The subject of this appeal is asbestos, a hazardous material found in public and private buildings everywhere. The presence of this substance has precipitated widespread litigation against its manufacturers, which have looked to their insurance carriers to defend and indemnify them. Before us on this appeal are a former manufacturer of asbestos products and the insurance companies that insured it. Here, the legal maxim volenti non fit injuria--that is no injury which is done with the injured party's consent--does not apply even though the owners had asbestos installed in their buildings. The maxim is inapt because neither the owners nor the individuals actually harmed by its carcinogenic effects consented to their injuries. The question before us instead centers on the bottom-line issue of who ultimately is liable in damages and litigation expenses to the owners of buildings contaminated with this mineral fiber. To resolve that question, we must make an initial legal determination as to when injury occurs to such a building.

We set forth first the background facts. W.R. Grace & Co.-Conn. (Grace), and companies it later acquired, manufactured and sold asbestos building products. These products were used for soundproofing and fireproofing buildings from the 1940s until 1973 when the U.S. Environmental Protection Agency (EPA) prohibited asbestos sales because inhalation of asbestos fibers caused serious health problems, particularly cancer. Grace then experienced a flood of litigation. As of May 31, 1989, 17,411 individual lawsuits had been filed against it for personal injuries caused by asbestos products, and building owners had filed another 212 lawsuits for property damage caused by the presence of asbestos in their buildings and its subsequent removal or containment.

The instant litigation began as a declaratory judgment action in October 1983 in the United States District Court for the Southern District of New York. There the parties agreed, pursuant to 28 U.S.C. Sec. 636(c) (1988 & Supp. III 1991), to refer the matter to U.S. Magistrate Judge Leonard Bernikow. Plaintiff Maryland Casualty Co. (Maryland)--which had sold comprehensive general liability insurance to Grace from 1955 to 1973--sued defendant Grace and its current primary insurer, defendant Continental Casualty Co. (Continental), to obtain a judicial declaration as to what obligation, if any, Maryland owed to defend or indemnify Grace in the thousands of underlying asbestos lawsuits. Jurisdiction was premised on diversity of citizenship. In April 1985 Maryland successfully added three other defendants, Royal Indemnity Co. (Royal), Aetna Casualty & Surety Co. (Aetna), and General Insurance Co. (General). Each of these defendants had sold primary insurance to asbestos manufacturers and distributors acquired by Grace between 1963 and 1967. Prior to 1963 Grace itself did not manufacture or distribute any asbestos products.

The insurance companies, with the exception of Continental, dispute when they insured Grace or its acquired companies. Continental has insured Grace from 1973 until the present; Maryland, Continental's immediate predecessor, insured Grace from 1955 to 1973. Grace contends that the following other insurance was also issued: Royal supposedly insured Zonolite Co., an asbestos manufacturer acquired by Grace in 1963, from 1950 to 1963; General is argued to have insured a Grace predecessor, Vermiculite Northwest Inc., between 1961 and 1966; and Grace asserts Aetna insured three Grace predecessors, California Zonolite Co., Ari-Zonolite Co. and Western Mineral Products Co., from 1951 to 1970.

In June 1987 the parties, with the exception of Continental, cross-moved for summary judgment on 13 separate issues. Grace and Continental reached a settlement in July 1990. In a 75-page opinion dated March 1, 1991 the magistrate judge granted summary judgment on several matters, including the issue of insurance covering the property damage lawsuits. Most of the insurance policies sold to Grace and its predecessors had based coverage on an "occurrence" of property damage during their effective dates, except for Royal's policies that based coverage on "accidents." This distinction was not relevant, according to the magistrate judge, because "Royal agree[d] to be bound by whatever trigger of coverage the court endorse[d]."

Applying New York law, which the parties agree controls interpretation of all the insurance contracts, the magistrate judge adopted a "discovery trigger" for property damage coverage. Damage to property did not "occur," nor was insurance coverage triggered, the trial court explained, under the terms of the insurance policies until the building owner discovered, first, that there was asbestos in the building and, second, that it was hazardous. In nearly all the property damage cases building owners did not discover the dangers of asbestos until the early 1970s when the EPA and other federal agencies alerted the public to its hazardous nature.

The magistrate judge reaffirmed adoption of this trigger in orders dated July 8, 1991 and October 22, 1991. Application of the discovery trigger obligates only Continental, which as noted has insured Grace from 1973 to the present date, and relieves all the earlier insurers of any liability. According to the trial court's October 22 order, "Grace presented no evidence of the discovery of property damage during Maryland's policy period," so Maryland was relieved of any coverage obligation even though it had insured Grace from 1955 to 1973. The financial consequences of this decision are substantial. According to Grace, as of early 1992 it had spent $184.6 million to settle claims or satisfy judgments in property damage asbestos lawsuits and $194.8 million to defend itself in such suits. Continental has exhausted the limits of its insurance coverage by paying $117 million in defense costs and $70 million in indemnity to Grace.

As a result of the March 1, 1991 decision, Royal and Maryland requested entry of final judgments on the property damage trigger issue. These judgments from which Grace subsequently filed this appeal, were entered pursuant to Fed.R.Civ.P. 54(b) on November 19, 1991. In addition to its 1990 settlement with Continental, Grace and Maryland reached a settlement in December 1991, but Maryland still is a party to the appeal because its liability to the other insurers depends upon the coverage trigger we ultimately adopt. During a February 1992 trial to determine whether Aetna had sold insurance to any Grace predecessors, Aetna also settled with both Grace and Maryland. The Aetna-Maryland settlement is contingent on the continued validity of a March 13, 1992 judgment of the magistrate judge dismissing all the other insurers' claims against Aetna.

On this appeal Grace challenges the adoption of the discovery trigger for property damage as a matter of New York law. Royal and Maryland advocate the discovery trigger because its adoption relieves them of liability in the underlying property damage lawsuits against Grace. Grace also renews a challenge to subject matter jurisdiction that impacts, among other things, the Aetna-Maryland settlement. We discuss the jurisdiction issue first.

Diversity jurisdiction requires that every plaintiff on one side of a civil action be a citizen of a different state than every defendant on the other side. See 28 U.S.C. Sec. 1332; 1 James W. Moore et al., Moore's Federal Practice p 0.71[5.-2] (2d ed. 1993). A corporation's citizenship is deemed to be that of the state in which it is incorporated and that of the state where it has its principal place of business. See 28 U.S.C. Sec. 1332(c). As the action currently is aligned, plaintiff Maryland is a Maryland citizen. None of the defendants is a citizen of that state.

But defendant Grace contends the parties should be realigned so that it stands alone against all the insurers. Such realignment would destroy diversity and deprive the federal courts of jurisdiction over this controversy because Grace, a Connecticut corporation doing business in New York at the start of this litigation, is not diverse from Aetna, a Connecticut citizen, or Royal, a Delaware corporation doing business in New York at the time the complaints were filed. According to Grace, realignment is appropriate because the insurance companies are united in a common effort to deny it insurance coverage in the avalanche of asbestos litigation filed against it. That this challenge to subject matter jurisdiction has been renewed by Grace nearly a decade after the litigation began is unexceptional because jurisdiction may be raised at any time during the course of litigation. See Fed.R.Civ.P. 12(h)(3); see also 1 Moore et al., supra, p 0.74, at 764-65.

The relevant focus for determining diversity is the parties' citizenship when suit is commenced. See Anderson v. Watt, 138 U.S. 694, 702-03, 11 S. Ct. 449, 451, 34 L. Ed. 1078 (1891). The legal principles governing this subject were succinctly summarized by the Supreme Court in Indianapolis v. Chase Nat'l Bank, 314 U.S. 63, 62 S. Ct. 15, 86 L. Ed. 47 (1941), where Justice Frankfurter wrote: "Diversity jurisdiction cannot be conferred upon the federal courts by the parties' own determination of who are plaintiffs and who defendants. It is our duty ... to 'look beyond the pleadings and arrange the parties according to their sides in the dispute.' ... Litigation is the pursuit of practical ends, not a game of chess. Whether the necessary 'collision of interests,' ... exists is therefore not to be determined by mechanical rules. It must be ascertained from the 'principal purpose of the suit,' ... and the 'primary and controlling matter in dispute'...." Id. at 69-70, 62 S.Ct. at 17; see also 1 Moore, et al., supra, p 0.74, at 771 ("The purpose of realignment is to ensure that the case truly involves the kind of adversarial relationship constitutionally required in a case or controversy in the federal courts."). Other courts applying the Indianapolis standard have focused on different phrases of the above-quoted passage, and as a result different tests have evolved.

Grace urges us to look to the single, primary purpose of Maryland's lawsuit. It argues that a lawsuit must be distilled to its single-issue essence for realignment purposes, and the parties must then "be aligned in accordance with the primary dispute in the controversy, even where a different, legitimate dispute between the parties supports the original alignment." United States Fidelity & Guar. Co. v. Thomas Solvent Co., 955 F.2d 1085, 1089 (6th Cir.1992); see also Employers Ins. of Wausau v. Crown Cork & Seal Co., 942 F.2d 862, 864 (3d Cir.1991); Continental Airlines, Inc. v. Goodyear Tire & Rubber Co., 819 F.2d 1519, 1523 & n. 2 (9th Cir.1987).

The insurance companies advocate a broader "collision of interests" test. Under this test courts require the existence of an actual, substantial controversy, or a collision of interests, see Indianapolis, 314 U.S. at 69, 62 S.Ct. at 68, but the conflict may in some cases concern an issue other than the so-called primary issue in dispute. See generally U.S.I. Properties Corp. v. M.D. Constr. Co., 860 F.2d 1, 4-5 (1st Cir.1988), cert. denied, 490 U.S. 1065, 109 S. Ct. 2064, 104 L. Ed. 2d 629 (1989); Zurn Indus., Inc. v. Acton Constr. Co., 847 F.2d 234, 237-38 (5th Cir.1988); American Motorists Ins. Co. v. Trane Co., 657 F.2d 146, 151 (7th Cir.1981); Farmers Alliance Mut. Ins. Co. v. Jones, 570 F.2d 1384, 1387 (10th Cir.), cert. denied, 439 U.S. 826, 99 S. Ct. 97, 58 L. Ed. 2d 119 (1978); Universal Underwriters Ins. Co. v. Wagner, 367 F.2d 866, 870-71 (8th Cir.1966). This approach is more flexible because it permits courts deciding whether diversity exists to consider the multiple interests and issues involved in the litigation.

We adopt the collision of interests test to resolve the realignment question. Although not having had occasion to adopt an explicit standard, we have stated that the Indianapolis rule requires realignment of parties "according to their real interests so as to produce an actual collision of interests." Lewis v. Odell, 503 F.2d 445, 447 (2d Cir.1974). Further, several district courts within this Circuit have employed the collision of interests approach when realigning parties. See Syms, Inc. v. IBI Sec. Serv., Inc., 586 F. Supp. 53, 56 (S.D.N.Y.1984); American Mut. Liab. Ins. Co. v. Flintkote Co., 565 F. Supp. 843, 846-47 (S.D.N.Y.1983); Irving Trust Co. v. Century Export & Import, S.A., 464 F. Supp. 1232, 1241 (S.D.N.Y.1979).

The primary purpose approach is not actually dictated by Indianapolis because though the facts of that case involved only a single controversy among the litigants involving the enforceability of a 99-year lease, the Supreme Court did not intend that all cases be forced into a single-issue posture. See Zurn Indus., 847 F.2d at 237. Indianapolis deliberately considered additional, subordinate controversies raised by the parties opposed to realignment and found that they were in fact non-issues. See 314 U.S. at 73 n. 3, 62 S. Ct. at 18-19 n. 3. Such discussion would have been wholly irrelevant were the realignment inquiry to concern only the primary purpose of the litigation. See Travelers Indem. Co. of Ill. v. Metropolitan Life Ins. Co., 798 F. Supp. 156, 158 (S.D.N.Y.1992).

Again, the collision of interests approach is consistent with the Supreme Court's chief concern in Indianapolis that parties not manipulate alignment to manufacture diversity jurisdiction. That Court directs us to examine "the realities of the record" to discover the "real interests" of the parties. Indianapolis, 314 U.S. at 69, 62 S.Ct. at 16. And, realignment is, of course, a fact-specific inquiry. The broader test we embrace does just this in its practical examination of the entire record, while the primary purpose test ignores "actual and substantial ancillary or secondary issues to the primary issue." Thomas Solvent, 955 F.2d at 1089.

The purpose in realigning parties is to make sure that there is a bona fide controversy between, as the statute commands, citizens of different states. In applying the collision of interests test therefore we must be mindful that actual and substantial conflicts in fact existed at the initiation of the lawsuit. Hypothetical conflicts manufactured by skillful counsel must not control because such an approach would reintroduce the notion of gamesmanship so disparaged by the Supreme Court.

In its effort to defeat diversity jurisdiction, Grace insists that the insurers are united in their efforts to deny Grace coverage. It states that inter-insurer disputes concerning contribution and indemnity are collateral because they will exist only to the extent that insurance coverage is established. This characterization of the facts we think ignores the practical reality of the litigation. Because the insurers are of a single mind to escape liability for paying claims does not mean that they are not in conflict with one another. In fact, the disputes among the insurers have been significant from the start of the lawsuit and produced the required collision of interests to sustain diversity.

Maryland's amended complaint suggests adversity among the insurers with the allegation that its liability "is directly affected by the interpretation of the rights and duties of Grace (or its predecessor companies) and each of the defendant insurers under their respective contracts of insurance." Disunity among the insurers is established further because Continental, Grace's current insurer, never has been adverse to Grace and, in fact, is actively pursuing a $60 million counterclaim against Maryland to be indemnified for money it paid to Grace. Continental, in addition, has filed cross-claims against Royal, Aetna and General. The existence of these claims suggests that realignment would be inappropriate even under the primary purpose test. Those courts that have employed the primary purpose test to realign parties in declaratory judgment insurance actions have relied on the fact that every insurer was antagonistic to the insured. See Thomas Solvent, 955 F.2d at 1091; Crown Cork, 942 F.2d at 866.

Concededly, each insurer--other than Continental--had a goal of escaping liability to Grace. But this interest does not eclipse the equally compelling interest each has to avoid liability to one another. See Trane, 657 F.2d at 150; Irving Trust, 464 F. Supp. at 1241. The insurance companies never doubted that some among their number would be held liable on their policies to Grace. But with millions of dollars at stake, which of them must pay or how far back coverage would extend inevitably pitted each insurer against all the others. Cf. Lumbermens Mut. Casualty Co. v. Connecticut Bank & Trust Co., 806 F.2d 411, 414 (2d Cir.1986) (affirming stay of federal insurance action in favor of pending state suit because "the determination of [insured's] coverage dispute with any particular insurer necessarily impacts upon the timing of the obligations of the other insurers"). Realignment hinges on those issues that divide the parties, not on those on which they agree. See American Motorists, 657 F.2d at 151.

Although the discovery trigger adopted by the magistrate judge relieves all insurers except Continental of liability, at the start of this litigation no insurer could have known what trigger would be applied and how far back in time that trigger would reach to make different insurers liable. The interest of the earlier insurers was only that coverage be triggered after their particular policies had expired, and this interest obviously did not extend to whether the next insurance company in line had to pay.

Recognizing that we examine collision of interests from the perspective of the lawsuit's initiation, the current posture of the parties nonetheless confirms that the insurers were at all times adverse. Although Maryland and Grace have settled, Maryland continues its suit against Royal, Continental and General. This fact belies Grace's contention that the insurers' adversity only comes into play once the fact of coverage is established because only Continental is presently liable to Grace for property damage claims. Hence, in applying the collision of interests test to determine realignment, we hold that actual and substantial controversies exist among the insurers and Grace to sustain diversity jurisdiction.

When it granted summary judgment, the magistrate judge ruled that the discovery trigger for property damage claims controlled as a matter of New York law. Under its ruling "the discovery of damage must occur during the policy period" in order to trigger the insurers' obligation to defend and indemnify Grace. Because Grace did not allege that any building owners in the underlying lawsuits discovered the presence and the hazard of asbestos in their structures before 1973 (Continental, as noted, has been the only insurer of Grace since 1973), there remained no material issue of fact regarding the insurers' obligation to defend or indemnify Grace in these disputes, that is, they had no such duty, and summary judgment was entered against the manufacturer denying coverage.

In reviewing the grant of summary judgment, substantive insurance law will determine which facts are material to the parties' dispute, and "[o]nly disputes over facts that might affect the outcome of the suit" will bar summary judgment. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S. Ct. 2505, 2510, 91 L. Ed. 2d 202 (1986). Because the adoption of a coverage trigger is a question of substantive insurance law, we review the magistrate judge's grant of such relief de novo.

We think that property damage insurance should be treated the same as insurance for bodily injury, which under New York law is governed by an "injury-in-fact" trigger. The relevant language in the insurance policies supports a damage-in-fact trigger for property damage claims. From adopting such a damage-in-fact trigger, it follows that insurers are obligated on the risk undertaken when asbestos was installed in the buildings involved in the underlying lawsuits against Grace. The reasons for which we reach that conclusion are set forth in the discussion that follows.

Because the question before us is one of contract interpretation, we begin by examining the language of the disputed insurance policies. The policies provide for "occurrence" coverage and, though their language varies slightly, their definitions of "occurrence" are similar. In this connection, we observe that throughout this litigation Royal repeatedly represented that it would be bound by the interpretation of an "occurrence trigger." It is now too late for it on appeal to attempt to change course and rely instead on accident-based language in its policies.

Thus, the meaning of the word occurrence is important because how that word is defined will determine what event will trigger the insurers' obligation to defend and/or indemnify Grace. See Abex Corp. v. Maryland Casualty Co., 790 F.2d 119, 124 (D.C.Cir.1986). The following language from Maryland insurance policies issued between 1961 and 1967 is representative:

"Occurrence" means either an accident or a continuous or repeated exposure to conditions which result during the policy period in injury to or destruction of (a) property including the loss of use thereof which is accidentally caused and (b) tangible or physical property, including the loss of use thereof. All damages arising out of such exposure to substantially the same general conditions shall be considered as arising out of one occurrence.

The above language incorporates a definition of property damage. The meaning of property damage is critical to our analysis because "for there to be coverage, there must be an occurrence, and for there to be an occurrence there must be property damage as defined in the policy." John P. Arness and Randall D. Eliason, Insurance Coverage for "Property Damage" in Asbestos and Other Toxic Tort Cases, 72 Va.L.Rev. 943, 951 (1986). The 1961-67 Maryland policy, the Royal policy and the General policy all define property damage as "injury to or destruction of property, including the loss of use thereof." The 1967-73 Maryland policy defines property damage simply as "injury to or destruction of property."

We turn first to the definition of occurrence. The magistrate judge--in a portion of its decision not appealed--adopted the injury-in-fact trigger for bodily injury claims under the Grace policies. This test obligates insurers on the risk when injury--as opposed to either exposure to disease-causing elements or diagnosis of injury--in fact began. The trial court relied on two federal cases interpreting New York law and adopting the injury-in-fact trigger for bodily injuries in the context of asbestos and pharmaceutical product liability, Abex, cited earlier, and American Home Prods. Corp. v. Liberty Mut. Ins. Co., 748 F.2d 760 (2d Cir.1984) (American Home I). See also Continental Casualty Co. v. Rapid-American Corp., 80 N.Y.2d 640, 651, 593 N.Y.S.2d 966, 609 N.E.2d 506 (1993) (New York Court of Appeals in an insurance coverage dispute for asbestos bodily injury claims applied the injury-in-fact trigger, "which rests on when the injury, sickness, disease or disability actually began," whether discovered or not); Cortland Pump & Equip. Inc. v. Fireman's Ins. Co., 194 A.D.2d 117, 121, 604 N.Y.S.2d 633 (3d Dept.1993) (adopting injury-in-fact trigger for an insurance contract covering property damage claims).

Both of the cases relied on by the trial court interpreted standard insurance policy language providing coverage on an "occurrence" basis, see Abex, 790 F.2d at 121; American Home I, 748 F.2d at 762, and found the contract language to be unambiguous. See Abex, 790 F.2d at 127; American Home I, 748 F.2d at 764. Interpretation of unambiguous contract language does not bring extrinsic evidence into play. See Seiden Assocs., Inc. v. ANC Holdings, Inc., 959 F.2d 425, 428 (2d Cir.1992). Because language of the disputed insurance policies is unambiguous, we need not consider arguments by Grace, Maryland and Royal with respect to extrinsic evidence of the drafters' intent.

We recognize of course that contracting parties may elect to define "occurrence" one way in the context of bodily injury and another way in the context of property damage. Although the insurers urge that we treat the occurrence definition for property damage differently than for bodily injury, their contracts have not done so. In the 1967-73 Maryland policies, for example, the same definition of occurrence is used for both types of injury, and at least some of the insurance issued by General uses the same definition of occurrence for both property damage and bodily injury. Moreover, the distinction the insurers seek is addressed more properly in the definition of property damage itself.

The language in the definitions of occurrence at issue is indistinguishable from that viewed as demanding application of the injury-in-fact trigger. See Rapid-American, 80 N.Y.2d at 651-52, 593 N.Y.S.2d 966, 609 N.E.2d 506; Abex, 790 F.2d at 125-27; American Home I, 748 F.2d at 764-65; Cortland Pump & Equip., 194 A.D.2d at 121, 604 N.Y.S.2d 633. By the plain language of the policies, injury must result during the policy period, but it need not be discovered during that time. See Continental Casualty Co., 80 N.Y.2d at 650-51, 593 N.Y.S.2d 966, 609 N.E.2d 506; American Home Prods. Corp. v. Liberty Mut. Ins. Co., 565 F. Supp. 1485, 1495 (S.D.N.Y.1983) (American Home II), modified, 748 F.2d 760 (2d Cir.1984).

New York courts also have held that injury must result during the insurance period to trigger coverage under occurrence-based policies. For instance, in National Casualty Ins. Co. v. City of Mount Vernon, 128 A.D.2d 332, 334, 515 N.Y.S.2d 267 (2d Dep't 1987), the Appellate Division had before it a policy that defined occurrence in language similar to that contained in the policies before us and that applied both to bodily injury and property damage claims. Rejecting the contention that an injury's cause must take place during the policy period, the court held that the injuries or damages resulting from the causative event must occur during the policy term for it to provide coverage. Id. at 336, 515 N.Y.S.2d 267. Accord Van Wyck Assocs. v. St. Paul Fire & Marine Ins. Co., 115 Misc. 2d 447, 450, 454 N.Y.S.2d 266 (N.Y.Sup.Ct.1982) ("[T]he policy does not include mere exposure to 'conditions' existent during the policy period, but rather focuses on the 'result' in 'bodily injury' during the policy period."), aff'd, 95 A.D.2d 989, 464 N.Y.S.2d 617 (2d Dep't 1983); American Motorists Ins. Co. v. E.R. Squibb & Sons, Inc., 95 Misc. 2d 222, 223, 406 N.Y.S.2d 658 (N.Y.Sup.Ct.1978) ("[C]overage is predicated not on the act which might give rise to ultimate liability, but upon the result."). But see Allstate Ins. Co. v. Colonial Realty Co., 121 Misc. 2d 640, 641, 468 N.Y.S.2d 800 (N.Y.Sup.Ct.1983) (holding that infant's exposure to lead-based paint chips was an occurrence).

The time at which injury was discovered was not at issue in these cases, but implicit in their holdings is the notion that injury can exist, or "result," independently of discovery or discoverability. Cf. Schultheis v. Centennial Ins. Co., 108 Misc. 2d 725, 727, 438 N.Y.S.2d 687 (N.Y.Sup.Ct.1981) (construing variant occurrence definition and holding that "[a]ny fair reading of the policy ... requires the conclusion that the covered risk does not depend on injury being discovered within the policy period"). We have stated, discussing personal injuries, that "[discoverability] need not coincide with the actual occurrence of injury; to add the requirement that an injury be [discoverable] limits the scope of the 'injury-in-fact' trigger-of-coverage clause in a way that is not justified by the policies' language." American Home I, 748 F.2d at 765-66. Consequently, under the occurrence definitions here at issue coverage is triggered upon the existence of property damage independent of its discovery.

In reaching a contrary conclusion, the magistrate judge relied on an unpublished New Jersey district court opinion applying New York law. It believed that the standard occurrence definition requires a discovery trigger in the property damage context and distinguished American Home I 's adoption of the injury in fact trigger on the ground that that case concerned bodily injury. The trial court reached this conclusion even though the disputed policies define occurrence in the same manner for both property damage and bodily injury. It also found that "[g]iven the long periods of time involved and the difficulty of measuring the exact time that damage to property might occur," the first discovery standard provided insurance carriers with more certainty. We think this reasoning flawed; first, because it fails to consider the plain language of the insurance policies and, second, as discussed in a moment, proper application of the injury-in-fact trigger does not create unduly burdensome uncertainty for carriers.

Thus, the definition of occurrence requires acceptance of an injury-in-fact trigger--perhaps better termed a damage-in-fact trigger here--in Grace's coverage disputes with its insurers. There remains a substantial question concerning correct application of this trigger, one that depends upon the definition of property damage. Only by clarifying the way in which asbestos injures a building can it be determined at what point actual damage occurs.

The insurance policies define property damage as "injury to or destruction of property, including the loss of use thereof." Under this definition, the injury to property need not be a physical injury. Acknowledging this, Royal and Maryland insist damage to property relates only to its market value, which suffers no decline until the owner discovers asbestos is present. Thus, according to these insurers, discovery of asbestos is the proper occurrence of property damage even under the damage-in-fact trigger. We are unable to agree with this proposition.

A reduction in marketability serves as a measure of damages, but it does not constitute an injury to property itself. Rather, it is an effect of injury. Under proper analysis, an event causes injury to property, and this injury in turn causes a decline in market value. In other words, the "decrease in market value merely reflects the recognition that something bad has happened to the building[, and that] 'something bad' is the incorporation of the defective product." Arness and Eliason, supra, at 955-56. Ignoring this point, Maryland avers that "property" is a legal abstraction and may suffer no damage other than a decline in the value of the legal rights tied to real estate. Royal suggests that lost market value affects only the property owners and actually is "wholly independent" of physical change in the property itself. Not only common sense but the underlying lawsuits belie these specious arguments. Building owners seek to remedy conditions directly concerning their structures either by removing or encapsulating asbestos. In making such changes, the owners will affect the buildings' market values. The damage that building owners are seeking to "undo" is not the fact that they discovered asbestos, but the fact of its incorporation in their buildings.

We rule therefore that damage-in-fact occurs upon installation in buildings of products containing asbestos. In interpreting an occurrence-based insurance policy that defined property damage as "injury to or destruction of tangible property," the New York Court of Appeals held that incorporation of a defective product into another product inflicts property damage. See Sturges Mfg. Co. v. Utica Mut. Ins. Co., 37 N.Y.2d 69, 72-73, 371 N.Y.S.2d 444, 332 N.E.2d 319 (1975); see also Marine Midland Servs. Corp. v. Samuel Kosoff & Sons, Inc., 60 A.D.2d 767, 768-69, 400 N.Y.S.2d 959 (4th Dep't 1977); Eljer Mfg., Inc. v. Liberty Mut. Ins. Co., 972 F.2d 805, 812 (7th Cir.1992) (applying New York law), cert. denied, --- U.S. ----, 113 S. Ct. 1646, 123 L. Ed. 2d 267 (1993). Market value is relevant only to the extent that the damage, i.e., the incorporation of asbestos products, must cause a decrease in market value of the entire product, here the building. See Sturges, 37 N.Y.2d at 71, 371 N.Y.S.2d 444, 332 N.E.2d 319; see also Arness and Eliason, supra, at 955-56.

We agree with the insurers when they insist that concepts of bodily injury may not be imported wholesale into the property damage context. Some types of property damage--such as the gradual contamination of earth and groundwater by leaking landfills--may be analogous to the slow progression of diseases such as asbestosis and cancer. See, e.g., New Castle County v. Continental Casualty Co., 725 F. Supp. 800, 809 (D.Del.1989), aff'd in part and rev'd in part, 933 F.2d 1162 (3d Cir.1991). Other sorts of situations, like the installation of asbestos building products, create a different kind of actual injury to property. These different circumstances must be carefully considered in attempting to determine when damage-in-fact occurs.

We fulfill this requirement by holding that installation of asbestos is an occurrence of damage-in-fact and triggers the insurance coverage in effect at that time. The actual injury to property--the presence of the asbestos hazard--occurs upon installation and exists regardless of whether it yet has been discovered by the building owners. See Stonewall Ins. Co. v. National Gypsum Co., No. 86 Civ. 9671, 1992 WL 123144, at * 14 (S.D.N.Y. May 27, 1992).

This is a case where measurable and compensable property damage has occurred, though not yet ascertained by the property owner. See American Home II, 565 F. Supp. at 1498. In fact, even where courts have adopted discovery triggers for property damage, they have acknowledged that damage actually existed before it was discovered. See Mraz v. Canadian Universal Ins. Co., 804 F.2d 1325, 1328 (4th Cir.1986); American Home Assurance Co. v. Libbey-Owens-Ford Co., 786 F.2d 22, 29 (1st Cir.1986). In those cases--concerning respectively the slow leaking of hazardous waste and the failure of a building's windows over an extended period of time--the difficulty of determining when damage in fact occurred influenced the trigger decision. In contrast to those factual situations, it is relatively easy--certainly not unduly burdensome--to determine at what point Grace's products were installed in the various buildings.

Next, it must be decided whether the injury to property continues after the discrete event of asbestos installation. If property damage is ongoing, then it is possible to trigger several successive insurance polices because the damage-in-fact occurs over a time continuum. See McGroarty v. Great Am. Ins. Co., 36 N.Y.2d 358, 365-66, 368 N.Y.S.2d 485, 329 N.E.2d 172 (1975). If such is the rule then injury need not be reduced to a single event fixed in time. See Mount Vernon, 128 A.D.2d at 337, 515 N.Y.S.2d 267. But asbestos property damage is unlike the gradual leaking of hazardous waste from a landfill or the gradual cracking and shifting of a building's foundation. Once installed, the damage that asbestos inflicts is complete. The underlying plaintiffs contend that asbestos fibers are constantly released and re-entrained into a building's atmosphere, creating further property damage. If such deterioration in fact exists, its damaging effect concerns solely the health of those persons who breathe the contaminated air. No further property damage occurs because the need to remove or encapsulate the asbestos, which occurred upon the product's installation, remains unchanged. See Arness and Eliason, supra, at 972-73; see also Stonewall, 1992 WL 123144, at * 17.

In their advocacy of a discovery trigger for asbestos property damage, the insurers also promote the need for certainty and a "bright-line test." Holding that damage-in-fact occurs only upon installation of asbestos products meets the goals of providing certainty for contracting parties and reducing administrative costs. Maryland nonetheless asserts that the discovery trigger is superior in this regard because it "helps to ensure that those policies were underwritten and issued when sufficiently current information was available to enable insurers to perform their predictive function, and to set premiums accordingly." This approach improperly would penalize Grace, a manufacturer that contracted and paid for insurance before 1973, because of the insurers' relative lack of information. Moreover, the very function of occurrence insurance recognizes the latent nature of the damages insured against and the possible insidious hazards of a manufacturer's products. American Home II, 565 F. Supp. at 1496; see also Eljer, 972 F.2d at 809 ("Once a risk becomes a certainty--once the large loss occurs--insurance has no function.").

Consequently, we hold that damage-in-fact occurs to property at the time asbestos products are installed, and injury to property does not continue after that event. Insurers on the risk at the time of installation are obligated under their policies with Grace.

The damage-in-fact trigger requires an examination of each underlying claim to determine when property damage--the installation of asbestos--occurred. See American Home I, 748 F.2d at 765. Importantly, the burden will remain with Grace to prove the existence of coverage. Abex, 790 F.2d at 129-30. Generally, the insured must prove the cause of the occurrence, the result, and that the result occurred during the policy period. American Home II, 565 F. Supp. at 1497. We are mindful of the fact that the occurrence policies at issue contemplate an injury resulting from a preceding cause, see American Home I, 748 F.2d at 764, but in the case of asbestos property damage, cause and effect are very close in time. On remand therefore the trial court must examine the facts of the underlying lawsuits to determine at what time Grace's asbestos products were incorporated into the various buildings that allegedly suffered property damage.

The federal courts have jurisdiction over this litigation because a sufficient collision of interests exists among the insurers to defeat Grace's request for realignment of the parties that would eliminate diversity. On the merits, we adopt the damage-in-fact trigger for the occurrence of property damage under Grace's insurance policies and reverse the magistrate judge's entry of summary judgment on this issue after it applied a discovery trigger for coverage. The case is remanded to that court for further proceedings consistent with this opinion.