While the question of trigger identifies which of the insurers may have an obligation to pay, most of the action in recent years has concerned the question of “how much”: how much does this particular insurer have to pay of the loss – is it responsible for the entirety of the plaintiff’s claim or only some portion? To the extent the insurer is obligated to pay, how do the limits of the insurance policy apply: is there an overall aggregate applicable to all asbestos claims or do different claims of different types of plaintiffs get bucketed into different hazards – and thus different aggregates or even no aggregate at all?

Another way for an insurer to limit its overall exposure – or potentially for a policyholder to avoid the application of multiple and cumulating deductibles or retentions – is to aggregate claims together on the ground they arise from a single “occurrence.” In that event, only one per-occurrence limit is available (or per-occurrence deductible required to be paid). Thus, even if the aggregate is not exhausted, the obligation to pay is limited to the single per occurrence limit. See Bomba v. State Farm Fire & Cas. Co. (N.J. Super. App. Div. Aug. 25, 2005)(holding that parents' negligence in allowing access to shotgun was a single occurrence, even though their son's use of the gun produced multiple victims).

The issues of aggregate and per-occurrence limits filter through into reinsurance disputes, where there have been a number of reported disputes concerning whether a carrier’s settlement with a policyholder is recoverable under a reinsurance policy: again, the context concerns per-occurrence retentions, allocation of loss to particular policies to which reinsurance attaches, and related issues. E.g., Travelers Casualty & Sur. Co. v. Gerling Global Reinsurance (2d Cir. Aug. 18, 2005).

Recently, the Third Circuit addressed the number-of-occurrences issue in the context of a policy that also had a non-cumulation of liability clause: a non-cumulation clause purports to confine the policyholder’s recovery to the single highest limit among the carrier’s triggered policies for a loss.

The Third Circuit’s case, Liberty Mutual Ins. Co. v. Treesdale Inc. (3d Cir. Aug. 15, 2005), involved a dispute where Liberty Mutual contended its obligation was limited to $5 million and the policyholder claimed Liberty Mutual owed $26 million.

The court applied the dominant “cause” test for determining the number of occurrences, that is, the number of occurrences is determined by the number of causes. The issue in Treesdale as in other asbestos cases is whether each worker’s exposure to asbestos is a separate occurrence or whether the company’s use of asbestos-containing materials in its products is “the” occurrence. The policyholder argued that the key language in the policy did not literally support batching all claims together: it noted that the manufacture and sale of asbestos-containing products – the cause of the losses according to the insurance company – did not itself constitute a “condition” from which injury could result or an “exposure.” Thus reasoned the policyholder the definition of occurrence – injurious exposure or the continuous or repeated exposure to a condition – could not be applied to the original decision to use asbestos.

The Third Circuit, however, turned mainly to the limit-of-liability section of the policy for the answer. Eschewing a literal application of the terms of the contract, the Court instead found that the language of the policy as a whole – and especially the limits section – indicated the intention that claims be grouped together more broadly than what the policyholder argued. The court found that “all injuries arising from the same source arise from one occurrence.” (Slip op. at 14).

This hardly answers the question presented but rather restates it: what is the “source” of the injuries. Other courts, like the California trial court in the noted Coordinated Asbestos litigation found that batching all asbestos claims together based on the original decision to use asbestos in the company’s operations is an unreasonable construction of the policy language, in part because that cause was too remote and is only a necessary but not sufficient cause to produce the injury in question. Granted that the policy does not apply, as the Third Circuit observed, on a per-person or per-claim basis, the plain truth is that the policy does not well answer the question of the number of occurrences.

Indeed, asking the question of the number of occurrences in the abstract is something that I’ve always found to be more confusing than elucidating. Much of most first-year torts classes focus on questions of “causation” and the differences among cause in fact, intervening causes, proximate cause, and legal causation – the latter being a conclusion rather than an empirical inquiry.

The better approach is to consider the “occurrence” question as one concerning the application of per-occurrence limits in an individual policy. Because insurance policies issued over time are independent agreements, each agreement should be read on its own terms – they are not codependent. Under this simple, contractarian approach, one looks at a triggered policy and then asks how the limit applies.

In the context of asbestos exposure, the cause of the person’s injury is his exposure to asbestos. Another “cause” can be said to be the company’s decision to use asbestos in its product, but it could also be said that any number of intermediate stopping points are part of the chain of causation. The question of which cause counts for determining the number of occurrences is not really answered by the policy language, nor are there inexorable insurance law rules on this point (to prove this assertion, consider whether the parties to contract out of any number-of-occurrence rule and the answer undoubtedly is that they can make whatever agreement suits them).

In the absence of a rule within the contract answering the question or a definitive insurance-law requirement (other than, perhaps, use of the “cause” test, though the language contemplates that approach), how should courts resolve the number-of-occurrence question?

As in questions of horizontal versus vertical exhaustion and other limits-related points, one should look at the policy language and fashion a theory that is sensible in the circumstances presented. What this means is that the court should adopt whatever result that the policy language permits – in other words, the policy should be construed against the drafter and in favor of coverage. I’m never very comfortable with arguments calling policy provisions “ambiguous” because I think that courts misunderstand what that claim means: I think courts often (mis)understand a claim of ambiguity to mean that the policy language is “hopelessly ambiguous” – that is, that one cannot make heads or tails of the policy language – which is a rhetorical mountain the policyholder does not need to climb. Instead, ambiguity is a lesser claim, viz., only that the policy language admits of more than one reasonable construction.

In the context of number-of-occurrences, the language is ambiguous in this sense: the policy does not ineluctably compel one result or the other, there is no meta-principle that resolves the matter, and no violence to contract language, insurance law, or public policy is worked from one result or the other. In those circumstances, insurance law (and contract law) dictate that the court adopt whichever result favors the policyholder, for the burden of ambiguity lies with the drafter who can protect itself through the selection of better contract language. Accordingly, no state should have a rule that all asbestos cases are one occurrence or are multiple occurrences – the answer should depend on the particular case. There is absolutely nothing wrong as a matter of the theory of insurance law to have different results on the same point where the policy language admits of more than one construction. This is one key difference between construing insurance policies and construing statutes.

The Third Circuit considered a secondary issue concerning policy limits: Liberty Mutual’s “noncumulation” clause – a policy provision whose intent manifestly is to prevent the cumulation of policy limits where more than one Liberty Mutual policy is triggered across time. The premise of this provision is that there can be a continuing occurrence spanning policy periods, and thus, in order to preclude more than one policy limit’s applying, each Liberty Mutual policy states that “if the same occurrence gives rise to person injury .. . which occurs party before and partly within any annual period of this policy, then each occurrence limit . . . shall be reduced by the amount of each payment made by the company with respect to each occurrence under [the earlier policy].”

This language contemplates that an occurrence will otherwise implicate the per-occurrence limit fully in each triggered policy. In such circumstances, the policyholder will collect more than one occurrence limit where injury continues beyond the policy period. (Note that the policy does not require a new occurrence in the policy period in question; rather, the occurrence can happen at any time but the policy responds only to the policyholder’s liability for injury during the policy period; if there is injury during the policy period, the policy is exposed to paying a per-occurrence limit (a result entirely consistent with taking each policy on its own terms).) The non-cumulation provision, however, results in only one total per-occurrence limit being available, even if injury occurs in more than one policy period. While the policy language employed by Liberty Mutual is not a model of clarity, the essential underwriting notion is a sensible reaction to how the policy language otherwise would work.

Of course, the noncumulation provision does not answer the question whether all asbestos claims are one occurrence or whether each worker’s exposure is a separate occurrence. Indeed, one can argue that because the worker’s occurrence is not complete until his actual exposure, the noncumulation provision seems inconsistent with the one-mega-occurrence approach adopted by the Third Circuit

Treesdale played into Liberty Mutual’s hand to some extent by adopting the unduly literalistic argument that the noncumulation should not apply because it intended to allocate the loss in reverse chronological order. In other words, because the noncumulation provision limits coverage in a later policy when an earlier policy is triggered, the policyholder argued that the provision was irrelevant if it was tapping its coverage first from the later policies and then the earlier ones. As the Third Circuit observed in rejecting the argument: “We are hardpressed to think that an insurance company would issue a policy with an anti-stacking provision, but intentionally include a provision that would void the anti-stacking provision by allowing the insured to invoke coverage in reverse chronological order.” (Slip op. at 27). Whatever the demerits to the noncumulation provision, sidestepping it via “reverse chronological exhaustion” reads the provision out of the policy. It is in fact more congenial simply to invalidate the provision on the grounds that the provision is abhorrent – which is what the New Jersey Supreme Court did in Spaulding Composites v. Aetna Cas. & Sur. Co., 819 A.2d 410 (NJ 2003). Carriers should welcome the reverse-chronology argument whenever it is advanced because it is so likely to wound the credibility of the policyholder and allow the carrier to capture the rhetorical high ground.

There are arguments to be made about noncumulation provisions, but it is important for the policyholder to advance a construction that gives some substance to the provision and bears some modicum of logic. It would be powerful to show that the pricing of the policy takes no account of the noncumulation provision and that this restriction on coverage was not explained in the marketing of the policy (and Liberty Mutual could have sold the provision as (i) confirming the policy’s intent and (ii) reducing the price of the coverage – but to my knowledge it never did that). Liberty Mutual surely knew what it was trying to do since some of its key personnel were intimately involved in the drafting of the 1966 CGL policy form (Gilbert Bean, for one, see Mayerson, Affording Coverage for Gradual Property Damage Under Standard Liability Insurance Policies: A History, 8 Coverage 3 (Sept./Oct. 1998)), and if Liberty Mutual was trying to slide the provision by its customers that would be grounds for its invalidation or nonenforcement on several grounds, including based on disproportionate forfeiture and other well-established contract doctrines concerning sharp practices.

The Treesdale court should have ruled that the number of occurrence question is ambiguous and thus adopt the result urged by the policyholder. That ruling would largely have mooted the noncumulation argument, but the court could have given life to that provision too by finding that where a single worker’s claim exceeded a per-occurrence limit and where the policyholder was electing to exhaust its coverage horizontally then the policyholder is unable to collect more from Liberty Mutual than the highest available per-occurrence limit among all triggered Liberty Mutual policies.

By its rulings, the court only adds uncertainty to coverage litigation by failing to adopt the principled, coherent approach to the number-of-occurrence question and instead ruling ex cathedral that the court knows an occurrence when it sees it without affording parties with clear principles for determining how the limits of the coverage apply.