Because insurance coverage disputes are governed by state, not federal, law, and because different states employ different tests to determine the number of occurrences, the answer to the question may differ depending on which state’s law is applied to the dispute. Under New York state law, for example, the number of occurrences under a given policy is not determined by the remote cause of the damage or liability, but rather by the immediate event that creates the loss or liability.

When the Number of Occurrences Matters
The number of occurrences under a given policy is only an issue if the policyholder faces multiple claims or losses arising from a given cause, event or series of events. The number of occurrences is most often an issue under two sets of circumstances.

One, the policy has a per occurrence limit that is lower than the policy’s aggregate limit (if there is one). In this case, the policyholder and the insurance companies providing umbrella or excess coverage advocate multiple occurrences in an effort to maximize the coverage available from the primary insurer.  

Two, the policy has a high self-insured retention, retrospective premium or deductible that is either not subject to an aggregate limit or has an aggregate limit that is not reached by the claims in question. In this case, the primary, umbrella and excess insurance companies tend to advocate multiple occurrences so that the claims at issue remain the responsibility of the policyholder.

Three Tests
Whether analyzing liability, property or employee dishonesty policies, courts asked to resolve the number of occurrences issue have generally applied one of three tests:

•cause (or negligent act or omission) test
•effects (or English) test
•unfortunate event (or liability triggering event) test

As a general rule, application of the cause test yields the minimum number of occurrences (frequently one), while the effects test yields the maximum possible number of occurrences (frequently one per claimant or instance of loss or property damage). The unfortunate event test tends to produce a number somewhere in between.

The Cause Test
The cause test has been adopted by a majority of states. These states have concluded that, where there are losses to more than one person or entity, the number of occurrences should be determined by looking at the underlying cause of the loss or liability. As long as multiple injuries or instances of property damage are the direct result of a single action or event, they are treated as one occurrence.

The courts justify not considering the individual instances of bodily injury or property damage as separate occurrences in part by the fact that policies provide coverage (and limits) on a per occurrence or per accident basis. By dint of the use of the words “occurrence” and “accident” rather than the word “claim,” the courts say policies show the intent of the parties to focus on the circumstances underlying the claims or losses rather than their individual components. Further evidence of the parties’ intention can be found in the policy’s definition of the term “occurrence,” which often includes language grouping together a “series of related acts” or a “continuous or repeated exposure to conditions.”

In addition, the limits of liability section in many policies contains restrictive language limiting the policyholder’s recovery to a per occurrence limit “regardless of the number of . . . persons or organizations who sustain injury or damages, or . . . claims made or suits brought.” This choice of wording also supports the position that the parties intended to gauge coverage not based on the individual lawsuits or properties damaged. These “unifying definitional principles,” the courts say, require the number of occurrences to be determined by looking at the underlying cause.

The cause test can best be understood by examining Owens-Illinois, Inc. v. Aetna Casualty and Surety Co. (1984). In the case, the U.S. District Court for the District of Columbia was asked to resolve the number of occurrences at issue for tens of thousands of lawsuits arising from the policyholder’s manufacture and sale of asbestos-containing thermal insulation over a ten-year period. The policies at issue were umbrella policies with deductibles of between $100,000 and $250,000 per occurrence. The parties agreed that the determination of the number of occurrences should be based on the cause of the injury, but they did not agree on what the cause was. The policyholder argued one cause (the decision to manufacture and sell asbestos-containing products), while the insurance company argued multiple causes (each claimant’s exposure to asbestos).

In a move that signaled the final outcome of the case, the court began its analysis by noting that treating each claimant as a separate occurrence would effectively deny the policyholder coverage because the deductibles were larger than any claim successfully brought against it. The court then examined the definition of occurrence, which contained standard form language (“continuous or repeated exposure to conditions”), and the unifying definitional provisions in the limits of liability sections. This language, the court concluded, mandated that any per occurrence analysis must focus on the underlying circumstances behind the claims rather than on the claims or lawsuits themselves.

Moreover, as the court pointed out, the insurance company’s position depended on establishing the circumstances behind each of tens of thousands of claims. The court rejected this approach as administratively unworkable and a clear distortion of the policy language.

Finally, the court looked at the language of the policy to determine the policyholder’s business purpose in purchasing the coverage. It found that the policyholder intended the coverage to relieve it from the risk of unknown liabilities. Since treating each claimant’s lawsuit as a separate occurrence would effectively eviscerate the coverage and deprive the policyholder of the security it had paid for, the court concluded that the policyholder’s asbestos-related liabilities all arose from a single occurrence.

The cause test has been adopted in what one court characterized as a long line of decisions holding that the manufacture or sale of a defective product constitutes a single occurrence, regardless of the number of people injured or pieces of property damaged by the product.

There are fewer decisions addressing the number of occurrences for other types of policies, such as employee dishonesty policies and property policies. (Because these policies typically contain per occurrence limits that are lower than any applicable policy aggregate, disputes generally have the insurance company arguing for a single cause and the policyholder urging multiple causes.) Under the cause test, courts analyzing a policyholder’s losses resulting from two dishonest employees working separately would find two occurrences. In the context of property insurance, two fires at two different places on the policyholder’s property, with two separate causal factors, would also be two occurrences under the cause test.

Massachusetts law is unclear as to whether its courts would apply the cause test when analyzing the number of occurrences. Although its courts appear to have applied this reasoning in previous decisions regarding the number of occurrences, none have affirmatively stated that they were adopting the cause test. If Massachusetts courts do apply this test, however, they could find that the events of September 11 were one occurrence, because they all resulted from a common, coordinated plan carried out by one group of terrorists.

The Effects Test
The effects test—clearly the minority test—was first used in the late nineteenth century. Under this test, courts look at the individual claimants or instances of property damage to determine the number of occurrences. The rationale behind the effects test is that, because insurance policies provide coverage for injury or loss, it is appropriate to equate “accident” or “occurrence” with “accident to any one claimant” or “occurrence harming any one claimant.”

Modern courts have adopted the effects test in part because they have difficulty accepting the theory that a pattern of behavior, conduct, inactivity or negligence by the policyholder justifies aggregating individual claims into a single occurrence. Courts often have difficulty categorizing an abstract pattern of behavior as a condition that would satisfy the continuous exposure clause. In addition, some view the effects test as more appropriate where the injuries or damage complained of arise at different locations and at different times, and therefore clearly do not constitute one occurrence.

In one case applying the effects test, Metropolitan Life Insurance Co. v. Aetna Casualty and Surety Co. (2001), the Connecticut Supreme Court examined the number of occurrences at issue in a series of asbestos-related bodily injury cases. Unlike Owens-Illinois, the claims underlying this lawsuit were based on the policyholder’s alleged failure to adequately publicize the health risks of asbestos exposure discovered through its medical research activities.

The court looked at the language of the policy and determined that the policyholder’s failure to warn could not be categorized as one occurrence because it did not fall within the definition of an event that took place “unexpectedly and without design.” Further, the court reasoned, the policy’s continuous exposure clause only combined claims into one occurrence when they arose from an exposure occurring at one place and at approximately the same time. The court rejected the one-occurrence idea that the claimants were continuously and repeatedly exposed to a failure to warn. It therefore concluded that each individual claimant’s exposure to asbestos amounted to a separate occurrence.

In light of the fact that the effects test is rarely applied, risk managers should only consider it relevant when facing claimants suffering injury or loss from liabilities that arose in many different circumstances over an extended period of time.

The Unfortunate Event Test
A number of jurisdictions, including New York, have rejected both the cause test and the effects test in favor of the unfortunate event test. This test is viewed as the most modern, because the concept of the unfortunate event as one of several possible factors contributing to injury or damage coincides with modern understanding of the scientific causation process. Under the traditional approach, causes were imagined as discrete, sequential incidents proceeding in a linear fashion to the ultimate injury. Such an approach makes it easy to attribute an event to one, isolated cause. A more realistic approach, however, is to look at the variety of interrelated and intermingled events as combining to affect the probability of other events.

This principle, which is at the heart of the unfortunate event test, dictates that the cause of injury or damage must be an event close to the injury or damage itself. More remote events, while certainly related to the ultimate injury or damage, merely cause the potential for liability or loss, and are not, in and of themselves, the direct cause—or unfortunate event—that results in the harm.

In many ways, the unfortunate event test is not wholly different from the cause test. Both justify aggregating individual claims or losses based on the policy language, especially on the per occurrence definition (if any) and the restrictive language in the limits of liability section; and both focus on the events that cause the liability or loss. Therefore, under both tests, a court would look at the same set of circumstances to determine the number of occurrences. But where the court applying the cause test might analyze the events near the beginning of the chain of causation—such as a negligent act or omission on the part of the policyholder—under the unfortunate event test it would analyze the events closer to the end of the causation chain (and closer to the actual injury or loss).

To make matters even more confusing, both unfortunate event test courts and cause test courts say that they are determining the “event” that “caused” the damage or liability. For the unfortunate event test, however, the cause is the immediate unfortunate event that resulted in the harm.

Negligent behavior, such as failure to warn of danger, cannot constitute an occurrence under the unfortunate event test because negligence does not cause injury or damage, it only creates the possibility for future injury. The actual occurrence, therefore, would be one of several other events that preceded and contributed to the resulting injury. In the products liability context, courts applying the unfortunate event test have found that the policyholder’s delivery of the product to its customers is the occurrence, because that is the last act over which the policyholder exercised control of the product.

In Arthur A. Johnson Corp. v. Indemnity Ins. Co. of North America (1959), New York’s highest court addressed how property damage caused by the collapse of two walls should be treated under a contractor’s liability policy written on a per occurrence basis. The policyholder had constructed two adjoining walls in front of two attached buildings. During a heavy rainfall, both walls fell within fifty minutes of each other, causing property damage to the two buildings for which the policyholder was held liable.

In analyzing the number of occurrences, the court considered and rejected both the cause test and the effects test, choosing instead the unfortunate event test. It determined that this test was the most practical, and that it corresponded with what the average person anticipates when buying and reading an insurance policy.

Under the unfortunate event test, the court concluded that the collapse of the two walls constituted two accidents because the property damage had occurred at separate times to separate buildings, and because the collapse of the first wall did not contribute to the collapse of the second wall. Further, as the court noted, if the walls and buildings were located blocks away from each other on different sites but subject to the same rainfall, “no one could contest that there were two accidents.”

In the property damage context, a California court applying the unfortunate event test has held that a series of four arsons committed by one person on four buildings were four separate occurrences, in part because each arson was a separate crime that would support a separate charge in a criminal proceeding. For property damage resulting from a natural event, a Connecticut court applying the unfortunate event test rejected as “ludicrous,” “def[ying] logic” and “particularly disingenuous” the insurance company’s position that two tornadoes that struck separate areas of the policyholder’s property within a forty-five minute period could not, as a matter of law, constitute two occurrences.

Although the ultimate resolution of the number of occurrences will depend on the language of the policies at issue, these decisions strongly suggest that a court applying the unfortunate event test would find that the World Trade Center attacks were two occurrences.

Ultimate Resolution
For the insurance coverage disputes arising from the events of September 11, 2001, the outcome of the occurrences question will depend on which state’s law is applied to the dispute. New York courts—or federal courts applying New York law—will decide this question using the unfortunate event test. Although Massachusetts law is less clear, its courts—or federal courts applying Massachusetts law—may use the cause test.

Whatever the ultimate resolution of this issue—and it will likely not be resolved for years—no one involved in the insurance industry will ever take the answer to this question for granted again.