity to state this intent directly-
Even if I were to agree with the majority that New York law is sufficiently clear to enable us to predict how the Court of Appeals would decide this case, I would disagree with the majority’s conclusion as to what the Court of Appeals’ resolution would be. I am reluctant to read the Court of Appeals’ decision in Schultz as a departure from Neumeier, as I believe the majority does, because of the strong arguments in favor of applying Neumeier Rule Three literally rather than engaging in a more subjective interest-balancing approach. In fact, the majority’s analysis in the instant case seems to me to provide a perfect example of the difficulties inherent in attempting to identify and balance states’ interests in seeing their laws applied. One such difficulty is that courts may not always accurately identify or properly weigh the interests of each state in the conflict. Here, for example, the majority takes a limited view of Connecticut’s interest in having its own charitable immunity law applied, failing to note that Connecticut has an important and compelling interest in protecting its state healthcare and insurance systems from bearing the uncompensated cost of Gilbert’s lifetime disability. This Court has consistently held that the state of a plaintiffs domicile “has an important and obvious interest in ensuring that its residents are fully and adequately compensated for tortious harm.” Pescatore v. Pan Am. World Airways, Inc., 97 F.3d 1, 14 (2d Cir.1996); see also Sheldon v. PHH Corp., 135 F.3d 848, 853 (2d Cir.1998) (“When the extent of victim compensation, is at issue, the plaintiffs domicile has an interest in applying its law because that forum is where the loss is felt and where the burden of the victim’s uncompensated needs may fall.” (emphasis omitted)).
A second difficulty is that the interests of the relevant states may not always be of a type that can easily be quantified and compared side by side. In the instant case, for example, the majority’s analysis fails to acknowledge that the economic trade-off between a state system of charitable immunity and one of non-immunity becomes difficult to calculate and balance where the parties to a dispute are domiciled in states with different rules. That is, New Jersey’s decision to provide charitable immunity to institutions within its *116state involves a trade-off that the state has decided to make — it reduces the cost to charitable institutions of providing services, thus encouraging and facilitating their operation; and it concomitantly increases the risk that the state will have to internalize (through its health-care, insurance, and other state service systems) the costs that state residents, injured as a result of a charitable institution’s negligence, are unable to bear individually. Connecticut has made the converse tradeoff in its decision not to provide charitable immunity. In a case like Gilbert’s, however, New Jersey only benefits (because its charitable institutions are immune from suit) and Connecticut only loses (because its state services must bear the uncompensated burden of Gilbert’s permanent disability).
Taking both of these additional considerations into account, it is far from clear to me that the majority has accurately or comprehensively weighed the relevant “substantive law purposes” at issue. The difficulty of accounting for all of the interests relevant to a state’s system of loss-allocation, and of balancing when the interests of several states are involved, provides a strong argument for applying the Neumeier rules literally. Direct application of Neumeier Rule Three in this case would lead to a straightforward result: the exception to Neumeier Rule Three has not been met, and New York’s law — as the lex loci delicti — must be applied, because displacing New York’s law with New Jersey’s would not advance the substantive law purposes of all the relevant jurisdictions.
However, because (as I have noted) New York law is not entirely clear regarding whether the Neumeier rules are to be literally applied, or whether courts are to conduct a more general interest balancing, I would certify the choice of law issue to the New York Court of Appeals.
. The magistrate judge considered Cook and declined to follow it because she found it to be "contrary to New York interest analysis.” This aptly illustrates both the tension between Neumeier Rule Three and a general interest analysis, as well as the confusion among courts as to which approach must be applied.