Summary
noting that judicial estoppel can apply if the party asserting a new position previously "convince[ed] the District Court to accept its earlier position"
Summary of this case from Singer Mgmt. Cons. v. MilgramOpinion
No. 07-2510.
Argued January 29, 2009.
Opinion Filed: October 26, 2009. As Amended December 4, 2009. Rehearing Denied December 4, 2009.
Appeal from the United States District Court for the District of New Jersey, Dennis M. Cavanaugh, J.
Anthony Bartell, Esquire, McCarter English, Newark, NJ, Stephan G. Weil, Esquire, Jerold Oshinsky, Esquire (Argued), Washington, DC, for Appellants.
Arnold R. Gerst, Esquire, Brad M. Weintraub, Esquire, Weiner Lesniak, Parsippany, NJ, Lawrence J. Bistany, Esquire (Argued), Celestine M. Montague, Esquire, White Williams, Philadelphia, PA, for Appellees.
Before: SCIRICA, Chief Judge, AMBRO, and SMITH, Circuit Judges.
OPINION OF THE COURT
G-I Holdings, Inc. ("G-I") purchased from Reliance Insurance Company a policy covering claims made against G-I's directors and officers between July 1999 and July 2002. Shortly after the policy was issued, Reliance encountered financial difficulties. In the summer of 2000, Hartford Fire Insurance Company took over claims administration for Reliance. Hartford also assumed some of Reliance's liabilities and reinsured other of those liabilities going forward. To protect itself against Reliance's impending insolvency, G-I split its initial Reliance policy with Hartford, keeping coverage for claims made up to July 15, 2000 with Reliance and shifting coverage for the remaining period (July 15, 2000 to July 1, 2002) to a new Hartford policy. Reliance went into liquidation and G-I sought coverage for three fraudulent conveyance suits against Samuel J. Heyman, its CEO, chairman of its Board of Directors and controlling shareholder. The first of those suits was brought during the amended Reliance coverage period, while the other two were brought during the Hartford coverage period. Under the policies' terms, because all three suits relate to the same allegedly fraudulent conveyance, their filing dates relate back to the date of the first suit (and thus fall within the Reliance coverage period).
We use this same term to refer to G-I's predecessor, GAF Corporation, to which G-I succeeded in 2000.
For convenience, all references to Hartford include Twin City Fire Insurance Company.
G-I filed a claim for coverage in Reliance's liquidation. But it also sued Hartford, arguing both that Hartford was liable under the policy it had issued to G-I (despite the fact that the first suit, to which the other two relate back, was filed during the Reliance policy period), and that Hartford was liable to G-I under the Reliance policy because agreements Hartford entered into with Reliance made Hartford responsible for Reliance's coverage obligations. In addition, G-I contended that Hartford was barred by judicial estoppel from arguing that the suits filed during its policy period related back to the suit filed during the Reliance policy period because, in an earlier stage of the litigation, Hartford had taken a position at odds with that argument. In granting Hartford's motion for summary judgment, the District Court rejected all of these contentions. We do so as well, and thus affirm.
I. FACTS AND PROCEDURAL HISTORY
In February 2000, G-I bought an insurance policy from Reliance that covered liability arising out of claims made by third parties against G-I's directors and officers (including Heyman) between July 1, 1999 and July 1, 2002. The policy included an "interrelated wrongful acts" provision, stating that the filing date of all suits arising from the same wrongful act would be the date on which the first such suit was filed. The coverage limit was $15 million. G-I appears to have paid a premium of $185,000.
In 1997, facing more than $200 million in existing asbestos liability and the prospect of hundreds of thousands of future claims, G-I distributed to Heyman the stock of a profitable subsidiary. As expected, asbestos claimants or their representatives filed fraudulent conveyance actions against Heyman and G-I: these were filed on (1) January 3, 2000 by an injured employee seeking class certification (the "Nettles action"); (2) September 19, 2000 by the Center for Claims Resolution, a non-profit entity created by asbestos defendants to pay claims (the "CCR action"); and (3) September 17, 2001 by the Official Committee of Asbestos Claimants in G-I's Chapter 11 bankruptcy case filed in 2001 (the "Claimants Committee action").
The United States District Court for the Southern District of New York and the Supreme Court of the State of New York, venues for the Nettles and CCR actions respectively, placed those actions on hold in 2001 pending resolution of material parts of G-I's bankruptcy case.
In early 2000, Reliance was in financial trouble. In summer and fall 2000, pursuant to an Asset Purchase Agreement, a Quota Share Reinsurance Agreement and two Claims Servicing Agreements, Hartford acquired renewal and other rights to, and became a reinsurer and servicer of, certain Reliance policies. In July 2000, after Reliance's rating fell below the minimum financial guidelines for insurers set by G-I's insurance broker, Marsh, G-I's risk manager (Robert Flugger) asked Marsh to arrange for G-I to acquire a directors and officers insurance policy from Hartford. Reliance changed the coverage termination date of its policy to July 15, 2000 from July 1, 2002, and Hartford issued G-I an identical policy with a period of July 15, 2000 to July 1, 2002. An endorsement to the Hartford policy limited the sum of coverage under it and the amended Reliance policy to $15 million. As part of the transaction, Reliance refunded, and Hartford received, $153,935.18 in premiums.
Marsh McLennan Companies is a professional services firm. Marsh is its insurance brokerage business. See Overview: The Businesses of MMC, http://www.mmc.com/about/index.php (last visited Oct. 21, 2009).
This new policy was issued by a Hartford subsidiary, Twin City Fire Insurance Company. For convenience, we refer simply to Hartford as the issuer.
In sum, there were now two policies: the amended Reliance policy, which covered claims made between July 1, 1999 and July 15, 2000; and the Hartford policy, which covered claims made between July 15, 2000 and July 1, 2002. The Nettles action filing date falls within the amended Reliance policy period and the CCR and Claimants Committee filing dates fall within the Hartford policy period. But because of the interrelated wrongful acts provisions in both the amended Reliance and Hartford policies, the filing date of the CCR and Claimants Committee actions relate back to the filing date of the Nettles action, placing them outside the Hartford policy period and within the amended Reliance policy period.
G-I and Heyman sought coverage for the three fraudulent conveyance actions. A Pennsylvania state court ordered the liquidation of Reliance in October 2001, and G-I has agreed to pursue coverage from Reliance in that proceeding. In this case, G-I seeks coverage from Hartford. The District Court in 2004 denied motions by G-I for summary judgment and by Hartford for dismissal. In June 2006, G-I again moved for summary judgment and Hartford did so as well. The District Court then granted Hartford's motion and denied that of G-I, which now appeals.
For the remainder of this opinion, we refer to G-I and Heyman collectively as "G-I."
The procedural history is complex. After Reliance denied coverage of the Nettles and CCR claims, G-I sued Reliance and its excess insurer, Great American Insurance Company, in November 2000 in New Jersey state court. G-I did not add Hartford as a defendant until August 2001 because, according to G-I, it was unaware of the agreements between Reliance and Hartford, had not seen the Hartford policy, and did not know that the amended Reliance policy moved up the coverage period end date from July 2002 to July 2000. Reliance removed the action to the United States District Court for the District of New Jersey. That action was subsequently dismissed and refiled by G-I in the Bankruptcy Court supervising its bankruptcy. In May 2002, the Bankruptcy Court referred this non-core proceeding back to the District Court. It stayed G-I's claims against Reliance. In 2004, the District Court dismissed claims against the excess insurer (Great American) without prejudice because G-I had not exhausted its rights against Reliance.
In August 2004, G-I added as defendants the New Jersey Property-Liability Insurance Guarantee Association and the New York Property/Casualty Insurance Security Fund. The District Court dismissed the claims against the former without prejudice in December 2006 because recovery against it depended on resolution of coverage issues pending against Reliance in Pennsylvania. In December 2008, the parties stipulated to dismissal without prejudice of G-I's claims and this appeal as against the latter. Claims against these insurers are therefore not before us on appeal.
II. JURISDICTION AND STANDARD OF REVIEW
The District Court had diversity jurisdiction pursuant to 28 U.S.C. § 1332 and we have jurisdiction under 28 U.S.C. § 1291. We exercise plenary review of an order granting summary judgment. Gonzalez v. AMR, 549 F.3d 219, 223 (3d Cir. 2008). Summary judgment is appropriate if there are no genuine issues of material fact and the moving party is entitled to judgment as a matter of law. Id. "[T]o survive summary judgment . . . [,] a non[-]movant must present sufficient evidence to allow a reasonable jury to find in [its] favor." Sovereign Bank v. BJ's Wholesale Club, 533 F.3d 162, 172 (3d Cir. 2008) (internal brackets and quotation marks omitted). We make our view of the evidence and inferences therefrom as favorable as possible to the non-movant. United States ex rel. Kosenske v. Carlisle HMA, Inc., 554 F.3d 88, 94 (3d Cir. 2009). The District Court applied New Jersey law; the parties do not appeal that choice and we follow it. We may affirm on any ground supported by the record. Azubuko v. Royal, 443 F.3d 302, 303 (3d Cir. 2006).
III. ANALYSIS
G-I argues that Hartford must cover some or all of the fraudulent conveyance actions because: (1) Reliance and Hartford agreed to provide insurance coverage for a single policy period, and thus the Hartford policy period includes the amended Reliance policy period; (2) the interrelated wrongful acts provisions in the policies should not apply, and therefore at least the two later-filed fraudulent conveyance actions fall within the policy period Hartford claims it covered; and/or (3) the purchase, servicing, and reinsurance agreements between Hartford and Reliance, and the close relationship of those parties in any event, make Hartford directly liable under the amended Reliance policy. It further contends that we should use the doctrine of judicial estoppel to bar Hartford from relying on the interrelated wrongful acts provisions to avoid coverage because Hartford asserted a contradictory argument in its motion to dismiss in the District Court.
A. The Hartford Policy Period Does Not Include the Amended Reliance Policy Period.
It appears that G-I argues that the Hartford policy period includes the entire initial Reliance policy period because (1) G-I requested a policy from Hartford that would cover the Reliance policy period and (2) Hartford behaved as if this were the case. We arrive at these arguments by construing in the most favorable way G-I's contentions that Reliance amended its policy period without G-I's consent and that Hartford effectively assumed Reliance's obligations.
G-I's argument that Reliance amended its policy without G-I's consent or otherwise failed to abide by the cancellation provisions of the policy is, strictly construed, of no help to G-I, as it suggests that the initial Reliance policy remains in force and Reliance, rather than Hartford, is therefore liable. We construe this argument as the claim that Hartford mistakenly provided G-I with a policy that failed to conform to a request by G-I for a policy that would cover the entire period of July 1, 1999 to July 1, 2002 (rather than the July 15, 2000 to July 1, 2002 period in the policy Hartford in fact issued).
We note at the outset that, in arguing that the Hartford policy period includes the initial Reliance period, G-I is asking us to disregard the plain language of the Hartford policy, which, by its terms, covers only the period between July 15, 2000 and July 1, 2002. That, however, is not necessarily fatal to G-I's prospects, at least not in this context. Under New Jersey law, an insurance policy that has been unilaterally drafted by the insurer (such as this one) will typically be treated as a contract of adhesion. See Doto v. Russo, 140 N.J. 544, 659 A.2d 1371, 1376 (1995). As such, New Jersey courts "constru[e] contracts of insurance to reflect the reasonable expectations of the insured in the face of ambiguous language and phrasing, and[,] in exceptional circumstances, [even] when the literal meaning of the policy is plain." Id. at 1377 (internal citation omitted). Thus, it is open to G-I to argue that (1) it was reasonable to expect that the Hartford policy period would include the period initially covered by the Reliance policy, and (2) this is one of those situations in which its reasonable expectations should trump the plain meaning of the policy. See Pizzullo v. N.J. Mfrs. Ins. Co., 196 N.J. 251, 952 A.2d 1077, 1089 (2008) ("[I]n some circumstances, we have recognized that it might be appropriate to permit an insured's reasonable expectation to overcome the plain meaning of a policy.").
This "reasonable expectations" approach to interpreting insurance contracts applies even where, as here, the insured is a sophisticated actor. See Doto, 659 A.2d at 1376-77 (company's commercial-umbrella liability policy); Sparks v. St. Paul Ins. Co., 100 N.J. 325, 495 A.2d 406, 414-15 (1985) (lawyer's liability insurance policy); Sealed Air Corp. v. Royal Indem. Co., 404 N.J.Super. 363, 961 A.2d 1195, 1206 (2008) (corporation's directors and officers policy).
In performing the reasonable-expectations analysis, we first ask whether, without making a request or receiving some affirmative signal, an insured can reasonably expect that the policy period of a new policy it takes out in response to the financial difficulties of a prior insurer will include the period of the old policy. If such an inclusion would follow as a matter of course, that would strengthen G-I's position considerably. However, G-I has provided no evidence to show that the process of acquiring a new policy from one insurer in response to the financial difficulties of a prior insurer is so standardized or driven by such determinate purposes that an insured (here G-I) can reasonably expect a specific relationship between the policy periods of the prior and new policies. As a result, to put at issue whether it was reasonable for G-I to have expected the Hartford policy to cover the entire Reliance policy period, G-I had to establish either that (1) it requested a policy covering the entire period and Hartford did not clearly refuse to provide it or, more generally, that (2) Hartford by its actions or representations otherwise created a reasonable expectation in G-I that Hartford would provide such a policy.
As noted, G-I employed an insurance broker, Marsh. Because G-I does not challenge the actions of its broker, we forgo analysis of agency relationships between or among Marsh, Hartford, and/or G-I. See Aden v. Fortsh, 169 N.J. 64, 776 A.2d 792, 800-01 (2001).
G-I has provided no evidence a reasonable jury could use to find that G-I actually requested a policy from Hartford covering the amended Reliance policy period of July 1, 1999 to July 15, 2000. The record contains only the statements of Flugger:
Q. . . . So you understood that you were going to have a Hartford policy from a certain date, it would be effective from a certain date agreed upon, and your Reliance policy would be canceled as of that date; is that correct?
A. Yes.
Q. . . . You understood that they were two separate policies[?]
. . .
A. They would have been two separate policies.
. . .
Q. . . . Did you have an understanding that the effective date would be some time in or about July of 2000 for [t]he Hartford policy? A. Yes.
Q. . . . [D]id you have an understanding that your Reliance policy would be in effect up until the date of the cancellation and rewrite?
A. Either that or they were just going to continue with the Reliance policy. Basically Hartford would handle the claims, and Hartford would collect the premiums and whatever. I believe there was a bifurcation between the two. . . . I was going to get Hartford paper for my D and O policy.
Based on this testimony, we cannot say that Flugger asked for a policy covering claims made during the amended Reliance policy period (July 1, 1999 to July 15, 2000).
G-I points to other statements by Flugger describing his expectations regarding the "prior litigation date" under the Hartford policy. "Prior litigation date" is a technical phrase referring to a date that precedes the start of the policy period. Flugger's statements are irrelevant because we are interested in his expectations concerning the policy period, not the "prior litigation date."
As there is no evidence that G-I requested that Hartford cover a period that included the amended Reliance period, we turn to whether G-I, based on Hartford's behavior, could reasonably have expected Hartford to include in its coverage the amended Reliance period. G-I has provided no evidence to support that proposition. It nevertheless points to two factors that it believes justify a reasonable expectation. First, the content and administration of the initial Reliance and Hartford policies were the same. Specifically: (a) in writing the policy, Hartford did none of the typical underwriting acts of a successor insurer ( e.g., requiring G-I to complete a new application and provide updated information); (b) the Hartford policy number remained the same as the initial policy number except for an "H"; and (c) claims administration pursuant to agreements between Reliance and Hartford transferred "seamlessly" from Reliance to Hartford. Second, G-I contends that the handling of premiums and the liability cap suggested that Hartford assumed the old Reliance policy — specifically, (a) G-I never paid additional premiums, (b) those it paid to Reliance were transferred directly to Hartford, and (c) an endorsement capped combined liability of Hartford and Reliance under the policies at $15 million.
G-I also points to an affidavit by Flugger that states that a Marsh representative "conveyed Hartford's offer to assume responsibilities under G-I's existing Reliance policy." Flugger does not state whether this offer included assumption of liability for claims made before July 15, 2000. Without more detail about what "responsibilities" Flugger meant, this lends no support to G-I.
The handling of premiums and the combined liability cap could not have created a reasonable expectation that Hartford would cover the amended Reliance policy period. There is evidence that Flugger knew that Reliance split the original policy premium with Hartford. That should have suggested to a reasonable corporate insured employing a risk manager that Reliance and Hartford were splitting the risk insured by the original policy. Such a splitting of risk should in turn have suggested to G-I that Reliance and Hartford were splitting the initial three-year policy period. G-I should have concluded that Hartford was not agreeing to cover the entire initial Reliance policy period.
The initial Reliance policy states the premium as $185,000, excluding a small state surcharge. The binder for the Hartford policy lists the premium as $153,935.18. This suggests that Reliance kept more than $30,000 of the initial premium when it amended its policy to cover the shorter period of July 1, 1999 to July 15, 2000. Hartford got the rest.
We note that there is some question whether G-I actually paid $185,000 for the initial Reliance policy. The premium listed for the Hartford policy is $121,422, before applying a small state surcharge. This amount differs from the Hartford policy premium of $153,935.18 listed on the insurance binder issued by Hartford. By analogy, a discrepancy might also exist between the premium listed on the initial Reliance policy and what G-I paid.
Regardless how much premium G-I actually paid, there is evidence that Reliance split it with Hartford. The record contains statements by both Flugger and a Hartford employee to that effect. Flugger stated in an affidavit that a Marsh representative told him that "unearned" premiums paid by G-I to Reliance would be forwarded to Hartford. The Hartford employee stated that "Hartford agreed without reunderwriting to write a new policy for the remainder of the term for the pro rata premium that was cancelled out of the Reliance account."
We are mindful that Hartford took over 80% of the initial premium, and this exceeded the portion of the initial policy period that Hartford assumed (roughly 67%). If, as a matter of industry practice, premium and policy period should be proportional (a possibility not addressed in the record), then this suggests that Hartford's share of the original premium was out of proportion to its period of coverage.
If Hartford's share of the original premium were grossly excessive relative to the portion of the policy period that it assumed, then it might be reasonable for G-I to expect Hartford to cover more of the original policy period. G-I has not made this argument and we can find no evidence that Hartford received a grossly excessive share of the initial premium in any event. We therefore do not believe that the mismatch between premium division and policy period coverages justifies denying summary judgment here.
The combined liability cap also should have suggested to a reasonable insured that Reliance and Hartford split the risk (and coverage period) of the initial policy. That policy charged a certain premium for coverage of $15 million. Because G-I did not add to the initial premium in obtaining the Hartford policy (rather, it split the premium between Reliance and Hartford), it makes sense that combined coverage under the amended Reliance and Hartford policies would not vary greatly from the $15 million commanded at the outset by the same premium (assuming the risk profile of G-I had not changed markedly since Reliance first wrote its coverage). In fact, the parties kept the combined limit at the first-set $15 million.
If Hartford had assumed all risk under the initial Reliance policy, then a reasonable insured would have expected Hartford to acquire the entire premium G-I had initially paid to Reliance and to have assumed coverage up to the full $15 million limit. That Hartford did not do these things should have put G-I on notice that, absent other affirmative signs, it could not reasonably expect Hartford to cover the entire policy period.
The identity in content and administration of the amended Reliance and Hartford policies does not change our view. We may take G-I's sophistication into account in deciding what was objectively reasonable for it to expect from its insurers. See Werner Indus., Inc. v. First State Ins. Co., 112 N.J. 30, 548 A.2d 188, 192 (1988). Any knowledge of G-I that Hartford had taken over claims administration for Reliance should not reasonably have caused G-I to expect coverage from Hartford for the entire policy period. A reasonable insured with enough sophistication to employ a risk manager would know the difference between claims servicing and the assumption of liabilities. Similarly, a sophisticated insured would recognize that use of identical policy language and a lack of additional underwriting do not necessarily signal that a new insurer has assumed liabilities under an old policy. Such an insured would also recognize that these attributes are consistent with a splitting of risk, policy periods and premiums between an old and a new insurer. We thus discern no basis to conclude that G-I could have reasonably expected its policy with Hartford to cover the Reliance policy period. B. The Interrelated Wrongful Acts Provision Applies.
As we discussed above, G-I's sophistication does not prevent us from construing the insurance policies it acquires in light of its reasonable expectations. It does, however, affect the analysis of what would count as a reasonable expectation for it to have had when obtaining a particular policy.
It is therefore unnecessary for us to decide whether this is one of those instances in which an insured's reasonable expectations can trump the plain meaning of the policy language, an issue with respect to which New Jersey law provides little guidance.
As an alternative, G-I claims that even if the Hartford policy does not cover the entire initial Reliance policy, it covers the CCR and Claimants Committee actions because plaintiffs in those actions filed them within the Hartford policy period of July 15, 2000 to July 1, 2002. In response, Hartford points to the interrelated wrongful acts provision in the policy, which (as noted already) deems the filing date for all claims arising out of the same wrongful act as the filing date of the first such claim. Because plaintiffs in the Nettles action (the first of the three fraudulent conveyance actions) filed on January 3, 2000, before the July 15, 2000 start of the Hartford policy period, the interrelated wrongful acts provision, Hartford contends, bars coverage for all three actions.
G-I appears to concede this in the abstract, but counters that the interrelated wrongful acts provision should not apply to this case because the purposes for which insurers include that provision in contracts do not apply here. Citing insurance treatises, G-I states that those purposes are (1) to ensure that risks arising out of the same wrongful act are subject to one policy and therefore one liability limit, and (2) to prevent changes in policy language from one policy period to another from creating disparate coverage determinations for the same wrongful act. According to G-I, these purposes do not apply here because the $15 million combined cap on the Hartford policy and amended Reliance policy already ensures that G-I cannot recover more than $15 million on all claims combined for both policies. Because the language of the amended Reliance and Hartford policies is identical, G-I continues, there is no danger of disparate coverage determinations, and thus we should treat the policy's interrelated wrongful acts provision as not triggered under the facts of this case.
We are not convinced. Even were we inclined to make application of the policy's interrelated wrongful acts provision contingent on the purposes behind that provision — as opposed to applying the contract as written — we would still apply the provision. That is because the interrelated wrongful acts provision does have a purpose that applies to this case. Such a provision not only allows insurers to cabin related wrongful acts to a single policy period (thus subject to one limit), it also, as one of the treatises that G-I itself cites points out, allows an insured (such as G-I) to obtain coverage under a new policy, despite facing additional liability exposure from its past acts, by "reserving the argument that any future claims arising out of the interconnected wrongful acts of a previously submitted claim will be covered by the former policy." John F. Olson et al., Director and Officer Liability: Indemnification and Insurance § 12:10 (2008). We thus have ample reason to give effect to the interrelated wrongful acts provision by barring coverage for the CCR and Claimants Committee actions under the Hartford policy.
As noted above, New Jersey courts, in interpreting insurance contracts, strive to give effect to the reasonable expectations of the insured, even in some instances where doing so would run contrary to the contract's plain meaning. See Pizzullo, 952 A.2d at 1089. We need not decide whether this is one of those instances, since an examination of the purposes for which insurers include interrelated wrongful acts provisions in contracts fail to reveal anything that would have justified G-I in thinking that the provision would not apply under the facts before us.
C. Other Agreements Do Not Make Hartford Directly Liable For The Fraudulent Conveyance Actions.
G-I also argues that, even if the Hartford policy does not cover the three fraudulent conveyance actions, Hartford still must cover them because agreements between Hartford and Reliance make Hartford directly liable under the amended Reliance policy. Specifically, G-I contends that (1) the agreements themselves create direct liability, or (2) the reinsurance relationship created by them brings this case within the ambit of Venetsanos v. Zucker, Facher Zucker, 271 N.J.Super. 459, 638 A.2d 1333 (1994), which held a reinsurer directly liable to an insured because of the reinsurer's close relationship with the insurer.
1. Hartford's agreements with Reliance
After Reliance encountered financial difficulties, Hartford entered into a series of agreements with Reliance concerning various coverage obligations of Reliance. Contrary to what G-I argues, however, none of those agreements creates direct liability of Hartford for the amended Reliance policy.
The Asset Purchase Agreement between Reliance and Hartford transferred books, records, and renewal rights (among other things) relating to the amended Reliance policy to Hartford. It also transferred liabilities associated with the policy "to the extent that any such liability is applicable to and accrues with respect to periods subsequent to the Closing [likely June 30, 2000, the date the parties executed the Asset Purchase Agreement]." We therefore read the Asset Purchase Agreement to transfer liability to Hartford on the amended Reliance policy only for claims made sometime after June 30, 2000. Because the plaintiffs in the Nettles action filed in January 2000, and the subsequent actions relate back to that date, the Asset Purchase Agreement did not transfer liability for them to Hartford.
We realize that the amended Reliance policy covered until July 15, 2001. The difference in dates does not affect our analysis. The same is true for the Quota Share Reinsurance Agreement and Claims Servicing Agreements discussed immediately below.
The initial Reliance policy permits modification only by endorsement signed by Reliance. The change in policy period at issue in this case probably required such an endorsement. But there is none in the record. However, whether the amendment thus is ineffective is irrelevant to this case because we are concerned with claims made (or relating back to) before July 2000, and it is undisputed that pre-July 2000 claims fall within both the initial and amended Reliance policy periods.
The Quota Share Reinsurance Agreement makes Hartford liable for a portion of amounts actually paid by Reliance under the amended Reliance policy "relating to claims . . . made on or after" July 1, 2000. As a threshold matter, the Reinsurance Agreement does not create direct liability to the insured for Hartford, but only liability of it to Reliance for amounts Reliance pays out under the amended Reliance policy. But even if it did, it would not cover the three fraudulent conveyance actions because the Reinsurance Agreement applies only to claims made on or after July 1, 2000.
Hartford entered into two Claims Servicing Agreements, one covering claims made on the amended Reliance policy on or before June 30, 2000, and the other covering claims made after that date. Hartford agreed to adjust, defend, and in some cases settle (on Reliance's behalf) claims under Reliance policies. Both agreements contain a provision stating that Hartford "is in no event financially responsible for payment or satisfaction of 'claims,' lawsuits, or any form of cause of action against any 'named insured' under the 'policy(ies).'" Thus, here too there is no basis for imposing direct liability on Hartford.
2. Hartford's relationship with Reliance
G-I argues additionally that, even if none of the specific agreements between Hartford and Reliance makes Hartford liable under the amended Reliance policy, the close relationship created by those agreements had the effect of making Hartford generally responsible for Reliance's coverage obligations. We disagree.
As noted above, this particular argument of G-I's is drawn from Venetsanos. In that case, the insurer, Homestead, "front[ed]" in New Jersey despite its lack of an insurance license by reinsuring a licensed insurer, Mutual, that it controlled. Venetsanos, 638 A.2d at 1334-35, 1337. Homestead (1) had "the entire exposure for [policy] liability," id. at 1336, (2) had final decision-making authority over whether to pay, id. at 1335, (3) conducted the insurance investigation, id., (4) was responsible for negotiation and settlement, id. at 1337, and (5) acted as the agent for service of process on Mutual, id. at 1336. Although Venetsanos recognized the "general rule that an original insured does not enjoy a right of direct action against a true reinsurer," id. at 1339, it held that Homestead was directly liable on the policy, id. at 1340, noting that "[w]here . . . the reinsuring agreement itself provides . . . that it takes charge of and manages the defense of suits against the original insured, the reinsurer may be held to be a 'privy'" to the insured's action against the insurer. Id. at 1339.
The Court in Venetsanos was concerned with protecting the ability of a New Jersey-insured to seek redress against a foreign "fronter" in New Jersey courts. Id. at 1338-39. The insurer, Mutual, had entered rehabilitation proceedings in Pennsylvania and the Court was concerned that the plaintiff would find it difficult to litigate its bad faith and coverage claims in a foreign jurisdiction, particularly where, as in that case, the reinsurance agreement had not been located. Id.
Our case differs much from Venetsanos because (1) Hartford did not have the same level of control over Reliance that Homestead had over Mutual and (2) there is no allegation of fronting. The Asset Purchase Agreement does not confer control over defense and settlement of claims under the amended Reliance policy and, as we remarked above, does not transfer liability for the fraudulent conveyance actions to Hartford. The Quota Share Reinsurance Agreement irrevocably makes Hartford Reliance's "lawful attorney-in-fact . . ., with full power of substitution[,]" in defending, settling and otherwise administering claims under the amended Reliance policy, but only "with respect to claims made on or after" July 1, 2000. The Reinsurance Agreement thus did not cede Reliance's control over the earlier-filed Nettles action or the other two fraudulent conveyance actions, the filing dates of which relate back to that of the first action. As we already noted, the Reinsurance Agreement also did not reinsure Reliance for the fraudulent conveyance claims. (Even if it did, it did not do so fully, as that Agreement commits only a share of losses to Hartford.) The Claims Servicing Agreements specifically preserved the authority of Reliance to direct Hartford in the handling of any claim. In this context, Venetsanos is simply off track from our case.
We consider misleading G-I counsel's conclusory statement in its opening brief that, under the Reinsurance Agreement, "Hartford will reimburse one hundred percent (100%) of any payments by Reliance." The relevant provisions of the Agreement are more nuanced. They read:
The Ceding Company [Reliance] cedes and the Reinsurer [Hartford] hereby accepts one hundred percent (100%) of the Ceding Company's Net Liability . . . for Ultimate Net Loss relating to claims . . . made on or after the Effective Time [July 1, 2000]. "Ultimate Net Loss" means that amount [of losses] that the Ceding Company has paid. . . .
"Net Liability" means . . . that portion of any Ultimate Net Loss that the Ceding Company has retained net for its own account after the application of all reinsurance . . ., which shall be a percentage of the Ceding Company's gross liability for losses, which percentage shall be the applicable quota share percentages indicated in Attachment A . . . and shall not exceed the limits set forth therein. Reinsurer's liability hereunder with respect to any loss or losses shall not be increased by reason of the inability of the Ceding Company to collect from any other reinsurers, whether specific or general, any amounts which may be due from them, whether such inability arises from the insolvency of such other reinsurers or otherwise. . . .
These provisions tie Hartford's liability to percentages (most less than 100) contained in Attachment A to the Reinsurance Agreement.
We note in addition that we are particularly reluctant to permit direct recovery by the insured against the reinsurer in this case because of Reliance's Pennsylvania liquidation proceeding. Doing so could expose Hartford to double liability because Pennsylvania law does not reduce a reinsurer's liability to the insurer's estate as a result of direct payments to the insured. 40 Pa. Cons. Stat. § 221.34. Direct recovery would also prevent the Pennsylvania liquidator from equitably apportioning the recovery among all insureds. See id. § 221.1(c)(iv). We believe G-I should stand in line in Pennsylvania with other insureds. See generally James E. Rudnik, Reinsurance as a Source of Recovery for Insured Losses, 15-Jan. Construction Law. 31, 33-34 (1995).
D. Judicial Estoppel Does Not Apply to Hartford's Invoking Of The Interrelated Wrongful Acts Provision.
In its first complaint, G-I argued that Hartford was liable on the amended Reliance policy solely because of its reinsurance and other agreements with Reliance. It did not yet argue, as it does here, that Hartford was liable for the CCR and Claimants Committee actions under the Hartford policy. In motioning to dismiss the complaint in November 2002, Hartford claimed that it was not bad faith to deny coverage of the CCR and Claimants Committee actions because, regardless whether Hartford's agreements with Reliance created direct liability for the amended Reliance policy, both actions were filed after the end of the amended Reliance policy period. In making this argument, Hartford ignored the interrelated wrongful acts provision in the amended Reliance policy. As we have seen, that provision relates the filing dates for the CCR and Claimants Committee actions back to the Nettles action filed during the Reliance policy period.
In February 2003, G-I amended its complaint to claim coverage under the Hartford policy. In response to G-I's new claim, Hartford changed its position. In March 2003, it obtained permission from the District Court to withdraw its initial brief and filed a new one invoking the interrelated wrongful acts provision in the Hartford policy to deny coverage for the two later actions. It now argued — as it does here — that, due to that provision, the two later actions relate back to the amended Reliance policy period. The District Court denied the motion to dismiss but, as we have seen, eventually granted summary judgment based in part on the interrelated wrongful acts provision.
G-I argues that, because Hartford initially argued that the CCR and Claimants Committee actions were not covered by the Reliance policy, it should not be allowed to invoke the interrelated wrongful acts provision to deny coverage for those actions under the Hartford policy. Under the doctrine of judicial estoppel, a court can defend the integrity of the judicial process by barring a party from taking contradictory positions during the course of litigation. See, e.g., Zedner v. United States, 547 U.S. 489, 504, 126 S.Ct. 1976, 164 L.Ed.2d 749 (2006); Scarano v. Cent. R.R. Co. of N.J., 203 F.2d 510, 513 (3d Cir. 1953); Ali v. Rutgers, 166 N.J. 280, 765 A.2d 714, 718 (2000).
As a threshold matter, we consider whether in diversity cases we should apply federal or state judicial estoppel law. In general, federal courts apply state law in diversity cases, at least where that law is substantive in nature. See Erie R.R. Co. v. Tompkins, 304 U.S. 64, 78, 58 S.Ct. 817, 82 L.Ed. 1188 (1938). But where an area of law implicates a "strong federal policy," federal law may apply. See Byrd v. Blue Ridge Rural Elec. Coop., 356 U.S. 525, 538, 78 S.Ct. 893, 2 L.Ed.2d 953 (1958). We have long avoided deciding whether federal judicial estoppel law applies in diversity cases. See In re Chambers Dev. Co., 148 F.3d 214, 229 n. 13 (3d Cir. 1998); Ryan Operations G.P. v. Santiam-Midwest Lumber Co., 81 F.3d 355, 358-59 n. 2 (3d Cir. 1996). But today we weigh in, as we believe that "[a] federal court's ability to protect itself from manipulation by litigants should not vary according to the law of the state in which the underlying dispute arose." Ryan Operations, 81 F.3d at 358 n. 2. In doing so, we follow five other Courts of Appeals. See Eastman v. Union Pac. R.R. Co., 493 F.3d 1151, 1156 (10th Cir. 2007); Ogden Martin Sys. of Indianapolis, Inc. v. Whiting Corp., 179 F.3d 523, 527 n. 1 (7th Cir. 1999); Rissetto v. Plumbers and Steamfitters Local 343, 94 F.3d 597, 603-04 (9th Cir. 1996); Guinness PLC v. Ward, 955 F.2d 875, 899 n. 20 (4th Cir. 1992) (quoting Allen v. Zurich Ins. Co., 667 F.2d 1162, 1167 n. 4 (4th Cir. 1982)); Edwards v. Aetna Life Ins. Co., 690 F.2d 595, 598 n. 4 (6th Cir. 1982).
Other Courts, however, have gone the other way or are uncertain. See, e.g., Monterey Dev. Corp. v. Lawyer's Title Ins. Corp., 4 F.3d 605, 608-09 (8th Cir. 1993) (applying state law); Konstantinidis v. Chen, 626 F.2d 933, 937-38 (D.C. Cir. 1980) (same); see also Alternative Sys. Concepts, Inc. v. Synopsys, Inc., 374 F.3d 23, 32 (1st Cir. 2004) (stating that "[a]s judicial estoppel appears neither clearly procedural nor clearly substantive, there may be a legitimate question as to whether federal or state law . . . should supply the rule of decision," but declining to answer it).
Though there is no rigid test for judicial estoppel, three factors inform a federal court's decision whether to apply it: there must be (1) "irreconcilably inconsistent positions;" (2) "adopted . . . in bad faith;" and (3) "a showing that . . . estoppel . . . address[es] the harm and . . . no lesser sanction [is] sufficient." Chao v. Roy's Constr., Inc., 517 F.3d 180, 186 n. 5 (3d Cir. 2008) (internal quotation marks omitted). We do not consider these factors, however, because, in our Circuit judicial estoppel is generally not appropriate where the defending party did not convince the District Court to accept its earlier position. United States v. Pelullo, 399 F.3d 197, 222-23 (3d Cir. 2005) (quoting New Hampshire v. Maine, 532 U.S. 742, 750-51, 121 S.Ct. 1808, 149 L.Ed.2d 968 (2001)); Dam Things from Denmark v. Russ Berrie Co., Inc., 290 F.3d 548, 559 n. 16 (3d Cir. 2002); Montrose Med. Group Participating Sav. Plan v. Bulger, 243 F.3d 773, 778 (3d Cir. 2001).
Here the District Court never accepted Hartford's prior position. Hartford withdrew that position and asserted its new one ( i.e., that the interrelated wrongful acts provision applies) before the Court ruled on its motion to dismiss. When it did rule, the Court did not rely on Hartford's initial position. Rather, it held against Hartford without discussing either Hartford's prior or new position. On summary judgment, Hartford reasserted its new position, and the District Court relied on it. Because the Court never relied on Hartford's first position, we shall not bar its new one.
We do not mean to suggest that where no court has accepted an initial position, judicial estoppel can never apply. We will apply it to neutralize threats to judicial integrity however they may arise. For example, in Krystal Cadillac-Oldsmobile GMC Truck, Inc. v. General Motors Corp., 337 F.3d 314 (3d Cir. 2003), a bankruptcy case, we applied judicial estoppel even though no court had ever relied on the debtor's initial position. Id. at 320-21. We did so because creditors almost certainly had relied on it, undermining the bankruptcy process by weakening their bargaining position. Id. at 324-25.
Here, G-I has provided no evidence of a threat to judicial integrity other than Hartford's inconsistent positions. We believe applying judicial estoppel here presents a greater threat to judicial integrity. We do not preclude arguments not accepted by the District Court in part to ensure that the order in which a party presents its claims does not determine the outcome of a case. Without this limitation, an amendment to a complaint can check-mate opposing counsel by introducing a new claim the defense of which contradicts the opposition's initial position. By amending its complaint, a plaintiff could (intentionally or not) force the defense to choose between conceding the old claim or the new one. That result is undesirable because a defendant ought to have the opportunity to put up the best possible defense in light of all the claims against it. Where, as here, a defendant has changed position in response to an amended complaint, there is no offense to the integrity of the judicial process warranting estoppel. There is only danger to that process averted.
IV. CONCLUSION
Because it was concerned that Reliance would become insolvent, G-I shifted part of the risk under an existing Reliance policy to Hartford. For a higher price, it might have shifted all of it. Because it did not pay that higher price and Reliance became insolvent, G-I must seek coverage for risks it kept with Reliance in that company's liquidation proceeding. Hartford, both in its dealings with Reliance and in the policy it issued to G-I, sought only to cover risks starting in July 2000 that were not otherwise excluded (such as by the interrelated wrongful acts provision). It shielded itself from the claims for which G-I now seeks coverage. We decline to tamper with this scheme. Although Hartford switched its position before the District Court, that Court did not rely on the earlier position, and thus as a threshold matter we will not bar the change. We therefore affirm in all aspects.