Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Occurrence-Based Liability Insurance Trigger Interpretation

Abstract. Occurrence-based liability insurance assigns the possibility of a defense or indemnity obligation to policy periods by reference to bodily injury or property damage and the temporal requirements of a particular insurance contract. Its meaning becomes contested when an exposure, harmful process, physical injury, diagnosis, claim, judgment, and payment occur in different years. This monograph reconstructs occurrence-based trigger interpretation from policy wording, judicial opinions, comparative legal material, and actuarial reporting concepts. It distinguishes the event or condition that produces harm from injury during a policy term; separates exposure, injury-in-fact, manifestation, and continuous-injury approaches; compares the treatment of long-tail disease and progressive property damage; and analyzes the later, separate determination of all-sums or pro-rata indemnity allocation and primary or excess-layer exhaustion. The principal evidentiary base comprises published United States court decisions, United Kingdom Supreme Court cases and compulsory-employers-liability legislation, Canadian authority, reproduced commercial general liability wording, and selected actuarial and accounting sources. Reproducible hypothetical timelines and arithmetic show why the same damage estimate can produce substantially different insurer-year obligations under different attachment and allocation assumptions. The central scientific conclusion is that a reliable temporal coverage decision requires an explicit sequence of contract identification, injury chronology, applicable legal trigger, coverage defenses, allocation, limits, and reserve treatment. The work contributes a documented comparative synthesis and decision-reproducibility analysis; it establishes no new proprietary legal doctrine or actuarial theory.

Keywords: Occurrence-based insurance; casualty insurance; third-party liability; commercial general liability; insurance contract interpretation; trigger of coverage; policy period; bodily injury; property damage; exposure; injury in fact; manifestation; continuous injury; asbestos; environmental contamination; long-tail claims; pro-rata allocation; all-sums allocation; insurance limits; excess insurance; duty to defend; actuarial reserving; comparative insurance law.

Introduction: the temporal attachment problem

An insured business may produce a substance in one year, expose individuals to that substance over several years, learn of a diagnosis decades later, receive a lawsuit another year, and satisfy the resulting settlement still later. A building contractor may complete an installation during one policy term while hidden water intrusion begins damaging materials in another. A spill may occur at a particular location on one day and continue to injure property through several renewals. Each chronology presents a question of temporal connection: which insurance contract, if any, responds to an injured third party's claim? The inquiry is central because successive annual policies may be issued by different insurers, carry different exclusions and deductibles, or have different liability limits.

The phrase trigger of coverage is legal shorthand for the timing requirements that bring a given policy's potential coverage into consideration. The Supreme Court of California's Montrose trigger analysis identifies exposure, injury-in-fact, manifestation, and continuous injury as competing approaches developed for certain long-tail losses. Crucially, trigger is an analytical description rather than an independently operative contractual promise. The wording, the applicable jurisdiction's interpretive law, and the evidence of injury govern whether a policy responds. The same decision distinguishes the possibility of defense from a final adjudication of indemnity and treats allocation among carriers as a further issue (Montrose Chemical Corp. v. Admiral Insurance Co., 1995).

This dossier asks three linked questions. First, how do occurrence, bodily injury, property damage, and the policy period connect under the controlling insurance form? Second, which dates and evidentiary propositions are required for an instantaneous accident, progressive injury, and latent disease? Third, after identifying implicated policies, which legal and contractual rules distribute a covered liability among annual periods and available limits? The answers have distinct epistemic status: an actual policy clause is documentary evidence of agreed text; a reported judgment establishes a legal holding in its jurisdiction and factual setting; a medical or engineering reconstruction is a potentially contestable finding; and an actuarial allocation is conditional on specified legal and financial premises.

The research uses documentary and comparative methods. The core United States decisions are selected because they articulate materially distinct interpretive approaches, not because they form a statistically representative sample of all states. The United Kingdom and Canada supply comparisons in which policy terminology, compulsory insurance arrangements, and judicial reasoning differ. Statements about contemporary standardized insurance wording rely on a reproduced 2013 commercial general liability coverage form in Locke's insurance-form source compilation; an executed specimen policy may differ through manuscript endorsements, retrospective clauses, and jurisdictional law. Accounting and reserving literature is used to study the consequences of attachment uncertainty, rather than to decide contractual coverage. Every hypothetical figure later in this work is labeled and independently recalculable.

Occurrence and injury timing

Reconstructing the chronology

The first analytical step is to construct a timeline independently of coverage opinions. Consider a hypothetical manufacturer whose product was sold in 2017; a hazardous exposure occurred in 2018; biological injury probably developed between 2019 and 2021; first symptoms were detected in 2023; the claimant filed suit in 2025; and an insurer paid settlement in 2027. Five different dates or intervals appear. An investigator must determine which facts each date establishes. The product sale establishes distribution; exposure establishes contact; injury evidence supports actual harm; diagnosis establishes recognition; filing and settlement establish procedural and financial events. Their chronological ordering does not decide whether injury occurred throughout the intervening period.

Under an injury-based CGL grant, the relevant factual interval is bodily injury or property damage during the policy term. The American Home Products appellate opinion explains why diagnosis and diagnosability do not set an automatic boundary on injury-in-fact: evidence may establish prior injury retrospectively. Under Don's Building Supply, the Texas Supreme Court held that physical property damage occurs when the property is actually injured, even if nobody discovers it until later; the court distinguished proof that deterioration occurred from assumptions about when the defect was created. These decisions concern actual harm, but the practical proof can differ dramatically between medicine and construction engineering (American Home Products, 1984; Don's Building Supply, 2008).

A chronological data table is often more useful than one asserted trigger date. The analyst should record start and end of potentially harmful conduct, first verified contact, first established physical change, later ongoing changes, discovery and claim notice, each supported by its source and confidence. Where two expert reconstructions differ, both should be retained with their assumptions. The degree of temporal granularity should match the quality of evidence: day-level precision is unwarranted when an expert can establish only that moisture damage existed within an eighteen-month interval.

Exposure as an evidentiary event

Exposure identifies contact between a person or property and an injurious condition. It is an essential causal datum for many long-tail losses but does not necessarily prove physical injury under every policy. One worker's inhalation exposure may occur years before recognized disease, while a chemical discharge might physically damage soil or groundwater on contact. A court may adopt exposure-based attachment for a particular disease and wording because the harmful process is understood to begin at exposure. The court's reasoning, medical evidence, and legal scope must be stated, because treating any presence of a hazard as established bodily injury would conflate risk with realized loss.

In asbestos litigation, the Keene multiple-trigger decision recognized multiple relevant periods for progressively developing disease. Another jurisdiction may formulate an injury-in-fact rule or continuous-trigger presumption differently. A historical file should record exposure dates, occupation and product, dose where documented, latency, diagnostic evidence, and the policyholder's insured operations. These facts can be combined with medical opinions to identify injury intervals, but a numerical exposure model does not itself establish the coverage law.

Injury-in-fact and retrospective inference

The injury-in-fact approach asks when actual bodily injury or physical property damage existed, regardless of when it was observed. Its apparent conceptual clarity hides an empirical challenge: a diagnosis often identifies the condition at the date of clinical observation while leaving uncertainty about its first biological onset. Historical tissue samples, imaging, occupational records, and accepted disease mechanisms may narrow the interval but seldom yield a unique calendar day. Courts can assign burdens of proof or permissible inferences under relevant law; an actuarial or scientific reconstruction should document those assumptions rather than silently embedding them in a single point estimate.

The American Home Products decision is particularly valuable because it recognizes retrospectively provable injury without requiring the claimant to have had diagnosable disease at the disputed policy date. It does not establish that exposure always creates injury instantly. Conversely, the Texas property-damage holding rejects discovery as a universal surrogate for the date physical damage first occurs. Both illustrate the difference between the date the analyst can observe today and the earlier date at which the insured peril may have produced injury.

Manifestation, knowledge, and preexisting damage

A manifestation threshold refers to the time harm becomes apparent or is discovered, depending on the legal or contractual formulation. Its administrative attractiveness is obvious: a first diagnosis, inspection, or documented complaint often has a clear date. Yet that clarity may be purchased by losing correspondence with the insurance wording. The Montrose court specifically declined to use a first-party manifestation rule as the answer for progressive third-party liability damages under the policies before it. Thus manifestation cannot be treated as a universal default for old occurrence policies.

Timing of an insured's knowledge requires its own analysis. Some later CGL forms condition the grant or coverage continuation on what the insured or specified responsible persons knew before inception. The reproduced 2013 CGL form contains such known-injury or known-damage provisions. Other doctrines concerning fortuity and known loss also operate in particular jurisdictions. A 1995 court discussing the contingency of unresolved future injury under its historical form should not be used as proof that a later policy lacks a prior-knowledge restriction. The correct sequence is to identify the version and apply both the injury timing and relevant knowledge terms.

Accident, cause, and effects

An accident can be a discrete event or a repeated exposure condition according to the policy definition. Whether an insured's negligent act occurred in 2016 may be irrelevant to a 2019 policy if injury did not occur in 2019, yet the same act may remain causally central to determining whether the insured is liable. Likewise, a gradual process may involve one causative condition while physically damaging different properties at different times. The policyholder's liability narrative, the insurer's temporal attachment narrative, and the causal proof should be kept consistent but separately legible.

For immediate harm—an explosion injuring a visitor at the insured premises on one day—the exposure, injury, discovery, and first notice may cluster together, making trigger disagreement relatively modest. For a construction defect where faulty work is finished in 2020 but dampness damages drywall during 2022–2024, the works-completion date alone is an unreliable substitute for physical damage. Under the Don's Building Supply analysis, the fact of physical injury over the relevant period remains necessary. Different completed-operations clauses and exclusions may independently narrow coverage.

Defense timing and the quality of allegations

A defense obligation may be triggered by a complaint that alleges potentially covered injury during a policy period even before the insurer proves its final occurrence date. Governing law may use the pleadings, extrinsic facts, or a mixture of both, and some policies have specialized defense arrangements. The Montrose duty-to-defend holding rested on the potential for ongoing covered injuries; the Progressive Homes reasoning likewise illustrates why allegations capable of falling within coverage can warrant a defense under the Canadian policy and law examined. An indemnity decision is ultimately concerned with what liability and covered facts are established or settled within the contractual framework.

When a claim spans three alleged years, a liability insurer cannot always defer defense merely because it disputes which year will bear final indemnity. Nor can a defense expenditure be used as proof that a given year owes the underlying damages. Defense costs may fall within or outside limits, and contribution or reimbursement provisions may change the ultimate distribution. This is one reason legal coverage databases need separate fields for potential_defense_trigger, indemnity_trigger_established, and indemnity_allocation_decided, even when software displays one overall claim status.

Continuous injury and multiple policy periods

Physical continuity and the problem of overlapping insurance years

A continuing harmful condition can produce a sequence of independently distinguishable harms, one gradually evolving injury, or multiple injuries to different claimants. The temporal evidence must establish which of these is being investigated. For physical property, contamination may spread from a source into groundwater, remain stationary at another site, and undergo remediation in later years. A claim for cleanup costs can encompass past physical injury, preventive measures for future injury, monitoring costs, and administrative expenditure that is not itself physical property damage. The analyst should map each expense to the legal basis on which the third party seeks damages and then to the covered harm; labeling every expenditure a continuation of the same occurrence prematurely resolves both causation and insurance coverage.

The Montrose continuous-injury ruling recognized that successive policies could be potentially implicated where bodily injury or property damage continued or progressively deteriorated through their terms. It also emphasized that the insured still had to establish the facts necessary to obtain indemnity; the holding concerned the potential duty to defend allegations and did not dispense with exclusions, coverage limits, or proof. This distinction is especially important when a claimant argues that an environmentally harmful condition persisted for decades: the existence of a contaminant on site and the occurrence of new or continuing physical injury are related evidentiary propositions but require attention to the court's reasoning.

Multiple triggered periods create a temporal insurance portfolio. Each period may have its own insurer, deductible, retention, primary limits, aggregate limits, exclusions, notice conditions, and excess tower. The parties may have chosen different contractual wording at renewal, or the same policy wording may be interpreted under a law that evolved after the policy was issued. Thus it is methodologically wrong to treat the presence of a named insurer in an historical period as a guarantee of recoverable indemnity for that period. A triggered grant is an entry point to the remaining coverage analysis.

Long-tail bodily injury: the asbestos decisions

Asbestos litigation illustrates the temporal complexity of progressive disease. Occupational exposure can continue across years; disease development can remain clinically silent; the first manifest illness, diagnosis, or claim may occur after all exposure-related policies expire. The Keene appellate judgment examined this structure and recognized injury across more than one stage of the disease process. The legal reasoning was not a biological finding that all possible diseases in all exposed workers follow an identical timetable. Each claimant still has a distinct medical history and underlying cause of action.

The J.H. France decision illustrates an all-sums response under the Pennsylvania interpretation of asbestos-related coverage: the insured was permitted to select among triggered policies subject to the operative wording and limits, rather than being compelled simply to average indemnity across all years. The Owens-Illinois judgment reached a notably different allocation outcome under New Jersey law while also recognizing continuous trigger. These cases together demonstrate that agreement on a broad continuous-injury trigger does not imply agreement about the financial distribution that follows. One must identify precisely which legal issue the court decides.

Industrial disease also calls for a distinction between exposure period and the duration of injury. A worker exposed from 1978 to 1981 could develop recognizable disease much later; a conclusion that all later years are triggered requires its own legal and factual premise. In some asbestos cases the courts employ evidentiary or legal presumptions around progression, but those presumptions do not support an unlimited assertion about every pathology. A reproducible file records exposure history, first medical evidence, plausible periods of tissue injury, the legal trigger test, and which inferences the court actually adopted.

Long-tail property damage: environmental and construction claims

Environmental claims present another combination of spread, migration, and deterioration. Pollution may enter groundwater at a discrete date, remain present for decades, or migrate to additional parcels. Some resulting physical changes can be tested with geology, hydrogeology, aerial imagery, monitoring wells, and contemporaneous records. Other portions of the alleged harm are reconstructed from sparse historical information. The resulting uncertainty is substantive: a one-year trigger conclusion may depend on whether an aquifer was injured when contamination first arrived, whether later propagation constituted additional covered injury, and whether the liability pertains to covered property damage as the contract defines it.

Under the Boston Gas allocation opinion, the Massachusetts high court discussed progressive environmental damage and the reasons for allocating indemnity proportionally among years. Its approach underscores that making several years available for coverage does not require every year to pay the full total. Consolidated Edison's New York reasoning also connected the allocation rule to wording limiting the grant to injury occurring during a particular policy period. That legal analysis can coexist with expert uncertainty about what fraction of contamination actually arose in each period; the legal denominator and the scientific measurement problem must be stated separately.

A construction defect may be completed long before consequential physical injury becomes observable. Under the Texas Don's Building Supply opinion, the relevant occurrence of property damage is physical injury, independent of the time of discovery. The outcome is not a general insurance guarantee for the cost of repairing the defective work itself. The policy may distinguish an insured's own work from damage to other property, and exemptions, business-risk exclusions, and endorsements may have significant effects. As a matter of research design, date of faulty installation, date of later physical damage, and date of repair demand should each have their own data field.

Multiple claimants and harm aggregation

A facility may face one action seeking remediation of several contamination sites, several actions concerning the same plume, or successive bodily injury claims from different workers. The legal concept of an occurrence used for the number of deductibles and per-occurrence limits can interact with but remains separate from the choice of triggered years. A single cause may generate property injury over several periods, while multiple causally distinct discharges may produce related injury in the same period. Policy definitions can group or separate incidents differently, and the exact aggregation clause may impose limits independently of the broad historical narrative.

The temporal unit of measurement must therefore be specified. For a single claimant's progressive disease, the injury interval may be analyzed at the individual level. For a portfolio of claims, aggregate injury histories may overlap in ways that do not justify a single common start date. An insurer reserving a portfolio must preserve individual claim dependencies as well as plausible group-level experience. One generic date of occurrence field cannot simultaneously represent exposure, injury in fact, a group aggregate, diagnosis, and a reporting lag.

The UK mesothelioma comparison

The United Kingdom provides an instructive comparison because some historical employers' liability policies used formulations concerning injury sustained or disease contracted during the insurance period. In the BAI v. Durham judgment of March 28, 2012, the UK Supreme Court considered exposure to asbestos and later mesothelioma in disputes over those historical wordings. Its construction treated the policies in their particular employer-insurance setting and the statutory and commercial background. The decision cannot be transposed by name into every general liability policy using a different injury-occurs clause.

The later Zurich v. International Energy Group litigation addressed an employee's mesothelioma injury across periods involving different insurers and an employer bearing some unprotected years. The Supreme Court considered indemnity and contribution in the setting of mesothelioma-specific law, including a Guernsey employer and the distinction between Guernsey and United Kingdom compulsory-employers-liability legislation. The official judgment specifies that the historical Guernsey coverage and statutory setting differed from the corresponding Great Britain framework. This comparison is valuable because it shows the potentially substantial consequences of an exposure-based legal response, statutory protections for affected employees, and distribution among contributing parties. Its outcome is linked to the special causation and compensation context, rather than a free-standing rule of international allocation.

The statutory anchor is the Employers' Liability (Compulsory Insurance) Act 1969, including its amendment history and coverage scope. It addresses the existence of compulsory coverage, while policy wording, judicial decisions, employers' obligations, and the specific time of exposure or injury determine the actual indemnity question. An international study that reduces the UK history to a generalized United States-style continuous trigger would erase the very distinctions that make the comparison informative.

A Canadian perspective on injury and the defense obligation

Canadian commercial liability interpretation has its own statutory, contractual, and judicial structure. The Supreme Court of Canada's Progressive Homes decision concerned alleged construction deficiencies, property damage, the interpretation of accident, and the insurer's duty to defend. It is a useful reminder that the underlying legal conception of accidental occurrence and the scope of alleged property damage may determine whether the temporal issue is reached at all. It does not furnish a standardized Canada-wide formula assigning every latent defect to a particular year.

The comparison also shows why research should not place every reported judicial decision into a single table of trigger rules without an issue column. One decision can answer whether a complaint potentially asserts property damage, another whether that damage happened in the policy period, another whether the resulting obligation is excluded, and still another how several implicated insurers share an indemnity. All are relevant to an occurrence-based dispute, but they operate at different stages.

Allocation across insured years

Attachment and indemnity allocation are separate legal decisions

Temporal attachment determines which annual policies might respond; allocation determines how a covered loss is distributed among them or charged to the insured. The distinction is clearest where several policies are triggered by one progressively deteriorating injury. An all-sums reading may permit an insured to seek its covered liability, subject to the limits and conditions of a chosen responding policy; a pro-rata reading may assign each year a portion based on its share of the injury period, its limits, or another legally appropriate measure. Neither method expands the underlying claimant's legally compensable damage, and neither automatically eliminates applicable deductibles, retentions, or exclusions. Academic perspectives on allocation also disagree. Hogg (1998) argues from a policyholder-centered reading of historical occurrence forms that some pro-rata allocation approaches impermissibly reduce coverage; this law-review critique of allocation is used as evidence of a contested scholarly position, rather than as a binding rule. Its objection must be evaluated against specific policy wording and later controlling authority.

Consolidated Edison interpreted New York policies whose coverage was tied to injury during the period and adopted a pro-rata approach consistent with that wording. Owens-Illinois developed an approach using years and policy limits in allocating continuous-injury loss in New Jersey. By contrast, J.H. France demonstrates a policyholder election among triggered insurers in Pennsylvania asbestos litigation. These outcomes cannot be averaged into a national formula. A sound comparative analysis identifies the governing state, choice-of-law conclusion, exact clause, policy period, and loss facts before classifying an approach.

The resulting distinction can be written conceptually. Let T_j denote whether policy year j is potentially implicated under its coverage grant and the applicable trigger test. Let D denote the amount of covered damages eventually established for the insured, before contractual allocation and limits. Then T_j is a coverage eligibility proposition; A_j(D) is a conditional allocation of the liability to year j; and P_j is the amount ultimately payable after applicable retention, exclusions, limits, exhaustion, contribution, and any other conditions. The analysis must not assume that T_j = 1 entails either A_j(D) = D or P_j = D. This notation is an explanatory bookkeeping distinction, not a proposed new legal theory or an actuarial valuation system.

Pro-rata allocation: time on risk

A simple pro-rata time-on-risk illustration partitions a covered injury interval into portions attributable to each year. When the relevant legal rule uses equal annual weights and there are n equally long years, a stylized annual allocation is D/n. More generally, if eligible periods are measured by exposure duration t_j, the proportional share is D × t_j / Σ t_j over the relevant denominator. This is only an arithmetic representation of one family of judicial allocation choices. Certain jurisdictions use relative policy limits as weights, others adopt different premises, and different treatment may apply to uninsured years.

For example, Boston Gas considered a time-on-the-risk approach to environmental contamination and the allocation of uninsured periods. KeySpan's New York decision held that, under the particular pro-rata regime before it, unavailability of liability insurance for earlier contamination years did not excuse assigning the policyholder those years. This conclusion is distinct from a failure to purchase otherwise available insurance. In an unavailability argument, the asserted cause of a gap is the market; in a voluntary gap, the asserted cause is the policyholder's purchase decision. A court may decide that these differences do or do not affect allocation. The jurisdiction-specific reasoning, and not an assumed policy preference, controls.

Pro-rata allocation: weighting by limits

The Owens-Illinois opinion is influential for allocating losses across triggered periods using a combination of time on the risk and the limits available in those periods. The method recognizes that unequal amounts of insurance were purchased in different years, and it addresses the policyholder's participation in periods when insurance was available but absent. The precise operative formula, treatment of excess layers, and limitations must be derived from the governing case and any later decisions in that state; a generic spreadsheet with a fixed weighting rule is not a judicial holding.

Suppose purely hypothetically that the same two years of physical injury each have one year of exposure, but the insured purchased a $1 million indemnity limit in the first and a $3 million limit in the second. A limits-times-duration illustration assigns weights 1 × 1 and 3 × 1, resulting in relative weights of 25 percent and 75 percent. A simple time-only formula would yield 50 percent each. The comparison demonstrates that a change in legal allocation convention can affect financial shares without changing when the actual injury occurred. It does not claim that a court must allocate the selected loss by these weights, nor that payment may exceed either year's policy limit.

All sums, selection of a responding year, and the limits of the phrase

Under some interpretations, a triggered policy may provide indemnity for the insured's entire covered liability arising from a continuing injury, up to its own applicable limits, even though some injury occurred outside that policy's dates. The Keene decision and J.H. France judgment illustrate the importance of this approach for asbestos claims. This is often called all sums because of particular grants to pay all sums for which the insured is legally liable as damages. Yet the words all sums alone do not remove conditions, exclusions, aggregate limits, exhaustion requirements, or the provisions of other policies. Nor does this approach mean that the policyholder receives duplicate reimbursement for the same covered damages.

All-sums decisions can coexist with complex carrier contribution disputes. A policyholder may choose an insurer for indemnity while that insurer seeks contribution from others according to contractual or equitable rules. The insured's recovery position and the ultimate distribution among insurers need separate columns. The Viking Pump court examined noncumulation clauses and prior-insurance provisions in excess policies and explained how those provisions altered the allocation analysis. The result was specific to that policy language and did not resolve all temporal-trigger issues. The scholarly lesson is that all-sums versus pro-rata classification depends on the legal interpretation of the complete text, not simply the existence of continuous injury.

Primary and excess layers: exhaustion

A long-tail claim can pierce several insurance layers. A primary carrier may owe defense and initial indemnity under its terms. An excess carrier typically responds after a specified underlying attachment point, but the identity of the relevant underlying insurance and the required form of exhaustion depend on the policy. Horizontal exhaustion broadly refers to exhaustion of available primary coverage across implicated years before accessing excess insurance; vertical exhaustion broadly permits access to an excess policy above an exhausted underlying year or tower, subject to wording and governing law. These descriptions summarize recurring coverage disputes, not universal defaults.

In Viking Pump, the New York Court of Appeals held that the excess policies at issue supported vertical exhaustion, in connection with its all-sums and noncumulation analysis. The California Supreme Court's later Montrose excess-insurance decision also adopted vertical exhaustion under the circumstances and policy language it considered. Neither holding should be treated as a determination that the same insurer must fund all historical asbestos or pollution damages, and neither replaces the first-stage injury-timing inquiry. A rigorous computation first identifies responding contracts, then determines the amount assigned to a year or selected policy, and only then applies deductible, retention, aggregate, and excess attachment rules.

Hypothetical allocation study: six years of progressive property damage

Assume an environmentally related bodily-injury-or-property-damage claim is fully supported as to liability, and the independently established amount of potentially covered property damage is $12 million. Further assume that continuous physical property injury existed during six complete equal-length calendar years, 2018–2023, and that a jurisdiction applies a six-year time-on-risk pro-rata rule with no exceptions, assigning uninsured time to the insured. Two hypothetical insurers issued policies in 2018–2019 and 2022–2023 respectively, while the insured had no applicable liability coverage in 2020–2021. For the sake of isolating temporal allocation, assume each insured year has at least $2 million available indemnity capacity after other terms, with no deductibles or exclusions and with legal expenses treated separately.

Injury year Policy status Time share Illustrative liability assignment
2018 Insured A 1/6 $2,000,000
2019 Insured A 1/6 $2,000,000
2020 Uninsured 1/6 $2,000,000 retained
2021 Uninsured 1/6 $2,000,000 retained
2022 Insured B 1/6 $2,000,000
2023 Insured B 1/6 $2,000,000
Total Six years 6/6 $12,000,000

The calculation is 12,000,000 × (1/6) = 2,000,000 per year. The two hypothetical carriers collectively receive an $8 million assignment across four years, and the policyholder retains $4 million for two uninsured years, before any actual policy terms are applied. This illustrates the effect of an assumed pro-rata regime only. Under an all-sums interpretation, one selected responding policy could in principle be asked to meet a larger share up to that policy's applicable limit; access to other layers and contribution would then require separate resolution. Without information about the actual coverage forms, no numeric all-sums insurance recovery can be derived from this table.

To test sensitivity, hold the $12 million covered amount fixed but suppose the evidence now supports injury only from 2020 through 2023, while keeping the simplified pro-rata rule and other assumptions. The denominator becomes four years and each year receives $3 million. The hypothetical two uninsured years then account for $6 million, and the insured years 2022–2023 for another $6 million, before capacity limits. Thus a change in the supported injury interval alters the unadjusted allocation from an $8 million insured share to a $6 million insured share while the overall covered loss remains $12 million. This difference is a consequence of changed factual and legal assumptions rather than a change in liability damages.

These examples deliberately omit policy-year aggregate limits, insurer insolvency, retrospective premiums, taxes, potential reinsurance recovery, deductibles, and discounting. They preserve their analytic usefulness by making the omissions explicit. A numerical illustration that silently assumes all these quantities away would be unsuitable as evidence for an actual insurer's liability.

A second sensitivity test: limits and noncumulative clauses

Consider another purely hypothetical two-year continuous injury and a covered liability of $2.4 million. Under an equal-duration two-year pro-rata assumption, each year is assigned $1.2 million. Now suppose only $1 million of applicable per-year coverage remains available in the first year and $2 million in the second. Before resolving any legal provision allowing an unused first-year share to transfer, the first year's payable amount is bounded above by $1 million; the $200,000 assigned beyond capacity requires separate analysis. The second year is within its nominal capacity. The allocation plus limits calculation therefore produces at most $2.2 million across the two individual assignments on those particular assumptions, leaving $200,000 unresolved by the stipulated coverage.

Under an all-sums alternative, if the second year's policy independently covers the entire continuing loss and carries an unexhausted applicable limit of $2.5 million, that selected policy could potentially meet the $2.4 million liability, subject to its wording and the relevant law. In contrast, if a noncumulation clause restricts cumulative recoveries across certain consecutive contracts, simply adding annual limits would overstate available cover. The Viking Pump holding demonstrates that such clauses can materially change the allocation analysis. These examples show the distinction between liability estimate, annual attachment, assignment, and the enforceable maximum payable under the applicable contract.

Reinsurance, subrogation, and accounting interfaces

An insurer's reinsurance treaty may distinguish accident year, occurrence year, risks attaching, losses occurring, or claims-made structures. The insurer's liability to its policyholder and the reinsurance counterparty's reimbursement obligation may therefore be governed by different temporal definitions and different wording. A direct insurer cannot infer reinsurance coverage from the fact that its direct policy responds. Similarly, subrogation or contribution may shift the eventual net loss after an insurer has indemnified, without changing the injury date or the direct insured's contractual attachment.

The Casualty Actuarial Society research summary illustrates long-tail asbestos and environmental liability as an estimation challenge for reinsurers. It supports an analytic distinction between reported claim developments and uncertain ultimate loss, although the cited accessible source is a research landing page rather than independent reproduction of all model results. Contract-specific reinsurance conclusions require treaty documentation, occurrence definitions, attachment, limits, commutation terms, and historical recoveries; they cannot be inferred from the primary insurance policy alone.

Evidence, uncertainty, actuarial implications, and validation

A reproducible evidence protocol

An occurrence-trigger file should be auditable from source documents to its ultimate policy-year conclusion. The minimum evidence trail comprises the executed contract and version history; primary pleadings and underlying orders; a chronology of conduct, exposure, and injury; an explanation of causation and physical harm; the applicable legal authorities; and separate computations for indemnity assignment, defense, contractual limits, and recoveries. Each material finding should identify both the proposition established and the proposition still uncertain. For example, a medical record dated 2022 may reliably establish diagnosis in 2022 while supplying only interval-censored evidence of biological onset before that date.

The source hierarchy is especially important in legal research. Published opinions have fixed holdings but may summarize or omit evidentiary details from the trial record. An opinion that decides a duty to defend based on allegations establishes a potential coverage result rather than the truth of every allegation. The Montrose judicial opinion explicitly preserves factual questions for the ultimate coverage determination. Likewise, the Progressive Homes judgment addresses allegations and the duty to defend under the conditions before that court. Accordingly, a source-to-claim table should state whether a proposition is a procedural holding, factual finding, legal interpretation, actuarial assumption, or analytic inference.

A verified citation also requires a precise coverage of what was actually read. The United States cases cited here were inspected through published opinion transcriptions, and the United Kingdom cases through Supreme Court case records and the accessible summaries and judgment pathways. The CGL wording is reproduced inside a 2017 professional paper, not represented as a signed policy from a claim file. The CAS research entry is an accessible scholarly catalog summary; its inclusion supports the scope and existence of long-tail actuarial inquiry, without a claim that inaccessible empirical microdata were reproduced. This distinction is preserved in the companion evidence ledger.

Classification of temporal uncertainty

At least five distinct kinds of uncertainty enter an occurrence-based assessment. Document uncertainty concerns lost policies, incomplete endorsements, uncertain policyholder names, and renewal gaps. Factual uncertainty concerns the first time bodily or physical injury occurred and whether it continued after a suspected exposure ended. Legal uncertainty concerns which policy language controls, whether governing law adopts a particular trigger, and whether allocation or excess exhaustion follows all-sums or pro-rata analysis. Contract-limit uncertainty concerns erosion by other losses, aggregate exhaustion, defense expenses, deductibles, and contribution. Model uncertainty concerns the selection of a reserve estimate and probability assumptions used to translate uncertain litigation results into expected cash flows.

These categories interact but should not be merged. If a scientific report cannot establish whether contamination began affecting neighboring property in 2019 or 2021, the legal choice of a pro-rata denominator may remain unstable even after the applicable legal rule is settled. If the injury dates are well established but choice of law is contested, factual precision will not resolve the legal allocation result. If the triggered years and legal distribution are known but historical aggregates are exhausted, the insurance recovery may still be materially different from the assigned covered loss. A single undifferentiated coverage uncertainty parameter masks the point at which further investigation could be valuable.

Historical claims and evidence preservation

Long-tail claims often arise after the disappearance of contemporaneous records. A policy issued decades earlier may survive only through a certificate of insurance, broker correspondence, secondary schedules, or a partial form. Claims handlers may possess settlement agreements while underlying court filings, scientific reports, or policy endorsements remain dispersed. A scientifically reproducible dossier should therefore track the custody and dates of recovered records, preserve the difference between originals and reconstructions, and show whether each missing document could change coverage conclusions. An insurer's internal claim code is convenient for analysis but is not a substitute for the executed policy text.

An archival reconstruction can also overfit to information created years later. A 2025 expert report may show that damage existed by 2010, but the insured's 2010 knowledge must be assessed using the contractual definition of knowledge and the evidence legally admissible for that issue. Similarly, later judicial opinions may illuminate the interpretation of historical policies without rewriting the facts known to the policyholder when the contract began. For the reproduced CGL preexisting-injury provisions, knowledge qualifications require attention to the exact role of the insured person whose knowledge is relevant and the form version purchased.

Claims triangles and the occurrence-year problem

Actuarial valuation of long-tail liability often uses development data grouped by accident year, report year, underwriting year, or other cohorts. A reported claim in 2025 involving physical injury across 2008–2012 can be assigned to a single operational accident-year code for convenience, while its insurance coverage may implicate several historical contracts. This is not necessarily a bookkeeping error; the statistical cohort and the legal coverage unit serve different purposes. The danger appears when a cohort code is treated as dispositive proof of which contract must pay.

For a liability portfolio, let R_{y,d} denote cumulative reported loss for cohort y measured at development age d. An ordinary development method can estimate an ultimate amount conditional on the reporting cohort and loss-development experience. It does not identify whether a disputed underlying injury is covered under year y, year y−1, or multiple years. The coverage allocation requires a policy-year bridge informed by facts and law. For asbestos and environmental exposures, this bridge is particularly consequential because claim reports can arrive long after exposure, and litigation can reopen estimated recoveries. The CAS asbestos-and-environmental study is used here as evidence of this actuarial problem domain, not as validation of any specific factor selection for the hypothetical examples. Cross and Doucette (1994) proposed an exposure-oriented treatment of asbestos bodily injury liabilities that connects estimated aggregate insured loss to policy years and insurance layers. Their asbestos liability measurement study supports the need for a documented policy-year bridge, while its scope, modeled inputs, and publication context delimit inference about other lines or later markets.

The bridge from claim estimation to cash flows can be expressed without introducing a novel statistical framework. Let U be a modeled ultimate covered-liability amount before application of a contested legal allocation. For each legal scenario s, compute contractual outflows P_{j,s} by policy and year; then add separate defense and allocated expenses, subtract recoveries where legally supportable, and incorporate future settlement and payment timing. This helps prevent double counting. For instance, two triggered primary policies do not each generate the full claimant settlement expense merely because both appear in the coverage scenario. Likewise, pending reinsurance reimbursement is an asset or recovery hypothesis with its own counterparty and contractual risk, not an automatic reduction of gross direct liability.

A bounded scenario exercise

A transparent reserve exercise can use mutually exclusive legal scenarios. Suppose the total expected underlying liability for a hypothetical portfolio is $10 million under an otherwise identical claims-development estimate. Legal scenario A, based on the assumed wording and governing law, assigns $6 million to contracts held by the insurer. Scenario B assigns $8 million, while scenario C assigns $4 million. If a research exercise stipulates scenario probabilities of 0.2, 0.5, and 0.3, the arithmetic weighted mean of the scenario-specific insurer assignments is \$6m × 0.2 + \$8m × 0.5 + \$4m × 0.3 = \$6.4m. It is a conditional illustration, not an estimate based on observed case frequencies or a statement about recognition under any particular accounting standard.

The same weighted average can conceal legal tail risk. Under those stipulated scenarios, the scenario maximum is $8 million and minimum is $4 million, giving a $4 million range while the underlying portfolio's $10 million liability estimate stays constant. A robust report would show the full conditional distribution, scenario rationale, insurer-year limits, expected legal expenses, settlement timing, and the sensitivity of probabilities to an adverse judicial decision. Historical court cases provide useful scenario descriptions; they do not supply empirically calibrated probabilities. Assigning exact percentages without data should be labeled expert judgment rather than statistical inference.

Accounting recognition is a different question

The IFRS 17 standard overview and official IFRS 17 terminology address measurement of insurance-contract cash flows and liabilities for incurred claims under an international accounting regime. These materials establish a reporting and measurement context for insurer obligations. They do not determine whether an old CGL policy is triggered under Texas law, which allocation rule a New York court must apply, or the correct asbestos disease date. The coverage conclusion informs estimated future cash flows; the accounting framework governs recognition, measurement, presentation, and disclosure using its own requirements.

This separation has practical consequences for long-tail reserves. An insurer may prudently hold an estimate for uncertain adverse coverage outcomes before a final judgment, while the law of the contract remains contested. Discounting, risk adjustment, and financial reporting can materially affect the carrying amount relative to an undiscounted legal demand. Conversely, an actuarial reserve recorded in a reporting system is not a concession that the policyholder has conclusively established cover. An editorially accurate research document should identify the financial measurement basis whenever it quotes a reserve amount.

Scholarly validation and negative evidence

Validation of an individual occurrence-trigger analysis requires testing its most consequential assumptions. First, replace the proposed injury onset with the earliest and latest medically or technically plausible dates and recompute the list of implicated years. Second, apply each legally supportable trigger interpretation and record which policy periods enter or leave the set. Third, rerun allocation under governing and plausible competing rules, retaining limits and uninsured periods. Fourth, test whether the duty to defend conclusion differs from ultimate indemnity because it uses allegations rather than proven facts. Fifth, check whether a different policy form, endorsement, or governing law changes the conclusion more than the injury chronology itself.

Negative findings are also research results. A cited opinion that never decided whether a particular insurer ultimately indemnified cannot establish that indemnity was payable. An insurer's absence from a historical insurance schedule may reflect missing documents rather than a proven uninsured period. A court's refusal to apply an unavailability exception in one New York case does not settle whether another state's law has an equivalent rule. The explicit boundary in KeySpan's allocation holding is therefore as important as the substantive result. Likewise, the facts and law behind Zurich's UK decision make its contribution to comparative analysis specific and intelligible rather than universal.

This monograph's evidence base is purposive legal-document selection with selected actuarial literature. It makes no assertion of a systematic search of every English-language decision or all international jurisdictions; there is no individual claims microdataset, quantitative estimate of insurer-market prevalence, or population-weighted comparison of judicial results. The numerical scenarios test logical consequences of assumptions and arithmetic, not the empirical frequency of court outcomes. These limits are stated to protect the conclusions from inappropriate expansion.

Integrative scientific synthesis and research conclusions

What occurrence-based attachment actually decides

The decisive analytical result is that occurrence-based attachment is a contractually specified relationship between actual or alleged harm and an insurer's coverage interval. Its content follows from the grant, definition of occurrence, actual bodily injury or property damage, applicable legal interpretation, and evidence. Under common CGL wording, the date of the negligent act, the purchase of a product, the filing of a claim, and the settlement date do not automatically substitute for the date that covered harm occurred. These dates can remain material for causation, defenses, statutory limitation, reporting, or financial settlement, but they do different work. The Montrose judgment and Don's Building Supply judgment provide strong illustrations of this temporal separation for different injury settings.

A scientific coverage conclusion should consequently identify its predicate explicitly: the insured is alleged to be liable for a bodily injury first diagnosed in year X; records support actual injury during interval Y; policy P covers occurrences causing injury during its interval Z; governing rule T implicates P insofar as Y and Z overlap; all remaining terms and ultimate proof continue to apply. This structure is an ordinary form of transparent documentary reasoning. It is useful because it reveals which inference would change if a new medical opinion, updated policy endorsement, or later binding decision emerged.

Why a multi-year trigger is not a multi-year payment formula

A continuing harm may implicate three policies and produce one indemnity obligation to a third party. The number of implicated contracts is therefore not the number of independent compensable losses. The courts' differing treatment of insurer choice, pro-rata distribution, insurer contribution, and uninsured periods confirms that attachment and allocation are separate legal questions. The contrast between Owens-Illinois's New Jersey rule, Consolidated Edison's New York rule, and J.H. France's Pennsylvania approach is substantive: the same broad long-tail injury category can produce distinct recoveries because contract wording and jurisdiction affect the distribution of an otherwise fixed covered damage amount.

The arithmetic consequences are measurable. In the six-year hypothetical, the loss amount is $12 million, yet the assumed four insured years receive an $8 million assignment while two uninsured years retain $4 million. A shorter four-year supported injury interval changes those allocations materially. Neither computation proves a claim's real-world coverage, but both provide a controlled test of how injury chronology and legal allocation affect the mapping from legal loss to insurer cash flow. Any model that reports only total underlying damages suppresses these coverage-sensitive variables.

The international comparison as evidence of contract specificity

The comparison among U.S. CGL cases, the UK mesothelioma employer-insurance decisions, and the Canadian construction-defect decision demonstrates the centrality of local policy language and judicial reasoning. The UK BAI trigger litigation concerns particular historical employer-liability formulations in a special disease setting. The Zurich mesothelioma litigation examines contribution and indemnity after an employee's exposure crossed coverage periods. The Canadian Progressive Homes judgment is informative about alleged property damage, accident, and defense, but is not an international allocation precedent. Their combined value is to discipline the comparative method: use like-for-like issues, preserve institutional differences, and resist extrapolation from the mere word trigger.

The evidence also identifies a recurring temporal asymmetry: a harmed person can discover injury after the industrial practice or faulty work has ended, while the insurer whose policy responds may be determined by an earlier period of actual harm. From the insured's standpoint, this makes records of old coverage and operating history economically valuable. From the insurer's standpoint, it creates long-duration reserving, adverse-development, and coverage-litigation risk. From the court's standpoint, it requires a persuasive account of facts, wording, and law rather than an administrative preference for whichever insurer is currently identifiable.

Interpretive consequences for legal research and insurance science

A comprehensive research file should preserve at least four independent outcomes: potential defense, established indemnity attachment, financial allocation, and insurer payment after application of limits and retentions. The structure prevents a duty-to-defend judgment from being mistaken for a final coverage ruling; it also prevents a favorable allocation holding from being mistaken for an unconditional contractual payment. The Viking Pump opinion and Montrose excess-insurance decision show why excess layers and noncumulation language require a further stage once the underlying injury and allocation questions are defined.

For insurance science, this has implications for the construction of historical claim data. The observed reporting year or cash payment year need not match the insured injury year. A triangle organized by claims reported can estimate late development while failing to reflect how a court distributes indemnity among multiple contractual periods. Adjusting the actuarial estimate therefore requires documentation of the legal bridge, not a mechanical reassignment based on one claim-system field. The IFRS 17 framework provides a financial-reporting context for incurred insurance liabilities, but the legal bridge remains rooted in the underlying insurance contract and its governing law.

For scholarship, the most consequential challenge is uncertainty about old injuries. New medical expertise, long-term contamination data, or archived contracts can reopen a seemingly settled timeline. Yet uncertainty is not unstructured. One can specify the intervals, evidentiary thresholds, and legal interpretations under which competing conclusions would follow. The second hypothetical reserve exercise illustrates how explicit scenario assumptions can expose legal sensitivity without pretending to measure the frequency of court decisions from a handful of cases. The method supports accurate inference and editorial traceability rather than a new doctrine.

Further research questions

Three lines of empirical inquiry follow. The first is a carefully stratified dataset of decided occurrence-trigger disputes with standardized fields for clause version, injury category, governing jurisdiction, alleged injury interval, judicial disposition, and ultimate indemnity result where available. Such a dataset would allow descriptive comparisons of outcomes, while a valid causal analysis would need to address selection into litigation and publication. The second is a longitudinal study of how missing historical policy documents and evidence of latent injury affect settlement timing and reserve error. The third is an examination of interaction between primary allocation decisions, excess attachment, and reinsurance recovery using insurer records with permission and adequate privacy protections. These are research questions for a future program, not claims of studies performed here.

Conclusion

Occurrence-based liability insurance trigger interpretation establishes the temporal connection between the bodily injury or property damage described by a liability contract and the insurer's period of coverage. In simple accidents, the relevant dates may converge. In latent disease, progressive contamination, and concealed construction damage, causative conduct, physical injury, diagnosis, notice, litigation, and settlement can be separated by years or decades. The resulting coverage decision depends on the complete policy text, a documented injury chronology, binding jurisdiction-specific interpretation, and the kind of legal determination being made.

The comparative cases support four findings. First, exposure, injury-in-fact, manifestation, and continuous-injury approaches represent legally significant ways of interpreting particular coverage grants, with no single judicial label sufficient on its own. Second, an allegation that potentially implicates a policy may support a duty to defend while ultimate indemnity still requires proof and examination of remaining terms. Third, an injury spanning multiple policy periods does not settle allocation: pro-rata and all-sums results, uninsured-year shares, noncumulation provisions, and primary/excess exhaustion are distinct legal and contractual determinations. Fourth, the actuarial and accounting consequences of temporal coverage disputes can be analyzed reproducibly only if the injury-year, claim-year, legal allocation, actual payment, and reporting-measurement variables remain distinct.

The original contribution of this monograph is a source-grounded comparative reconstruction of these relationships, with explicit limits on jurisdictional inference and reproducible sensitivity calculations. Its governing scientific conclusion is that dependable long-tail liability analysis requires an evidence-traceable sequence from the executed policy and actual injury through legal attachment, financial allocation, payment constraints, and financial reporting. The documentary record and the calculation assumptions supporting that conclusion are disclosed in the References and accompanying EVIDENCE.json.

References