RESEARCH DOSSIER
Liability Insurance Primary Excess and Umbrella Layers
Abstract. Liability insurance programs distribute the economic consequences of legally compensable third-party injury across retentions, primary policies, successive excess policies, and sometimes umbrellas whose insuring agreements also extend the scope of protection. The division is simultaneously contractual, legal, operational, and statistical. This monograph analyzes the distinct functions of underlying insurance, vertical limits and aggregates, follow-form incorporation, true umbrella gap coverage, attachment by exhaustion, defense costs, multi-year allocation, settlement, insolvency, and recoveries. Its primary documentary evidence comprises court opinions from the United States and United Kingdom, insurance supervisory terminology, statutes, professional actuarial literature, and published accounting and prudential standards. Its quantitative analysis derives layer-payment identities and examines illustrative claims using explicit assumptions rather than observed insurer data. Judicial decisions establish that a loss may exceed an attachment amount without satisfying a policy-specific condition for payment; that the rules governing vertical or horizontal exhaustion follow the governing words and law; and that a primary insurer's insolvency does not itself necessarily convert its insured exposure into an uncovered exposure. The central analytical finding is that layer attachment, coverage scope, and cash recoverability are distinct coordinates of an insurance program. Reliable evaluation therefore reconstructs individual policy wording, legal period, covered loss, exhaustion condition, defense treatment, and economic distribution before calculating insurer obligations. This synthesis provides a reproducible account of layered liability insurance without asserting a new doctrine or an original empirical discovery.
Keywords: Primary liability insurance; excess liability insurance; commercial umbrella; umbrella drop down; follow-form excess; attachment point; exhaustion; underlying insurance; vertical exhaustion; horizontal exhaustion; aggregate limits; defense costs; self-insured retention; claims-made; long-tail liability; actuarial layer pricing; insurance coverage litigation; reinsurance.
Introduction
The essential economic purpose of a layered liability program is to protect an insured from qualifying third-party obligations of increasing severity while assigning portions of the resulting financial exposure to identified contracts. A liability claim can pass through a deductible, a self-insured retention, one or more primary policies, several excess contracts, and an umbrella policy. That description specifies an order of potential financing; it does not itself specify which insurer owes money. The answer depends on the legal character of the obligation, the relevant policy period, the applicable exclusion or grant, the loss definition, the underlying insurance schedule, the exhaustion language, the treatment of defense expense, and the governing law. The NAIC insurance glossary expressly recognizes excess and umbrella cover above a specified underlying amount and protection potentially involving gaps in basic coverage.
This research addresses insurance issued to an insured against liability to others. It distinguishes primary and excess insurance from excess-of-loss reinsurance, which is ordinarily a separate contract under which a reinsurer indemnifies a ceding insurer; an insured's right against an excess insurer does not automatically create a corresponding direct right against a reinsurer. It also separates an indemnifiable judgment or settlement, the insurer's contractual duty to defend, the estimated accounting reserve, a payment already made, and an ultimate reinsurance recovery. These entries may relate to the same accident while differing in timing, debtor, legal basis, and amount.
The analysis is international in comparative reach but relies on explicitly identified legal materials rather than treating a United States coverage decision as world law. England and Wales and several U.S. states provide especially informative publicly accessible decisions. UK disclosure and regulatory rules, international supervision, and international financial reporting provide complementary institutional contexts. Published actuarial work supplies mathematical tools for modeling attachment points and the cost of higher limits. The comparison is organized around three substantive questions: how underlying and follow-form limits are defined; how a layer attaches or drops down; and how actual gaps, aggregation and legal-year allocation affect the resulting economics. The method is doctrinal analysis plus transparent illustrative calculations, with no proprietary claims dataset or causal identification exercise.
Definitions, History, Legal Sources, and Subject Boundaries
Contractual architecture and terminology
A layer is the part of a covered loss financed between a lower attachment point and an upper limit, subject to the terms that make the layer available. A primary policy ordinarily begins at the contractually defined insured retention or deductible boundary and may provide initial defense protection. An excess policy adds a stated amount of capacity above listed underlying policies, a specified retained amount, or both. An umbrella policy can combine higher limits with independent protection for defined liabilities outside the scope of underlying coverage; whether a particular policy does so is a matter of its insuring agreement and exclusions. The American Bar Association's coverage analysis distinguishes these functional types and emphasizes that rights and duties depend on policy language.
The nominal phrase “$10 million excess of $5 million” describes an excess limit of $10 million with a nominal attachment at $5 million, providing a potential covered band from $5 million to $15 million. It does not promise the policy will pay every dollar in that numerical band. An exclusion may restrict the subject, a named underlying policy may have a different trigger, a defense cost may be outside or inside the stated limit, and the excess policy may specify that only certain payments count toward exhaustion. A layer's width is a numerical quantity; its coverage is a legal predicate. Both must be satisfied before a layer payment can be calculated.
A per-occurrence limit caps the amount for a single contractually defined occurrence; a per-claim limit may count claims independently, subject to related-claims provisions; an aggregate caps payments across all covered claims in the designated category or period. These limits can coexist. A self-insured retention is generally a portion of loss contractually retained before insurer attachment, while a deductible can be administered as an insurer payment subject to reimbursement by the insured. Their operational effects differ by policy form, including whether expenses, judgments, settlements, and direct defense costs erode them. Neither label alone answers which cash flows count.
Primary, excess, and umbrella insuring commitments
The NAIC defines commercial umbrella and excess insurance broadly, but the terminology is not a substitute for the operative contract. A strictly follow-form excess policy may mirror the underlying policy's substantive coverage while adding its own attachment, limits, notice and consent provisions. A stand-alone excess policy may state a complete coverage grant, independently chosen definitions and exclusions. A commercial umbrella may offer supplementary breadth for events excluded by the scheduled primary but included by its own wording, often subject to an umbrella retention. The policy might also provide only excess coverage over the underlying risk in specific situations.
The phrase drop down warrants two distinct definitions. First, when an underlying aggregate is exhausted through qualifying payments, an upper insurer may begin responding to later covered losses at an effective lower position, but only under the applicable wording. Second, a genuinely broader umbrella may operate as primary coverage over a retained amount for a class of risk excluded by underlying policies. The first concerns exhaustion; the second concerns coverage breadth. A third proposed use—requiring an excess insurer to pay because a primary insurer has become insolvent—raises a separate collectibility question and has no universal affirmative answer. This taxonomy follows the operative distinctions articulated in the 2026 Colorado Supreme Court decision, not a newly proposed legal doctrine.
The development of modern commercial liability towers reflects the interaction of large, potentially long-tail losses, liability caps, insurance market capacity, reinsurance and contracts structured for different claim bands. An upper layer may insure less frequent, more severe claims, yet be highly exposed to mass-tort accumulation or inflation in litigation cost. The CAS exposure-rating study shows why increased limits and policy exposure cannot be priced by attachment arithmetic alone: assumptions about claim frequency and severity matter. The historical observation is structural rather than a claim that one standard form governs all markets.
Sources of law and research design
Policy wording has priority in determining what a given insurer promised, interpreted under the applicable law. Courts have disagreed on policy-year allocation and conditions for exhaustion in circumstances involving continuous harm, multiple insurers and negotiated settlements. A court's conclusion about an excess policy is bounded by the particular forms, procedural posture, and governing law. The official New York decision in Matter of Viking Pump (2016) explicitly bases all-sums allocation and vertical exhaustion on the policies before it. The California Supreme Court applied a related contract-sensitive analysis in Montrose Chemical (2020), although the cases did not establish an internationally uniform rule.
The UK Supreme Court in Teal Assurance (2013) examined an engineering professional-liability program incorporating a primary insurer, successive excess layers, and a top-and-drop cover with geographical exclusions. Its reasoning is especially useful because it identifies the distinction between the order in which insured liabilities are ascertained and the later timing of insurer payment. This is a court interpretation of the identified policy terms, not a universal rule overriding all other policy wordings.
Methodologically, the dossier uses publicly inspectable primary judgments, official regulatory and legislative texts, and professional literature at the level actually accessible. Coverage and pricing identities are derived independently. No insurer policy file, confidential placement slip, claim database, complete census of national statutes, or licensed full-text journal collection is represented as having been reviewed. Comparative conclusions are correspondingly conditional on contract and jurisdiction. The scope excludes separate comprehensive treatments of tort liability, reinsurance tower placement, industry-wide reserving, litigation finance and insurer insolvency regulation, except insofar as each changes a primary, excess or umbrella layer.
Underlying Limits and Follow Form
Limits, retentions, and scheduled insurance
A complete layer analysis begins with the declarations and the schedule of underlying insurance: named insurers, policy numbers, applicable limit categories, deductibles or retentions, effective dates, claims-made or occurrence triggers, and any minimum underlying insurance condition. Schedules serve multiple roles. They identify the expected financial layer beneath the excess insurer, specify coverage assumed to remain available, and can determine how contractual references to actual underlying payment apply. A schedule that names multiple underlying policies raises the further question of whether their limits are separate, overlapping, alternative, or cumulative. Naming two primary policies does not by itself make both fully collectible for the same loss.
For a simple single occurrence, suppose the insured has a $250,000 retained amount, a $2 million primary limit above that retention, and a $5 million excess limit above the $2.25 million total retained-and-primary threshold. A qualifying $4 million loss would, under a stated simplifying assumption that no exclusions, aggregates, defense expenses or special exhaustion conditions apply, produce $250,000 retained, $2 million primary and $1.75 million excess. This allocation is a model calculation, not an interpretation of a particular issued policy. If the primary has paid only $1 million because its coverage is disputed, the excess obligation cannot be inferred merely from the illustrated severity band.
A policy may require the insured to maintain scheduled underlying limits in force. If the insured buys less primary insurance than specified, cancels a scheduled policy, or permits a gap between underlying and excess effective dates, the excess insurer's liability may continue to be calculated as if the scheduled underlying amount remained; alternatively, precise wording may provide for a different response. The economically relevant amount is thus the contract's attachment basis, not merely today's collectibility of a primary payment. A robust schedule review treats a minimum underlying condition as an independently testable clause.
Incorporation and independent exclusions
Follow form generally means incorporation by reference to an underlying policy's terms to an identified extent. The operative inquiry is whether the excess policy follows the underlying insurance in all terms other than stated exceptions, and which underlying contract controls when multiple policies are scheduled. “Follow form” is an instruction to analyze the relationship between documents; it is not evidence of perfect identity. The excess may have its own pollution, professional services, contractual liability, punitive damages, territorial, sanctions, cyber, or prior-known-circumstance exclusions. Changes to the primary after binding may also require consent to bind the excess, depending on contract and applicable law.
Consider a U.S. primary commercial general liability contract covering bodily injury but excluding a specified professional service. A stand-alone umbrella might affirmatively insure that service, subject to its own exclusions and retention; a purely follow-form excess policy ordinarily would not acquire coverage for that excluded service simply because the resulting damages exceed the primary limit. Conversely, a primary policy could cover the risk while an upper excess policy independently excludes it. A claim can then traverse a monetary layer that is unavailable as insurance. Without identifying these coverage grants and exceptions, describing a program as a continuous $25 million tower is only a statement about advertised limits, not a tested statement about financial protection.
Direct policy comparisons should therefore use a clause-by-clause matrix: insuring agreement; definitions of occurrence, claim, loss and ultimate net loss; insureds and additional insureds; exclusions; territories; period; aggregation; expense erosion; underlying maintenance; exhaustion by specified party; and notice, cooperation and consent. Differences are not administrative trivia. A single change in a policy's definition of loss or payment can alter both the insured's incentive to settle and the upper insurer's contractual obligation. The Teal Assurance judgment illustrates how the words describing payment, ascertainment and the identity of underlying claims function together within a tower.
Defense and claims management
Defense is often a distinct performance from indemnity. In many primary liability policies, an insurer undertakes to defend potentially covered claims, subject to the applicable policy terms and law. An upper excess insurer may disclaim an immediate defense duty until exhaustion; may possess a right to associate in defense; may pay covered defense expenses as part of ultimate net loss; or may acquire an express defense obligation in circumstances not covered by scheduled primary insurance. All such possibilities require contract review. The ABA multilayer coverage discussion connects the primary insurer's defense position with duties and coordination involving other layers.
Suppose a $3 million indemnity settlement is accompanied by $900,000 in defense and adjustment costs. If the policy's covered loss includes defense expenditure within its $3 million liability limit, the indemnity capacity remaining after defense differs materially from a form that pays defense outside limits. If upper layers follow the former's definition but change its treatment of expenses, upper attachment must be recalculated using the exact incorporated and independent terms. Court costs are not a generic universal addition to limits. Their allocation may also determine when exhaustion occurs and when excess insurers acquire practical control or consent rights.
Settlement behavior adds another dimension. A primary carrier may receive a demand within its policy limit whose rejection exposes an excess layer, the insured's retained interests and potentially itself to dispute. Governing law may impose standards of good faith or reasonable settlement in an insurer's handling of claims; those duties and enforcement rights vary. Equitable contribution and subrogation theories depend on the jurisdiction and whether a contribution claimant is a co-primary, excess or other carrier. A legal analysis should separately identify the party possessing the duty, the party entitled to enforce it, and the effect of any settlement consent clause. A conceptual tower cannot substitute for that legal analysis.
Temporal triggers and occurrence aggregation
The triggering event for insurance coverage is not always the date a lawsuit is filed or a payment is made. Occurrence-based and claims-made coverage organize the temporal risk differently. The NAIC medical-malpractice overview identifies the broad distinction: occurrence coverage concerns harm occurring within the policy period, while claims-made forms focus on claims made within the relevant coverage window, potentially including an extended reporting period. Actual claims-made forms can also contain retroactive dates, notice requirements, or related-claims rules that materially narrow the apparent time window.
A multilayer program becomes especially vulnerable to mismatch when a claims-made primary policy is placed under occurrence-based excess wording, or where retroactive dates, extended reporting arrangements or definitions of related wrongful acts are not harmonized. The question is whether each layer independently responds to the same claim under its own temporal trigger, not whether all cover pages show the same calendar year. Multi-year bodily injury or environmental damage requires another step: deciding which periods are triggered and whether contractual and jurisdictional rules produce proportional allocation or all-sums recovery against triggered policies. The comparison between Viking Pump and Montrose Chemical demonstrates that non-cumulation, prior insurance and attachment wording can matter to that result.
An occurrence may aggregate several claims into one limit when the policy definition and applicable causation rule treat them as related; another wording may assign distinct per-claim limits. The distinction affects both lower-layer erosion and upper-layer exposure. A $12 million event comprising six $2 million claims could exhaust a $10 million occurrence aggregate if aggregation applies, while none of the individual claims might exceed a hypothetical $3 million per-claim primary threshold. An asserted amount of aggregate damages cannot decide the coverage result without the relevant definition. These examples show why temporal and aggregation coding belongs in both legal review and actuarial claim data.
Drop Down and Attachment
Exhaustion mechanisms
An excess attachment point is both a quantitative threshold and a contractual condition. In the idealized layer function, covered loss above an attachment amount reaches the next carrier. In actual contracts, a clause might require exhaustion by payment by the underlying insurer, by payment by the insured or another party, by an admitted obligation, or by a judicially established liability. Other policy clauses may treat aggregate exhaustion, defense costs, reinstatements, or reductions of underlying limits differently. The exact wording determines whether settlement with a lower insurer for less than nominal limits satisfies the condition. It is necessary to separate the size of a legally covered obligation from the event that makes the upper insurer's payment duty mature.
The English Teal Assurance opinion examined policy language that deferred excess liability until underlying insurers had paid, admitted, or been held liable for their full indemnity; other clauses addressed the reduction and exhaustion of underlying aggregate limits. Lord Mance explained why a mechanism governing when upper-layer liability attaches does not permit the insured or its captive insurer to rewrite the identity and order of claims falling into earlier layers. The case is important because an insured liability can become ascertained at one time and a payment under a policy may occur later. The claimant's ascertained obligation, the exhaustion of a lower indemnity, and the contractual maturity of the next insurer's obligation occupy different points on the chronology.
Where a policy reduces its attachment point after the underlying aggregate is exhausted, the exact cause of exhaustion matters. Consider an underlying $5 million annual aggregate exhausted by six covered claims paid during a policy year; an upper layer may contractually respond to a later claim within the same policy year if its aggregate drop-down clause so provides. If the underlying policy was instead canceled, rescinded or never purchased at the scheduled limits, the excess contract may continue to require the originally contemplated lower amount. If a court finds that defense costs reduce aggregate insurance, a materially different exhaustion trajectory may result from identical indemnity payments. “Drop down” therefore cannot be determined from a ledger of indemnity checks alone.
Insolvency and noncollectibility
Insolvency is the most direct test of the distinction between a coverage gap and an inability to collect. A scheduled primary policy can insure a loss while the primary carrier lacks resources to meet its obligation. Whether an umbrella or excess carrier must assume that obligation depends on the excess policy's treatment of scheduled underlying insurance and insolvency. In A.R. Wilfley & Sons v. National Union (2026), the Colorado Supreme Court answered a certified question concerning umbrella/excess wording under which certain coverage could attach for occurrences not covered by scheduled underlying policies. It held, on that wording, that the scheduled primary insurer's insolvency did not render an otherwise covered occurrence uncovered and did not oblige the excess insurer to provide the primary insurer's first-dollar indemnity or defense.
The decision has a narrower and more valuable implication than a blanket assertion that an umbrella never drops down. It directs attention to the distinction between risk outside the scope of underlying coverage and payment uncollectible under a policy that covers the risk. A policy may address the second circumstance expressly or differently. The applicable insolvency law and guaranty arrangements may create additional recovery questions, but they are analytically separate from the excess insurer's insurance contract. A model of expected reimbursement should distinguish probabilities of coverage, attachment, and collectibility, with contractual and credit-risk assumptions separately recorded.
In the same judgment, the Colorado Supreme Court described the broader protective function of umbrella cover and the additional limit function of excess cover. That functional distinction is useful but must not be mistaken for a representation that every product marketed as “umbrella” contains identical insuring agreements. The name on the policy is weaker evidence of drop-down scope than the complete wording. A research file should store the declarations and endorsements, not just the retail product category.
Settlement below limits
Suppose a primary insurer with a $5 million limit resolves a claim with its insured for $3.8 million, while the insured pays another $1.2 million to settle the underlying third-party liability. The actual loss may satisfy the monetary $5 million threshold, but the excess insurer's obligation also depends on who must pay the lower limit under the attachment wording. One interpretation treats the lower layer as exhausted when the insured has absorbed the shortfall; another requires full payment by the specified insurer. The choice affects allocation of the $1.2 million compromise and upper-layer obligations even when total third-party recovery is identical.
The historical discussion of Zeig v. Massachusetts Bonding describes the Second Circuit's 1928 acceptance of exhaustion by negotiated settlement under language that did not mandate actual full collection of underlying insurance, while preserving the excess threshold. This ARIAS-US account is a secondary case analysis, not a substitute for treating the 1928 holding as textually universal. The official New York decision in JP Morgan Chase v. Indian Harbor (2012) discussed more restrictive exhaustion terms and rejected an argument that a below-limits settlement combined with insured-funded shortfall necessarily satisfied the language before it. Both lines of reasoning illustrate the primacy of the specific insurance wording.
The practical result is that negotiations with underlying insurers can change the position of the upper layer even if the ultimate third-party settlement is unaffected. An insured that accepts a discounted lower-layer settlement before obtaining upper-layer consent may bear a gap that the excess insurer is not contractually required to pay. Conversely, an excess insurer should not be assumed to receive a windfall from all compromised settlements: some policies expressly accept payment by the insured or others as qualifying exhaustion. The relevant clause, jurisdiction, and actual payment record must be examined before declaring either outcome.
Judicial analysis of layered towers
The US cases Montrose Chemical, Viking Pump, and Truck Insurance Exchange v. Kaiser Cement (2024) are best understood as distinct inquiries into complex towers affected by long-tail injury. Montrose permitted access to an otherwise available excess policy after the directly underlying excess insurance in that year was exhausted. Viking Pump interpreted specific non-cumulation and underlying policies to permit all-sums allocation and vertical exhaustion. Truck addressed first-level excess coverage under the wording before the California court and held vertical exhaustion applicable, while leaving additional contribution questions for proceedings on remand. None removes the need to read the terms of the excess contract or the precise status of underlying coverage.
A legal reconstruction must ask: who brought the action (insured against insurer, insurer against insurer, or another party); what relief was sought (declaration of attachment, reimbursement, contribution, or defense); which policy years were triggered; and which exclusions and endorsements were before the court? A holding about an insured's ability to select a triggered tower does not automatically determine how insurers subsequently contribute among themselves. Equally, a decision about the time at which a liability is ascertained does not necessarily decide whether a settlement below underlying limits qualifies as contractual exhaustion.
This distinction has operational consequences. Claims departments may treat lower-layer exhaustion as an administrative event; litigation can demonstrate that it is the result of a legal interpretation of several contract documents. A legal-opinion reserve may be needed for coverage risk in addition to an actuarial reserve for the expected amount of covered liability. The New York liability-insurance statute also illustrates the importance of class-specific rules: part of its policy-limits disclosure provision expressly distinguishes certain policies from excess liability or umbrella contracts. That statutory distinction is a matter of New York law, separate from an insurer’s contractual payment obligation. Neither is a substitute for maintaining the actual payment and consent chronology.
Coverage Gaps and Layer Allocation
Vertical and horizontal exhaustion
A vertical exhaustion construction allows access to a particular upper policy after exhaustion of its directly underlying insurance, commonly within the same policy period. A horizontal exhaustion construction can require exhaustion of policies at a lower level across additional periods before a higher policy attaches. These are interpretive alternatives used especially in continuous-injury disputes; the governing text and law determine which applies. Horizontal and vertical describe relations among financial layers or years. They should not be conflated with an umbrella's separate breadth of coverage across different insured perils.
In Matter of Viking Pump, the New York Court of Appeals considered excess policies that either followed form to non-cumulation provisions or contained non-cumulation and prior insurance clauses. Its conclusion favored all-sums allocation and vertical exhaustion under the relevant language. Montrose Chemical likewise used textual and background-law analysis to permit the insured to proceed vertically through directly underlying coverage. These holdings show that the physical picture of an insurance tower becomes misleading when the actual disputed program has separate towers for separate years with cross-year allocation rules.
The 2024 Kaiser Cement opinion addressed first-level excess coverage in a contribution dispute, explaining that the first-level policy wording was materially similar to the higher-level wording analyzed in Montrose. The court found vertical exhaustion appropriate for the relevant excess policies and remanded questions that remained regarding contribution. The distinction matters for secondary analysis: an insurer may have a duty to indemnify the insured once attachment occurs, while another insurer's attempt to recover an equitable share involves a separate cause of action and potentially separate defenses. Coverage access and inter-insurer allocation are not interchangeable accounting entries.
Allocation across years and events
A single injury may generate alleged responsibility across many calendar years. For certain kinds of environmental, asbestos or other progressive harm, a jurisdiction's trigger rule may implicate several occurrence policies. The analyst must first specify exposure or injury chronology, then identify each relevant period's terms, then examine whether the loss is allocated proportionally to years, whether an insured can seek covered all-sums indemnity against a triggered policy, and whether stacking or non-cumulation clauses alter the available amount. It is inappropriate to apply one allocation rule across jurisdictions and policy years without documenting its source. The official New York Viking Pump judgment supplies a concrete example of a non-cumulation clause affecting the allocation choice.
An illustrative continuous-injury loss of $24 million might implicate three contract years. Under an intentionally simplified pro rata time allocation, and assuming three equal covered periods, $8 million would be tentatively allocated to each year before applying retentions and each year's limits. Under an all-sums model with legally permitted selection of one triggered year, a $24 million covered amount could be presented against one year's tower, with subsequent contribution issues to be resolved separately. These are different scenario assumptions; neither calculation asserts that a particular jurisdiction necessarily mandates it. A non-cumulation clause could make naive addition of all three towers wrong even if their declared nominal annual limits appear to sum to a much larger number.
A useful claim record has at least four dates: alleged first injury, alleged last injury, first claim or notice, and ascertainment or settlement of the insured's liability. Payment, attachment, and reinsurance settlement dates are additional independent fields. A data file with only the “loss date” cannot reliably distinguish claims-made from occurrence liability or model the sequence in which annual aggregates erode. The temporal classification is therefore a precondition of quantitative layer pricing rather than merely a post-loss legal detail.
Excess-of-loss mathematics and aggregates
Let X be a nonnegative, gross, otherwise covered monetary claim amount at a specified valuation date, after applying a stipulated coverage interpretation. Let a be the attachment amount measured on the same basis and let w be the width of the excess layer. For a single isolated claim, no deductible interaction, no annual aggregate, full qualifying exhaustion, no independent exclusion, and no separate defense-cost treatment, its simplified excess payment is
This is a mathematical payment identity, not a contract clause. It follows that 0 ≤ Y ≤ w and that Y is nondecreasing in X. The function is zero for X ≤ a, grows dollar-for-dollar when a < X < a + w, and is capped at w thereafter. Under the same assumptions, a primary coverage band from retained amount d to limit d + p has payment Y_{d,p}(X). An umbrella gap-coverage grant for a different peril would require a different covered-loss variable and potentially a different retained amount; it cannot be obtained by mechanically shifting a in the original formula.
The expected pure loss cost of a single-claim layer can be written in survival-function form, assuming E[X] is finite,
If N is a claim count independent of identically distributed severities X_i and there is no cross-claim aggregate, the expected annual layer loss is E[N] × E[Y]. If frequency and severity vary with a latent risk factor or are dependent, the correct expectation is E[∑_{i=1}^{N}Y(X_i)] with the actual joint distribution. A portfolio annual aggregate A further changes the insurer payment to min(A, ∑ Y(X_i)), subject to claims timing and legal coverage conditions. In general E[min(A,∑Y)] is not equal to min(A,E[∑Y]); exchanging the nonlinear cap and expectation understates or otherwise misstates the model depending on the comparison. The CAS Basic Ratemaking text and the CAS layer-rating literature provide the actuarial context for policy limits, exposure curves and loss distributions.
For reproducibility, consider four independent illustrative claims, X = $0.4 million, $1.8 million, $4.0 million, and $12.0 million. For a $5 million excess of $2 million band, the modeled payments are $0, $0, $2.0 million, and $5.0 million; the total is $7.0 million. If the layer also has a $6 million annual aggregate, the modeled total becomes $6.0 million. These computations are conditional on all four claims being eligible and on attachment occurring as assumed. A real tower could produce a different outcome if defense costs erode the lower aggregate, if the $4 million claim is outside the excess coverage grant, or if contractual exhaustion requirements remain unmet.
An increased limits factor (ILF) for basic limit b and higher limit l can be framed, at the pure-loss level, as E[min(X,l)] / E[min(X,b)] when comparable covered-loss definitions and exposure assumptions apply. The CAS monograph on actuarial distributions discusses derivation of ILFs from claim-size distributions. The expected price of an excess layer is related to differences in limited expected values, but policy price adds expenses, reinsurance, capital, risk margin, profit or contingency provisions, and contractual scope. It should not be described as a universal simple multiplication of a primary policy premium.
International comparison and reinsurance
International comparisons begin with differences in liability law, policy form, claims procedure and market supervision. The UK Supreme Court's analysis of claim ascertainment and payment in Teal Assurance is grounded in the program's specified professional-liability wording and English-law reasoning. New York and California opinions address distinct contracts and sometimes distinct procedural questions. Consequently, two programs with identically expressed dollar attachment points may respond differently because they operate under different covered-loss triggers, doctrines of construction, aggregate rules, or reinsurance structures.
The UK Insurance Act 2015 requires fair presentation for non-consumer insurance contracts within its statutory scope. The principle affects how a commercial insured discloses risk across placements; it does not itself rewrite the stated attachment condition of each excess policy. The FCA's insurance conduct rules establish market-conduct obligations involving clear, appropriate information, with their application dependent on regulated activity and policy/customer classification. Neither framework supplies a generic judicial answer to a disputed US vertical-exhaustion clause.
The EU's revised product-liability directive changes the prospective landscape of product liability. Article 2 states that the directive applies to products placed on the market or put into service after December 9, 2026, a future date as of this dossier's November 15, 2026 cutoff. This prospective rule should not be retroactively applied to historical loss years, and its effect on liability insurance depends on transposition, national law and individual contract language. The directive is cited for evolving liability exposure, not as a harmonized European code of primary-versus-excess attachment.
Reinsurance is a second, analytically distinct layering system. A primary or excess insurer may cede a share of its liability to reinsurers, and its own reinsurance may be structured in proportional or excess-of-loss layers. The insured-facing insurance contract and the insurer-facing reinsurance contract can use different attachment definitions, aggregates, occurrence constructions, exclusions and payment provisions. A $10 million excess insurer's expected outward reinsurance recovery is therefore not implied by its $10 million gross insured liability. Solvency and accounting studies must reconstruct the insurer's gross claim obligation, ceded recovery, counterparty risk and timing separately. The IAIS Insurance Core Principles identify insurance risk and capital supervision as matters requiring effective supervisory systems; they are not a standardized substitute for contract-by-contract reinsurance analysis.
Evidence, Uncertainty, and Model Validation
Sources, interpretation, and inference
A statement that a specified insurer “covers” a loss can refer to at least five different propositions: the underlying tort or statutory liability has arisen; the insured is within an insurance grant; the amount falls into the excess layer's numerical band; policy conditions for attachment are satisfied; and the insurer has sufficient funds or legally available recovery mechanisms to pay. These propositions are related but independently defeasible. The research method therefore treats the insured's legal liability, contractual coverage, insurer performance obligation, and actual payment as separate observable or adjudicable variables.
The documentary evidence has unequal epistemic status. A court opinion directly establishes what the identified court decided, on what record and under which jurisdictional law. It does not by itself prove the actuarial distribution of liability claim severity. An insurer glossary describes market terminology without creating operative coverage in a particular issued policy. A professional actuarial paper supplies derivations or assumptions for rating models but does not establish the scope of a particular exclusion. A regulatory standard states supervisory or reporting requirements within its scope and effective period, but it is not a private contract. Each source is linked in the claim's prose and recorded with its specific claim and source category in the accompanying evidence ledger. The method deliberately avoids presenting these different sources as interchangeable.
Legal evidence can be tested using an extraction protocol. For each disputed policy, record the quoted operative clause only to the limited extent necessary for analysis, identify the court and decision date, the parties' roles, the coverage period, the issue actually decided, and any explicitly reserved questions. Distinguish an opinion's holding from its discussion of cases from other jurisdictions. In Truck Insurance Exchange v. Kaiser Cement, for example, the remand and unresolved equitable contribution issues matter because the vertical-exhaustion holding alone does not conclusively allocate contribution among carriers. Treating the opinion as the final determination of all cost-sharing issues would exaggerate its scope.
Market evidence also demands proportional interpretation. The Swiss Re social-inflation study identifies rising US liability litigation costs and claim severity pressures. Its estimated index is the publisher's empirical research result under its chosen definitions; the present dossier does not independently reproduce that index. Its relevance is mechanistic: a change in the upper tail of claim severity can increase attachment probabilities and expected excess-layer payments even if the mean number of claims changes little. The result cannot be extrapolated without qualification to every jurisdiction, legal system or commercial line.
Pricing, reserving, and stress testing
An excess underwriter needs data on both the frequency of covered events and the probability that their covered severity will exceed each attachment point. Claims should be adjusted to a consistent valuation basis, with claim development, reporting delay, truncation at prior policy limits, large-loss corrections, inflation assumptions and changing retentions made explicit. Observations from primary insurer paid claims often provide an incomplete view of claims above the primary limit: a high loss may be censored at that limit. To fit the tail of severity responsibly, the analyst needs additional excess loss experience, large-claim reports or defensible parametric assumptions. The CAS loss-layer paper explains why a severity and frequency distinction is central when moving between limits.
The fundamental pricing sensitivity is the survival probability at each dollar of excess attachment. Differentiating the expected layer loss with respect to its lower attachment a, for a continuous severity distribution under the simplified assumptions, gives d E[Y_{a,w}(X)] / da = P(X>a+w) − P(X>a). The derivative is nonpositive. The higher the attachment, the lower the expected cost of an otherwise identical fixed-width layer, holding the covered-loss distribution and all contract terms constant. In actual pricing, higher bands can still demand large capital charges, sparse-data margins or risk loadings because their outcomes are severe and dependent; the pure-loss derivative does not predict the full commercial premium.
Sensitivity of upper layers to severity inflation can be pronounced. If every covered loss increases by a positive factor k, an attachment a set in nominal dollars becomes a/k on the original loss scale, making previously sub-threshold losses potentially attach. If the original severity variable is X, the transformed payment is Y_{a,w}(kX), rather than simply k × Y_{a,w}(X): the lower threshold and upper limit prevent a general proportionality. A stress test that multiplies the expected upper-layer loss by a general inflation percentage can therefore miss the changing number of losses entering the band. The Swiss Re social-inflation analysis provides a documented motivation for testing changing severity, litigation cost and large verdict exposure, while the mathematical conclusion follows directly from the layer function.
Claims reserving needs additional separation of liability and timing. A claim may be fully reserved by the primary insurer but remain unpaid and disputed. The excess insurer may face an incurred-but-not-reported exposure under an occurrence form even when no claim has yet been made, or a reported claim requiring a distinct claims-made trigger. A prudential reserve must reflect available information and applicable measurement rules; the accounting recognition of an estimate is neither legal exhaustion of underlying insurance nor an insured settlement. The IFRS 17 insurance contracts standard provides the international financial-reporting context for risk-adjusted measurement of future contractual cash flows; it does not itself decide whether a disputed claim satisfies the words of an excess policy.
A reinsurance credit or expected cession should be monitored separately from gross liability. If an insurer owes a qualifying excess payment but the corresponding reinsurance recovery is contested or delayed, its liquidity demand may precede collection from reinsurers. Capital management must assess gross and net exposures, counterparty default, aggregation risk and duration mismatch. The IAIS supervisory framework and IFRS 17 address different but complementary oversight and financial-reporting purposes. Claims, reinsurance, actuarial and legal departments must reconcile their records so that “gross ultimate loss,” “covered contractual claim,” “paid loss,” “recoverable reinsurance” and “recognized liability” are not used as synonyms.
Counterexamples and robustness
The strength of a layered-insurance conclusion can be tested by altering one assumption at a time. Start with a $2 million primary layer above a $250,000 retention and a $5 million excess layer. Under the simplified covered-loss calculation, a $4 million loss triggers $1.75 million from the excess. Now suppose the primary policy excludes the relevant liability and the excess simply follows form. The covered-loss input into both contracts may become zero despite the unchanged gross damages. Suppose instead an umbrella independently covers the excluded activity subject to a $500,000 umbrella retention; the umbrella's response will follow its separate coverage grant, loss definition and limit, and the previous $2.25 million attachment model will not describe it. The numerical severity has not changed; the contract classification has.
Next keep all coverage predicates constant but change the lower insurer's financial health. The primary insurer's insolvency changes its ability to perform, while the upper insurer's legal obligation may remain unchanged. This is precisely the kind of distinction clarified for specific wording in Wilfley (2026). The financial model must consider an unrecovered primary amount even if the excess band is mathematically unaffected. Alternatively, vary the excess policy's exhaustion clause to accept payment of the underlying shortfall by the insured; the result may differ from a contract requiring full payment by named underlying insurers. That sensitivity is supported by the divergent case analyses associated with JP Morgan Chase and Zeig.
Finally vary the time aggregation. Assume a $9 million long-tail claim potentially attaches to two successive years, each with different primary and excess structures. A model that sums both towers without checking the non-cumulation language may double-count contractual capacity. A model that applies a one-year-only allocation without establishing the governing rule may understate coverage. Viking Pump and Montrose Chemical show why model validation must test alternative legally supportable allocation interpretations. In this dossier, neither construction is assumed to apply to a hypothetical claim absent its underlying wording.
A reproducible coverage-and-pricing study should preserve the following linked datasets: policy text and endorsements with version identifiers; governing jurisdiction; period and retroactive dates; underlying schedules; loss event facts and relevant claims dates; liability and defense estimates; exhaustion evidence; insurer payments and settlements; and claims assumptions for frequency, severity and inflation. A legal-review change log should distinguish an actual amendment from a later reinterpretation of unchanged text. The actuarial run should retain its model code, parameter estimates, exposure definitions, stochastic seed if simulation is used, and sensitivity outputs. None of those proprietary insured or claims records was available for this study. The present quantitative illustrations have therefore been checked only against their explicit arithmetic and stated assumptions.
Integrative Scientific Synthesis and Research Conclusions
Contractual and financial reconstruction
The findings support a sequential reconstruction of liability insurance programs. The first analytical operation is identification of the underlying legally compensable harm, covered claimant and insured entity. The second is identification of the relevant policy year and contractual coverage grant, including exclusions, notice and defense. The third is identification of the applicable primary limit, retention, underlying insurance schedule and upper-layer wording. The fourth is assessment of the legal preconditions for attachment and their factual satisfaction. The fifth is calculation of covered loss in each eligible layer, with aggregates and claims sequence. The sixth is financial reconciliation of actual insurer payments, retained insured costs and possible reinsurance reimbursement. These are ordinary analytical operations derived from the evidence; their presentation here is an organized synthesis and does not establish a proprietary scientific system.
The order matters because an early accounting shortcut can create an invalid legal conclusion. For example, applying a $5 million layer excess of $2 million to a $6 million judgment first produces an attractive modeled payment of $4 million. But if the asserted loss is outside the excess policy's coverage grant, the correct payable amount under that policy is zero. If coverage exists but the primary has not satisfied an actual-payment exhaustion condition, the modeled amount may represent economic exposure rather than an immediately enforceable excess payment. If the claim spans two periods subject to aggregation or non-cumulation language, the model must reconstruct eligible policies before using layer functions. Mathematical precision cannot repair an incorrect coverage predicate.
A second result concerns the value of carefully defined coverage gaps. Three materially different gaps can appear: an affirmative underwriting gap where an event is outside the primary coverage grant but inside a broader umbrella grant; a limit gap where lower and upper contract thresholds fail to meet as intended; and a collectibility gap where a covered insurer obligation cannot be paid. Legal and financial remedies for the three gaps differ. The distinction is supported by the vocabulary of the NAIC insurance glossary and the contract interpretation in Wilfley. It is best treated as an interpretive classification, not as a newly claimed general theory of insurance.
A third result concerns the unit of analysis. The appropriate object is usually a versioned contract operating on a defined claim in a legally identified period, embedded within a program whose other contracts may become relevant. Broad program labels such as “$100 million tower” omit the information needed to test defense, exclusions, period alignment, attachment and collectibility. Financial capacity and insured risk coverage are different measurements. A program may have a large nominal limit while providing no relevant excess protection for a particular claim. Conversely, a true umbrella may cover a claim that the scheduled primary excludes, although its own retained amount and conditions must first be satisfied.
Governance, disclosure, and risk transfer
The policy-level findings have practical implications for underwriting and risk governance. At placement, insureds and insurers should establish schedules of underlying insurance that reconcile limits, dates, insured entities and trigger types, and should record how primary amendments affect excess wording. A named follow-form relationship should specify the exact primary form and exceptions. A claim-handling plan should identify who controls defense and settlement before and after exhaustion, and whose consent is necessary for a settlement that implicates upper limits. An insolvency-sensitive program should document whether its umbrella or excess wording responds to noncollectibility, rather than inferring an insurer guarantee from the word “umbrella.”
In a UK commercial context, the insured's material risk disclosures also intersect with the Insurance Act 2015 fair-presentation duty. For insurers and intermediaries subject to relevant regulatory standards, FCA insurance product information requirements emphasize comprehensible disclosure of policies and their complexity. These institutional sources support the importance of making coverage boundaries understandable, while the law applicable to any particular policy controls its legal interpretation.
For international firms, changes in substantive liability exposure can migrate losses into upper layers even without amendments to the insurance contract. Product liability developments, litigation trends, changing settlements, and economic or social inflation affect the severity distribution at which a fixed attachment point is reached. The EU product-liability directive and Swiss Re's comparative analysis identify different kinds of drivers: future legal changes and empirically observed claims-cost pressure. The proper response is scenario analysis with geographically and temporally defined assumptions, not wholesale transplantation of one market's data into another.
Research findings
The documentary evidence establishes several durable results. Underlying insurance limits are simultaneously measures of intended risk financing and inputs into contractual attachment conditions; neither one exhausts the other. A follow-form undertaking is defined by its actual incorporation clauses and independent terms. A true umbrella's additional breadth is a coverage question distinguishable from exhaustion following an aggregate limit. Insurer insolvency creates a collectibility problem that a given umbrella may or may not contractually absorb. Contractual settlement language can alter whether a compromised underlying settlement unlocks excess coverage. Multi-year allocation and layer exhaustion must be reconstructed from wording and applicable jurisdictional rules. Actuarial layer functions describe conditional covered amounts and require separate legal and operational inputs before they become forecasts of collectible cash flow.
These findings are derived from authoritative coverage cases, market terminology and elementary actuarial identities. The original contribution of the dossier lies in making their interactions explicit, reconciling the apparently conflicting terms commonly used for attachment and drop down, and establishing a transparent comparison between legal conditions and quantitative payment calculations. No original legal doctrine, pricing theory, causal estimate or new scientific school is claimed.
The strongest unresolved research requirement is access to a large, anonymized body of full policy forms, claims and settlement histories covering several jurisdictions and multiple insurer layers. Such a dataset could permit empirical estimation of how frequently upper-layer collections are delayed or reduced by attachment disputes, and whether gaps arise predominantly from misaligned forms, financial default, temporal triggers or explicit underwriting choices. It would also allow comparison of contractual excess attachment with realized distributions of loss paid and recovered. This work establishes the definitions and documentary context for such research; it does not anticipate its empirical conclusions.
Conclusion
Layered liability insurance is a contractual method of distributing liability risk over retentions, primary insurance, excess capacity and potentially broader umbrella protection. Its performance depends on the agreement between covered claim characteristics and the conditions of each participating policy. Attachment cannot be inferred from damages alone; an upper insurer's duty depends on the coverage grant, underlying structure and exhaustion wording. The leading examined cases show the significance of this distinction in disputes concerning settlement below limits, multi-year injury, the order of policy exhaustion, and primary-insurer insolvency.
A complete evaluation of a liability tower therefore proceeds from legally established loss and covered insurance events to contractual attachment, numerical layer allocation, annual aggregates, defense and settlement duties, insurer credit exposure, and reinsurance reimbursement. The mathematical layer formula identifies conditional loss cost; legal examination identifies whether the assumed layer is available for the particular event. The combination supports defensible pricing, reserving, governance and claims analysis. Its essential scientific conclusion is that a liability insurance program can be understood rigorously only when contractual eligibility, numerical limits, timing and cash realization are reconstructed together for each relevant claim and policy period.
References
American Bar Association. (n.d.). Principles of Multi-Layer Insurance Coverage: Duties, Allocation, and Risk. ABA Litigation Section. https://www.americanbar.org/groups/litigation/resources/newsletters/insurance-coverage/principles-of-multi-layer-insurance/
A.R. Wilfley & Sons, Inc. v. National Union Fire Insurance Co. of Pittsburgh, PA, 2026 CO 64 (Colorado Supreme Court, September 21, 2026). Official opinion. https://www.coloradojudicial.gov/system/files/opinions-2026-09/25SA265.pdf
Bahnemann, D. (2015). Distributions for Actuaries. Casualty Actuarial Society Monograph Series. https://www.casact.org/sites/default/files/2021-02/02-Bahnemann.pdf
Evans, J. P. (2001). Exposure Rating Loss Layers: Unifying the Property Perspective of Severity with the Liability Perspective of Frequency. CAS research abstract, Fall 2001, pp. 493–500. https://www.casact.org/abstract/exposure-rating-loss-layers-unifying-property-perspective-severity-liability-perspective
Financial Conduct Authority. (updated 2025). Insurance Conduct of Business Sourcebook (ICOBS). https://www.fca.org.uk/firms/insurance-conduct-business-sourcebook-icobs
International Association of Insurance Supervisors. (updated December 2024). Insurance Core Principles and ComFrame. https://www.iais.org/activities-topics/standard-setting/icps-and-comframe/
International Accounting Standards Board. (2017, as amended). IFRS 17 Insurance Contracts. IFRS Foundation. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/
JP Morgan Chase & Co. v. Indian Harbor Insurance Co., 2012 NY Slip Op 04702 (New York Court of Appeals, June 2012). Official New York State Law Reporting Bureau text. https://www.nycourts.gov/reporter/3dseries/2012/2012_04702.htm
Matter of Viking Pump, Inc., 27 N.Y.3d 244, 2016 NY Slip Op 03413 (New York Court of Appeals, May 3, 2016). Official New York State Law Reporting Bureau text. https://www.nycourts.gov/reporter/3dseries/2016/2016_03413.htm
Montrose Chemical Corp. of California v. Superior Court, 9 Cal. 5th 215 (California Supreme Court, April 6, 2020). Court opinion, hosted by Justia. https://law.justia.com/cases/california/supreme-court/2020/s244737.html
National Association of Insurance Commissioners. (n.d.). Glossary of Insurance Terms. https://content.naic.org/glossary-insurance-terms
National Association of Insurance Commissioners. (2026). Medical Malpractice Insurance. NAIC Insurance Topics. https://content.naic.org/insurance-topics/medical-malpractice-insurance
New York State Legislature. (current legislative text). New York Insurance Law § 3420: Liability Insurance; Standard Provisions; Right of Action by Injured Person. https://www.nysenate.gov/legislation/laws/ISC/3420
ARIAS•U.S. Law Committee. (n.d.). Zeig v. Massachusetts Bonding & Insurance Co. Case analysis of Second Circuit decision, 23 F.2d 665 (2d Cir. 1928). https://www.arias-us.org/law_committee_report/zeig-v-massachusetts-bonding-insurance-co/
Swiss Re Institute. (2024, September 7). Sigma 4/2024: Social Inflation—Litigation Costs Drive Claims Inflation. https://www.swissre.com/institute/research/sigma-research/sigma-2024-04-social-inflation.html
Teal Assurance Co. Ltd. v. W R Berkley Insurance (Europe) Ltd., [2013] UKSC 57 (Supreme Court of the United Kingdom, July 31, 2013). BAILII full judgment. https://www.bailii.org/uk/cases/UKSC/2013/57.html
Truck Insurance Exchange v. Kaiser Cement & Gypsum Corp., S273179 (California Supreme Court, June 17, 2024). Court opinion, hosted by Justia. https://law.justia.com/cases/california/supreme-court/2024/s273179.html
United Kingdom. (2015). Insurance Act 2015, Part 2: Duty of Fair Presentation. Legislation.gov.uk. https://www.legislation.gov.uk/ukpga/2015/4/part/2
European Parliament and Council of the European Union. (2024). Directive (EU) 2024/2853 of 23 October 2024 on Liability for Defective Products, Article 2 (future application). EUR-Lex. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=celex%3A32024L2853
Werner, G., & Modlin, C. (2016). Basic Ratemaking (5th ed.). Casualty Actuarial Society. https://www.casact.org/sites/default/files/2021-03/5_Werner_Modlin.pdf