RESEARCH DOSSIER
Per-Risk Excess of Loss Reinsurance
Abstract. Per-risk excess of loss reinsurance is a nonproportional indemnity arrangement under which a reinsurer assumes an agreed portion of the ceding insurer's qualifying loss above a specified retention, subject to a limit measured separately for each contractually defined risk and loss. The expression appears straightforward, but its operational meaning depends on the identity of a risk, the composition of ultimate net loss, the ordering of underlying reinsurance, the treatment of multiple claims from a common cause, and further limits on an occurrence or annual aggregate. These dimensions jointly determine the recoverable cash flow and the cedent's residual financial exposure. This monograph examines treaty architecture using publicly filed American contracts from 2008, 2013, and 2024; primary supervisory materials; actuarial studies of experience and exposure rating; and comparative legal analysis. It formulates transparent payment functions, tests hypothetical single-risk and multi-risk losses, distinguishes exhaustion from reinstatement, and explains why the same nominal layer can transfer materially different amounts of risk under different wordings. A detailed analysis of the 2024 publicly filed property treaty shows how a $2.25 million layer excess of $750,000 per risk is further constrained by an occurrence cap and an aggregate cap. Separate comparisons show that the order in which ceded proportional reinsurance is recognized can materially alter excess recoveries, and that claims development and reinsurer default can separate contractual recoverable value from cash available to pay policyholders. The original contribution is a source-grounded analytical synthesis of contract construction, nonlinear payment mathematics, underwriting exposure, claim allocation, and regulatory recognition. It establishes no new proprietary theory or doctrine. The main conclusion is that a per-risk treaty must be evaluated as a complete system of definitions and interacting financial constraints rather than as a pair of retention and limit figures.
Keywords: Per-risk excess of loss reinsurance; property per risk; nonproportional reinsurance; risk definition; ultimate net loss; cedent retention; attachment point; indemnity layer; inuring reinsurance; gross and net loss; per-occurrence cap; annual aggregate limit; reinstatement; exposure rating; experience rating; excess loss; claims allocation; salvage and subrogation; reinsurance credit risk; Solvency II; reinsurance contract law.
Introduction
The economic function of reinsurance is to allocate insurance loss between an original insurer and another risk-bearing entity. The original insurer remains responsible under its direct contracts, while the reinsurance contract specifies the circumstances in which the reinsurer indemnifies the insurer. Within nonproportional reinsurance, the attachment mechanism is sensitive to the amount of loss. For per-risk cover, it is additionally sensitive to the unit over which that loss is measured. An amount paid under three damaged policies can be three separate risk losses, one combined risk loss, or one catastrophe occurrence involving multiple risks, depending on the language and facts. Those characterizations can produce different recoveries even when the physical damage is identical.
The U.S. Federal Insurance Office's report on global reinsurance distinguishes excess-of-loss contracts protecting individual risks from contracts protecting a single occurrence or accumulated losses over a period (Federal Insurance Office, 2014). The U.K. tax authority's worked excess-of-loss example separately demonstrates the threshold-and-limit operation for individual policies (HM Revenue & Customs, 2016/2023). Gary Patrik's actuarial Reinsurance chapter clarifies the analytical difference between subject loss for one covered policy and subject loss aggregated over all policies affected by one event (Patrik, 2001). These descriptions provide a point of departure; the parties' executed treaty remains central to determining an actual recovery.
The investigation proceeds from documentary definition to actuarial valuation and finally to enforceability. Three real treaty exhibits are used as contract evidence: a 2008 Florida homeowners treaty, a 2013 property treaty, and a 2024 property treaty (Homeowners Choice, 2008; United Property & Casualty, 2013; Lemonade and Metromile, 2024). They are evidence of particular transactions and negotiated language, not universal contract forms. The 2024 filing redacts commercially sensitive provisions, including premium information. Its disclosed attachment and caps can be examined, but its undisclosed premium cannot be inferred. Earlier filings are historical legal artifacts; they cannot establish the terms currently offered by their parties.
The study uses four methods. First, it identifies the contract's operative words and their order of application, treating the policy, reinsurance agreement, endorsements, claims documentation, and applicable law as distinct evidentiary layers. Second, it expresses those words through loss functions and deliberately hypothetical examples. Third, it examines the credibility of experience-based and exposure-based pricing with reference to published actuarial work. Fourth, it compares supervisory recognition and contract enforcement without treating model standards as directly applicable statutory law everywhere. No proprietary insurer claims triangles, confidential placement slips, litigation files, or actual treaty prices are supplied; the absence of such data constrains empirical inference but leaves the contractual and mathematical analysis complete.
The central questions are therefore concrete. What is one risk? Which underlying payments compose the net loss that may attach? In what sequence should distinct coverages, expenses, recoveries, and caps be applied? How much apparent capacity survives a large event or series of losses? What contractual and institutional conditions make a mathematical receivable an economically reliable indemnity?
Terminology, Source Genealogy, and Scientific Scope
The nonproportional architecture
A reinsurance treaty is a standing agreement covering a defined book of business or class of contracts during a specified temporal and geographic scope. Per-risk excess cover normally sits above an amount that the cedent must bear on a qualifying subject risk loss. If the agreed retention is $1 million and the maximum reinsurer payment per risk is $4 million, a conventional description is "$4 million excess of $1 million," or "$4 million xs $1 million." A loss of $600,000 does not enter the layer, one of $3 million produces $2 million before other treaty constraints, and one of $8 million exhausts the $4 million layer, leaving $3 million above it to be borne or otherwise transferred.
This construction differs mathematically from a quota share, which pays a designated percentage of a defined loss from the first dollar. It differs from per-occurrence catastrophe cover, in which losses on distinct risks may be pooled to exceed a common event retention. It also differs from aggregate stop-loss, where attachment depends on cumulative qualifying claims or a contractual loss ratio. The distinctions follow the U.S. government reinsurance taxonomy and the Reinsurance Association of America glossary (Federal Insurance Office, 2014; RAA, n.d.).
The phrase "working excess" often describes relatively low layers expected to be hit more frequently. The phrase "catastrophe excess" generally directs attention to accumulation across covered risks. Yet the label does not settle the contract's true boundary. A property per-risk treaty may include an event cap so that several risk-level indemnities arising from one catastrophe cannot accumulate without restriction. The U.K. GIM8070 illustration expressly notes that per-risk treaties may impose a separate event limit to avoid inadvertently supplying broad catastrophe capacity (HM Revenue & Customs, 2016/2023). Accordingly, categorization and contract analysis must occur together.
Subject risk, loss, and financial layers
Four objects must be identified before pricing or interpreting a claim. The insured exposure is the physical or legal interest that gives rise to the original insurer's responsibility. The subject policy or book specifies which such exposures are included in the treaty. The contractual "risk" determines the unit to which the per-risk retention and limit apply. The "loss" identifies the amounts and periods to be aggregated for that unit. These objects may coincide in simple single-building insurance but diverge where one location contains several buildings, multiple policies cover related interests, or one event damages many premises.
"Ground-up loss" refers here to loss measured before the contemplated excess layer and before specifically identified insurance or reinsurance deductions; "subject loss" is the narrower amount produced by the treaty's definitions and prerequisites. "Ultimate net loss" (UNL) is a defined contractual quantity, not an automatically invariant actuarial statistic. The 2008 public treaty's Article VI definition includes loss-adjustment expense and specified extracontractual amounts while directing deductions for salvage, recoveries, and certain prior reinsurance (Homeowners Choice, 2008). The 2024 treaty's ultimate-net-loss clause likewise treats loss-adjustment expense, qualifying extracontractual liability, salvage, and inuring recoveries as express calculation elements (Lemonade and Metromile, 2024). A model that ignores these words models a different contract.
In contracts that cover only a cedent's net retained lines, the reinsured amount cannot be enlarged solely because another insurer or reinsurer has failed to pay. The 2008 treaty's net-retained-lines article expressly addresses the retained share and inability to collect from other reinsurers (Homeowners Choice, 2008). The financial question then divides into two parts: the size of the contractual UNL and the probability, timing, and recoverability of payments arising from that UNL.
Documentary genealogy and epistemic hierarchy
Excess-of-loss techniques developed within an insurance market seeking to separate frequent manageable losses from infrequent severity claims. That is a functional description rather than a claim that any particular printed contract first invented per-risk protection. Patrik's actuarial account of treaty structures documents the coexistence of quota share, surplus, facultative, per-risk working, and catastrophe covers in layered programs (Patrik, 2001). Swiss Re's educational treatment, Proportional and Non-Proportional Reinsurance, presents examples of per-risk and event-based limits operating in a common program (Swiss Re, n.d.). Such expositions describe practice, but specific policy wording and adjudicated disputes must govern actual obligations.
The strongest evidence for what two parties agreed is their operative treaty and endorsements, interpreted with governing law. A public SEC exhibit is especially useful because it permits examination of entire articles rather than a glossary alone, although executed subscription shares and omitted confidential information can limit a third party's reconstruction. The 2008 contract exhibit identifies separate subscribing reinsurer participations and a 2008 premium; the 2024 contract exhibit includes redactions and describes several named reinsured companies (Homeowners Choice, 2008; Lemonade and Metromile, 2024). Neither is a standard market specimen without qualification.
An institutional standard answers another question: which controls and risk transfers a supervisor expects to recognize. IAIS Insurance Core Principle 13 addresses governance, risk appetite, counterparty credit, concentration and effectiveness of risk transfer (IAIS, 2024). The IAIS's April 2026 peer-review report assesses implementation of the relevant principle across participating supervisors; it does not certify any specific private treaty (IAIS, 2026). The legal status of a supervisory principle also differs from enacted national rules and judicial interpretation.
For pricing, empirical experience and models have a different evidentiary function. David Clark's Basics of Reinsurance Pricing explains how a bespoke reinsurance transaction demands judgments beyond generic loss-cost estimates (Clark, 1996). Snoussi and Walhin's per-risk pricing study investigates the different evidence used by historical experience rating and exposure rating, with attention to exposure curves and sparsely populated layers (Snoussi & Walhin, 2012). Those methodological findings inform an estimate; they cannot substitute for the definition of the insured layer.
Definition of a Single Risk and Attachment
The operative unit of indemnity
The central contractual problem is the meaning of "one risk." The per-risk retention does not apply to an abstract dollar amount detached from an insured unit. It applies to the agreed unit, subject to the loss aggregation rule. Under a treaty in which a separate building and its contents constitute one risk, simultaneous destruction of two buildings may create two risk-loss calculations. Under a different treaty, a whole location, an insured complex, or an individual policy may be one risk. If two buildings under one policy are combined into one risk, loss amount and retention application change even if the policy's total paid indemnity remains constant.
Actual filed contracts demonstrate the scope of drafting choices. The 2008 treaty's definition of one risk gives the cedent judgment over the unit while expressly constraining treatment of a building and its contents as separate risks (Homeowners Choice, 2008). The 2013 treaty's retention-and-limit article also states that the company judges what constitutes one risk (United Property & Casualty, 2013). The 2024 treaty's one-risk definition includes a specific treatment of renters' and condominium-unit policy exposures at a single location (Lemonade and Metromile, 2024). Such wording is a contractual allocation of initial classification authority within the specified contract; it should not be mistaken for a power to disregard other restrictions, applicable law, or a demonstrably inconsistent claim presentation.
Consider a hypothetical contract offering $2 million xs $1 million on each risk, each loss, with no aggregate restriction. One storm damages two separate insured buildings, generating qualifying subject losses of $2.2 million and $1.8 million. If each building is one risk, recoveries are $1.2 million and $0.8 million, totaling $2 million. If the treaty instead defines the two buildings as one combined risk and treats the storm damage as one loss, the qualifying loss is $4 million and the recovery is limited to $2 million. The equality of those results is accidental to the selected values. If each building instead suffered $2.8 million, separate-risk treatment would yield $1.8 million twice, or $3.6 million, while combined treatment would yield only the $2 million per-risk cap. Conversely, splitting a loss into multiple risks can multiply retentions where every individual loss is near or below attachment. The directional effect depends on the size distribution of losses.
A precise layer function
Let X denote the qualifying subject loss for one contractually established risk and covered loss after applying the relevant UNL definition. Let A be the nonnegative retention and L the positive maximum reinsurer payment for the relevant per-risk layer. Ignoring any additional restrictions, the indemnity is
This formula states the contract's ordinary vertical layer operation, not a universal legal rule. Equivalently, the reinsurer pays zero when X is at or below A; pays X − A between A and A + L; and pays L at or above A + L. The insurer's residual subject loss before other cover is X − c(X;A,L). The retention A is the threshold; L is the layer width and indemnity cap; A + L is the exhaustion point. Calling a "$2 million xs $1 million" cover "$3 million of reinsurance" would confuse the exhaustion level with the reinsurer's maximum payment.
An elementary identity makes the retained tail explicit:
For X ≥ 0, the identity decomposes the cedent's subject loss into the low-severity part below attachment and the uncovered upper tail beyond exhaustion. It is an algebraic re-expression of the agreed layer. It explains why this form of reinsurance limits a range of losses rather than placing an absolute cap on the cedent's maximum loss; loss above the top of the layer remains in the cedent's account unless other valid coverage attaches.
Causation, temporal attachment, and risk-level accounting
A risk-level limit must be combined with a temporal grant. A treaty can cover losses occurring within the treaty term, policies attaching during a treaty period, or designated claims-made exposure. The temporal basis affects reporting tail, runoff, premium allocation and which reinsurer is responsible as the portfolio renews. The 2024 treaty's term article provides a defined losses-occurring period and a conditional run-off mechanism linked to policies in force at expiry (Lemonade and Metromile, 2024). This is an illustration of operative text, not evidence that all per-risk treaties use the same basis.
Per-risk classification is also distinct from identifying the cause or event that groups claims for other purposes. The 2008 treaty includes a loss-occurrence article addressing events and time windows (Homeowners Choice, 2008). Such a clause may matter because a treaty can calculate individual per-risk liabilities but impose a separate event-level restriction. The accounting records must preserve policy, location, risk, claim, event, coverage period, and any allocated related expenses as separate fields. Without those fields, different forms of aggregation can become irreversible once the claims are reduced to a portfolio total.
The uncertainty in the unit
Per-risk coverage yields a deceptively strong visual promise when a schedule contains high sums insured. Sums insured describe potential exposure but do not by themselves determine reinsurance recoveries. In the HMRC numerical example, a £20,000 xs £10,000 treaty pays according to the size of the qualifying individual loss, up to its limit, rather than as a proportion of an insured value (HM Revenue & Customs, 2016/2023). Exposure values still matter for estimating loss distributions and layer price. This twofold role of the schedule—defining who is covered and informing expected severity—must be kept analytically separate.
A well-specified treaty should therefore identify what one risk means for multiple insured structures, locations, interests, policyholders and insurers; whether multiple losses on a risk within an event are combined; whether repeated events can lead to independent layer recoveries; which policy grants and exclusions are followed; and how risk classification interacts with other cover and any aggregate cap. The evidence required to resolve those questions includes location schedules, policy limits, endorsements, peril and cause codes, engineering information, claim chronology, loss-adjustment expense ledgers, and reinsurance bordereaux. They are claims and underwriting facts; adding up policy numbers alone does not establish the contractual recovery.
Ground-Up and Subject-Loss Allocation
From the original claim to treaty subject loss
One of the largest conceptual and practical errors in reinsurance analysis is to apply a per-risk excess formula directly to a direct insurer's gross claim ledger without first reconstructing the contractual subject loss. The claims ledger can include insured damage, deductibles and sublimits, defense expenses, bad-faith-related payments, salvage, subrogation, primary insurance, and previously negotiated facultative or proportional reinsurance. The per-risk treaty may admit some and exclude others. A dollar is not automatically part of the layer simply because it appears in a cedent's total financial loss.
A useful analytical sequence for a conventional indemnity property contract is: establish the insured peril and original policy obligation; resolve the insured's policy deductible, limits, and claims-adjustment items; identify the property, location, risk, and loss occurrence; allocate covered claim expense as the treaty directs; apply eligible salvage and subrogation; recognize any inuring insurance or reinsurance; establish the remaining subject ultimate net loss; apply the per-risk attachment and limit; finally apply treaty-wide occurrence and aggregate restrictions and payment mechanics. This sequence is an analytical checklist. Individual clauses can change its order and produce materially different numbers.
The significance of contractual order becomes concrete in the SEC-filed 2008 Article VI, where ultimate net loss includes defined expenses and qualifying excess-of-policy-limits and extracontractual obligations, but deductions for salvage and inuring recoveries are required. The same contract's Article VII on other reinsurance allows specified underlying protection to inure to the cedent's benefit and be disregarded in applying that treaty (Homeowners Choice, 2008). The resulting interpretation must read the two articles together rather than announcing that every prior cession is always deducted. The 2024 definitions article expressly orders certain salvage and inuring reinsurance recoveries to be subtracted in deriving ultimate net loss (Lemonade and Metromile, 2024). The answer is specific to the treaty.
Formal allocation with eligible components
For one risk and one qualifying loss, let G be the original insured gross amount payable by the cedent under the direct policy; E be permitted loss-adjustment expense; Z be permitted additional contractual or extracontractual liabilities; S be credited salvage and subrogation; and I be eligible inuring insurance or reinsurance recoveries. A transparent schematic UNL is
The terms are included only to the extent the particular treaty authorizes them. If the treaty defines UNL before actual receipt of an inuring recovery, the model should treat I as the recoverable amount dictated by contract even where collection is disputed. If expense is apportioned pro rata at a different stage, E cannot be included wholesale. If the subject loss is based on a net retained percentage, that percentage may have to be applied before the loss is subjected to the excess function. Therefore the formula describes a family of contractual calculations, not a default rule capable of overriding wordings.
The order matters because the layer function is nonlinear. In general
and a proportional share applied before a retention is not equivalent to a proportional share of a recovery calculated on the full amount. Take a hypothetical $5 million covered loss and a proportional arrangement that leaves the original insurer bearing 60%. Suppose the per-risk treaty covers the retained loss after that cession at $3 million xs $1 million. Subject loss is $3 million and the excess indemnity is $2 million. If the $5 million ground-up loss is incorrectly fed into that layer, the model produces $3 million. The $1 million error results solely from confusing the basis of the contract. The Clark reinsurance pricing note discusses the need to reconstruct experience net of inuring surplus-share reinsurance when pricing per-risk layers (Clark, 1996).
There are other contractual possibilities. A treaty may attach on gross rather than net amounts and direct which insurer receives the benefit of underlying cover; a reinsurer may cover only a specified fraction of the layer; or the contract may specify a preferential treatment of particular recoveries. The analyst should not silently impose the previous hypothetical ordering on another treaty. Each input in the claim calculation should be linked to its originating contract provision, transaction, and accounting date.
Expenses, claims handling, and the timing of UNL
Loss-adjustment expense may attach inside a per-risk limit, outside the layer, proportionately, or under a separate contractual condition. The 2024 public loss-adjustment-expense definition includes specified adjustment, appraisal, defense, litigation and recovery-related costs, together with qualified employee costs, while excluding ordinary overhead outside its specified exception (Lemonade and Metromile, 2024). This level of specificity is commercially material: counting a $300,000 legal defense bill as treaty subject expense can push an otherwise sub-retention claim into the reinsured layer. A subsequent court-cost recovery may reduce the ultimate liability, requiring a later adjustment between cedent and reinsurer.
For a numerical example, assume a hypothetical treaty treats $1.4 million of insured indemnity, $200,000 of permitted allocated expense, $100,000 of subrogation receipts and $150,000 of inuring reinsurance recovery as subject items, and provides $1 million xs $500,000. Then the subject loss equals $1.35 million, yielding an excess recovery of $850,000. If the $200,000 expense were excluded by actual treaty wording, subject loss would equal $1.15 million and the excess recovery would be $650,000. A change in expense eligibility, by itself, produces a $200,000 difference. If the subrogation recovery were still pending, whether it is estimated or treated only when collected must be resolved from the treaty; the example cannot decide that for a real contract.
A further distinction is between UNL becoming sufficiently established to support a reinsurance claim and the entire final cost of the underlying claim being determined forever. The 2008 and 2024 filed contracts contain provisions anticipating proof, settlement, later receipts, and adjustments. The 2008 settlement provisions and the 2024 reimbursement wording illustrate that contractual payment timing must be read from the agreement, rather than assumed to require final exhaustion of all claim developments (Homeowners Choice, 2008; Lemonade and Metromile, 2024). The accounting concept of an estimated recoverable, the legal entitlement under a treaty, and an actual bank receipt occur at different points in time.
Risk-level versus event-level accumulation
A catastrophe can damage many individually reinsured risks. Let X_i denote the defined subject loss of risk i in event e, with A and L common for simplicity. Without higher constraints, the preliminary ceded loss is
This sum deliberately occurs after each risk loss has been subjected to its own attachment. Treating the aggregate ground-up event loss as one risk and applying the attachment only once is a different mathematical object:
The two amounts generally differ. Five properties with subject losses of $1.5 million each under $1 million xs $1 million yield total per-risk reinsurance of $2.5 million (five recoveries of $500,000). If the same $7.5 million total is erroneously handled as one risk loss, the layer pays only $1 million. Conversely, many small risks each under attachment can give zero recovery even though their combined catastrophe damage is enormous.
Some treaties solve the exposure problem from the reinsurer's perspective by placing an additional occurrence cap above the sum of all eligible per-risk recoveries. In the 2024 SEC property treaty, the per-risk layer is $2.25 million excess of $750,000, while the reinsurer's liability is also restricted to $6.75 million per loss occurrence and $20.25 million over the contract term (Lemonade and Metromile, 2024). These three numbers refer to distinct dimensions. The event cap is not the per-risk limit, and the term cap is not the contractual attachment threshold. They must be represented as separate calculations.
Inuring priority and multi-layer programs
A reinsurance program commonly consists of primary insurance, facultative or surplus protection, one or more working per-risk layers, catastrophe layers, and perhaps aggregate protection. Some are expressed in terms of ground-up loss; others on the cedent's net retained loss after identified earlier cover. A treaty may expressly state that one prior cover inures to the benefit of the excess reinsurer, so that the reinsurer assumes loss only after the previous recovery has reduced the cedent's exposure. It may instead grant the benefit of the earlier cover to the cedent, leaving the per-risk excess computation unaffected by that earlier amount, subject to other provisions.
Consider an expressly hypothetical two-layer excess program: first layer $1 million xs $500,000; second layer $2 million xs $1.5 million; both apply to the same subject loss and have independent legal capacity. At X = $3 million, the first layer pays $1 million and the second pays $1.5 million, so the combined cover is $2.5 million and the cedent retains $500,000. At X = $5 million, the second layer reaches its $2 million ceiling; total recovery is $3 million and $2 million remains outside the two layers. This simple tower uses aligned exhaustion and attachment points. If a lower layer excludes a peril that an upper layer covers, the upper layer's contractual attachment does not necessarily disappear; the contract must say whether the gap is retained by the cedent, covered by a different insurer, or treated otherwise.
A valid waterfall thus requires a chart for each cover specifying what is measured, when it attaches, which earlier payments are deducted, whether the cover is exhausted and reinstated, and which events are aggregated. The Aon 2025 reinsurance training outline identifies practical skills in calculating recoveries across excess per-risk layers and understanding the direction in which reinsurance inures (Aon, 2025). Training material documents the relevance of those operations; the treaty wording determines their exact legal meaning.
Data integrity and the recoverability audit
The minimum auditable claims file should maintain stable identifiers for original insurer, original policy, risk, location, claim, insured event, loss date, paid amount, case reserve, allocated expense, salvage, subrogation, facultative recovery, proportional recovery, and the applicable treaty and subscription share. An event register should link risk-level claims without eliminating their separate identities. An audit should reproduce both the original gross-to-net calculation and every subsequent recovery allocation. If losses are restated for inflation or development, the original and adjusted amounts should remain independently retrievable.
The key methodological implication is that loss data are not automatically interchangeable across the primary-insurance and reinsurance accounting systems. A historical loss triangle prepared on a gross accident-year basis may be unsuitable for pricing a losses-occurring per-risk treaty net of proportional cover without remeasurement. Similarly, a catastrophe-event database that records one total event amount does not by itself reveal the distribution of losses across reinsured risks. The credibility and granularity of evidence determine how much can be learned from past claims, especially about rarely reached higher layers.
Limits and Recovery Waterfall
Four distinct upper constraints
Layer descriptions frequently emphasize the per-risk retention and indemnity width because these are easiest to draw. The contractual capacity can nevertheless be restricted along four different axes: the loss on an individual risk, the combined losses of one event, the total of all eligible losses in the contract period, and the percentage subscription of a participating reinsurer. The last dimension matters when multiple reinsurers underwrite shares of one treaty; it changes the allocation of the reinsurance payment among reinsurers and may also create counterparty concentration.
For the 2024 publicly filed treaty, the first three limits are expressly stated as $2.25 million per risk, $6.75 million per occurrence, and $20.25 million for all losses during the contract term (Lemonade and Metromile, 2024). To examine their interactions, suppose for illustration that eligible claims are ordered chronologically, all qualifying losses are undisputed, subscription percentages are collectively 100%, and no separate exception modifies the disclosed limits. For each event e, define
If G_e is the remaining contract-period aggregate available immediately before event e, its paid reinsurance recovery is
with initial G_1 = $20,250,000. These are a transparent representation of one possible claim-processing sequence for the disclosed caps. Actual contracts may specify how simultaneous events, adjustments, cash calls, changes in order, and reinstatements affect the accounting. Therefore the formulas do not replace the full legal text.
Worked example: individual loss versus occurrence cap
Assume five insured risks are damaged in the same qualifying occurrence and each has a subject ultimate net loss of $3 million. On each risk, $3 million minus the $750,000 retention equals the full $2.25 million per-risk layer. Preliminary reinsurance is five times $2.25 million, or $11.25 million. The event cap of $6.75 million reduces the amount payable for that occurrence to $6.75 million. The combined subject loss is $15 million. Before any other coverage, the cedent's residual amount is therefore $8.25 million, comprising the five per-risk retentions totaling $3.75 million plus $4.5 million of eligible layer indemnity prevented by the event cap.
This example demonstrates an analytical asymmetry: the maximum nominal per-risk coverage of an individual high-value property does not imply that the entire portfolio can recover that amount for every damaged property in one event. If the contract instead insures three qualifying risks with the same $3 million loss each, the preliminary $6.75 million sum exactly matches the occurrence cap; subject loss totals $9 million and the residual is the combined $2.25 million retentions. If ten qualifying risks each have a subject loss of $500,000, every loss is below the $750,000 retention and the aggregate reinsurance recovery is zero, however severe the total portfolio shock appears relative to an individual household.
There is no single universal method for allocating a binding event cap among individual risk records. A contract, cedent accounting practice, or loss settlement agreement could specify a priority, pro rata share, or another consistent approach. For a hypothetical pro rata method, five equal individual preliminary recoveries of $2.25 million become five allocated paid recoveries of $1.35 million, totaling $6.75 million. This allocation preserves totals and is computationally transparent; it should not be asserted as an automatically binding allocation rule where the treaty is silent. It also cannot manufacture additional coverage for a later reinstatement calculation.
Worked example: contract-period aggregate exhaustion
Now take three independent qualifying occurrences in the contract term, each generating preliminary per-risk amounts above the $6.75 million event cap. Under the assumed payment sequence, each occurrence consumes $6.75 million. After the first event the annual available amount is $13.5 million; after the second it is $6.75 million; after the third it is zero. A fourth eligible occurrence would produce no additional payment under this exhausted term cap, even if each individual risk exceeded its attachment and would have otherwise qualified.
At a lower level, consider the 2013 SEC contract. Its disclosed property layer covered $500,000 excess of $500,000 per risk and imposed a $2 million total limit during the term, including any specified run-off (United Property & Casualty, 2013). Four successive independent $1 million subject losses would each generate $500,000, jointly consuming the stated aggregate ceiling. The fifth equivalent loss would remain outside available cover. This is a direct consequence of the aggregate constraint; it is not evidence that those particular losses actually occurred under the historical contract.
In the 2008 treaty, a $1.5 million excess of $500,000 per-risk layer is paired with a $3 million total term limitation (Homeowners Choice, 2008). Thus the same high-level architecture occurs across public agreements with materially different quantitative parameters. A comparison across three filings establishes variation in negotiated terms; it does not establish the distribution, average, or standard price of industry contracts.
Distinguishing insurer retention from the sum of retained loss
The stated $750,000 retention in the 2024 example is a per-risk attachment threshold, not necessarily a ceiling on the cedent's total net cost. When one loss penetrates beyond the top of its layer, the cedent again bears the severity excess. When several losses emerge in one event, an occurrence cap can remove coverage that was otherwise available within individual layers. When several events occur in a year, exhaustion of the annual amount may eliminate later reinsurance. The total portfolio retention therefore depends on the full joint distribution of individual severity, event clustering and contractual aggregate usage.
For an actuarial estimate, the expected preliminary per-risk payout for a nonnegative random subject loss X can be represented as
This is the area under the survival function between the attachment and exhaustion levels. For an individual risk it shows how the expected cost of a layer reflects the tail of loss severity rather than the full insured value. For a portfolio of independently specified risk losses, linearity of expectation gives the sum of the individual expected layer costs before further caps, even if losses are dependent. However, after application of an occurrence cap or aggregate cap, expected paid reinsurance is generally below the sum of those preliminary expectations. Its value then depends on the joint distribution and event structure, not just the marginal severity curve of each risk.
Jensen-type intuition is helpful but should be applied correctly: for nonnegative Q, . The event cap truncates the payment distribution, and estimating alone does not tell us how often the cap will bind. Two portfolios with identical marginal per-risk loss distributions can have identical uncapped expected payouts but different capped expectations if their events create different degrees of contemporaneous loss clustering. This is a direct implication of applying a nonlinear event cap to a sum of risk-level amounts.
Claims settlement and contractual limits
Settlement provisions often permit the cedent to handle original claims, then bind the reinsurer within the agreed treaty scope. The 2008 liability and settlement articles expressly preserve the reinsurance contract's own terms and limits even while following primary policy obligations (Homeowners Choice, 2008). A judicial illustration is the New York Court of Appeals decision in Travelers v. Certain Underwriters (2001), which addresses the function of follow-the-fortunes arrangements while recognizing that such provisions do not override the agreement's stated limitation of liability. The source is jurisdiction-specific. It supplies a caution about contractual construction, not a global standard for allocation.
The Third Circuit's Travelers v. INA decision explains that ordinary reinsurance indemnifies an insurer rather than creating direct claims by its underlying policyholders, subject to exceptions actually established by law or contract (2010). The question of who may demand payment therefore differs from the question of how much loss has entered a reinsurance layer. The cedent retains its original insured obligations and must manage liquidity even where the reinsurer's ultimate payment obligation is reasonably clear.
A reliable recovery analysis should show the sequence of entries: policy-indemnity determination; allowable expense; applicable salvage, recoveries and inuring insurance; per-risk amount; shares of participating reinsurers; any event cap; annual or term cap; previous paid and outstanding allocations; and net balance recoverable. When a claim is reopened or adjusted, the same sequence must be re-run and changes reconciled against prior cash payments. Proper recovery accounting is a continuous contract-based process rather than a one-off calculation at the first reported claim.
Reinstatement and Aggregate Exposure
Meaning and contractual operation of reinstatement
A reinsurance layer may have an initial limit for each qualifying risk or event and a further maximum amount available across the treaty period. Reinstatement describes the contractual restoration of coverage capacity consumed by a covered loss, in the amount, at the time, and for the additional consideration stated in the agreement. It cannot be presumed from the phrase "excess of loss." An agreement can provide free reinstatement, paid reinstatement, only a specified number of reinstatements, or no reinstatement at all. The RAA reinsurance glossary defines reinstatement in relation to restoration of the amount of reinsurance protection following a loss (RAA, n.d.). Its exact pricing and operation are treaty-specific.
Reinstatement also addresses a different dimension from the risk-level attachment itself. A $2 million per-risk layer can theoretically respond repeatedly to independent risks if no event or annual cap restricts it. If the treaty imposes a finite term aggregate, an individual recovery may consume part of the total amount available. The critical question is whether the contract replenishes that exhausted capacity, under which conditions and on which axis. Restoring an occurrence limit need not restore a separate contract-period limit. The words "one reinstatement" alone do not answer whether the reinstatement restores a per-event cover, a per-risk capacity, or the contract aggregate.
The three SEC-filed contracts illustrate limits on aggregate exposure that demand this analysis. The 2008 per-risk contract imposed a $3 million term cap on a $1.5 million per-risk layer excess of $500,000; the 2013 property contract disclosed a $2 million term cap on $500,000 excess of $500,000; and the 2024 property treaty specifies an aggregate of $20.25 million in addition to $6.75 million per occurrence (Homeowners Choice, 2008; United Property & Casualty, 2013; Lemonade and Metromile, 2024). The contract's reinstatement and termination clauses, rather than the ratios of those numbers, control the availability of renewed protection.
A hypothetical replenishment calculation
Suppose a treaty covers an initial $4 million aggregate amount during one year with one paid reinstatement of up to a further $4 million of capacity, provided the qualifying loss has reduced the initial available amount. Assume, expressly for this illustrative calculation, a full original premium P of $600,000 for the initial capacity, a reinstatement charged at 100% of that original premium pro rata as to the amount of exhausted cover, no time adjustment, and no other conditions. If the first eligible claim consumes $1 million, a proportional reinstatement of that $1 million would charge
If the first loss instead exhausts all $4 million, restoring all $4 million would cost $600,000 under those assumptions. Where a contract also applies a time-based pro rata term, the proportion may change according to the remaining time or the elapsed time as actually specified in the treaty; "pro rata" must be examined for both numerator and denominator. A dollar-denominated replenishment can be expressed as , with an explicit contractual maximum. The original premium and reinstatement premium are distinct cash flows; the latter may be payable immediately, deducted from a loss settlement, or subject to another arrangement. This example asserts no market premium quotation or actual clause from the three filed contracts.
The maximum liability under an agreement allowing one full replenishment of an initially available aggregate K might reach 2K if all other terms permit it; it would not follow that the per-risk limit, event cap and any other aggregate are each doubled. If a contract instead states an absolute cap on all reinsurer indemnities, a purported reinstatement cannot be assumed to expand that cap. Contract comparison must therefore distinguish a provision that describes total available limits including reinstatements from one that describes initial limits plus replenishment. The accounting system should record original available capacity, consumed amounts, eligible reinstatement, charged premiums, newly available capacity and ultimate exhaustion separately.
The dynamics of successive claims
Consider a hypothetical program with a $5 million occurrence cap and $10 million term aggregate, and no reinstatement. If three successive events generate otherwise eligible risk-level recoveries of $2 million, $5 million and $4 million, the first two events consume $7 million. The third event can collect only $3 million of its $4 million preliminary recovery, because only $3 million remains under the aggregate cap. The term total is $10 million. If the same events occurred in a different sequence, the amount paid to the individual third event might change although total treaty-period indemnity would still reach $10 million under these simplified assumptions. That observation matters for claims-level allocation, actuarial cash-flow timing and the classification of outstanding recoverables. It does not establish a legal chronological priority rule for policies with simultaneous or disputed claims.
Introduce a conditional aggregate reinstatement, and the distribution of remaining capacity becomes path-dependent. A claim may generate both reinsurance recovery and an additional premium liability. The insured portfolio has now acquired a contingent cash outflow dependent on the occurrence and size of loss. The value of reinstatement to the cedent must therefore be assessed net of its cost, subject to whether the cover is restored before future losses and whether the premium obligation is legally collectible.
Expected indemnity alone cannot resolve the liquidity question. Under severe claims clustering, a treaty may satisfy its actuarial expected loss function but pay too late to prevent liquidity strain. In addition, a reinsurer's failure during the claims-settlement period can weaken the effective value of the restored cover. Contract wording should specify timing, notices, premium adjustment, credit and security where relevant; risk management should reflect these facts in catastrophe stress tests and counterparty exposure.
Actuarial valuation of an excess layer
An exposure-based expected loss cost for per-risk coverage starts with a model of each subject risk's loss severity and frequency. For a risk whose nonnegative loss severity is random variable X, the expected pure layer payment is , where is the loss survival function. Where N is a frequency random variable and X_1,...,X_N are independent identically distributed severities independent of N, the expected uncapped annual recovery is
This expectation identity is an illustration of a specified probabilistic model. Actual risk-level events need not be independent, annual claims need not have an identical severity distribution, and adjustments, expense treatment and treaty aggregates can alter the payment process. In particular, the expected capped amount cannot be obtained by merely applying the cap to the expected uncapped amount. For a whole portfolio, simulation or suitable exact distributional methods must reproduce correlation of losses, multiple exposures per event, coverage exclusions and limit erosion.
Clark's reinsurance pricing work outlines practical experience-based and exposure-based approaches, including challenges arising from loss development and changes in retentions (Clark, 1996). Snoussi and Walhin's ASTIN analysis revisits the combination of exposure and experience information in per-risk excess pricing (Snoussi & Walhin, 2012). Their work supports the methodological point that a portfolio's risk distribution must be inferred using evidence appropriate to high-severity and sparsely observed claims. The present monograph does not replicate their proprietary curve fits or assert an empirical superiority for a particular fitted curve without the data needed to evaluate it.
Experience rating reconstructs past claims under the proposed treaty terms. This is especially important where a historical policy's original deductible, participation, exposure limits or original reinsurance inuring clauses differ from the prospective arrangement. Claims must be trended, developed and transformed to a comparable subject basis, with a materiality assessment for inflation, catastrophe clustering and portfolio composition. Exposure rating uses insured values, exposure characteristics and modeled severity or loss-frequency curves to estimate layer loss even when the actual historical record contains few qualifying losses. The relative credibility of the two is not a universal constant; it depends on sample size, stability of mix, data quality, tail extrapolation and changes in the policy and treaty terms.
Price, rate on line, and capital
The expected value of indemnity under the payment function is a pure risk cost, rather than the full premium that a commercially sustainable contract necessarily requires. A price may include expenses, commissions and brokerage, cost of risk capital, credit and liquidity effects, uncertainty margins, operational costs and expected reinstatement premiums. A premium divided by the nominal layer limit is often called rate on line; for example, a hypothetical premium of $400,000 for $4 million of nominal layer capacity gives a 10% rate on line. That ratio says little by itself about the risk distribution or available aggregate capacity. If the same layer is subject to a $4 million annual cap instead of an unrestricted annual aggregate, the expected payment and tail may change markedly without the headline per-risk limit moving.
The 2024 property reinsurance filing redacts commercially sensitive premium details. The numerical layer limits can be identified from the public filing; any rate-on-line inference for that particular transaction would therefore be unsupported. Nor can a public annual report's total reinsurance expense be allocated mechanically to one treaty without transaction-level accounting. Contract limits and premium consideration are separate evidence fields.
In portfolio terms, per-risk excess protection can dampen the distribution's range between attachment and exhaustion, but it leaves the first-loss layer with the insurer and may leave an extreme tail above exhaustion. Event caps and annual exhaustion can simultaneously concentrate the retained tail in catastrophe states. A rigorous capital analysis must evaluate the joint distribution of gross loss, recovered amounts and counterparty performance, including the timing of claims settlement, rather than measuring only expected ceded loss. Risk-capital recognition under a supervisory standard is conditional on contract effect and regulatory requirements; it cannot be derived mechanically from nominal coverage.
Reinsurance credit and actual collection
An actuarially modeled reinsurance recoverable is an expectation of payment conditional on the agreement's terms and counterparty performance. The treaty's legal value additionally depends on enforceability, collateral, counterparty financial strength, concentration, liquidity and settlement mechanisms. The International Association of Insurance Supervisors' ICP 13 reinsurance standard recognizes that supervision must examine reinsurance arrangements and related risk transfer, including the effectiveness and manageability of arrangements (IAIS, 2024). Its 2026 peer review examines supervisory implementation across assessed jurisdictions (IAIS, 2026). The report cannot supply a credit rating or contract enforcement outcome for any particular participating reinsurer.
The IFRS 17 standard overview requires distinct accounting treatment for insurance contracts issued and reinsurance contracts held, subject to the standard's scope and accounting rules (IFRS Foundation, 2023). A reinsurance recovery is not simply a cancellation of an original insurer's claims obligation. An IFRS Foundation reinsurance contracts guide clarifies the treatment of reinsurance contracts held and related recognition and measurement issues (IFRS Foundation, 2018). Accounting and solvency recognition depend on the applicable reporting framework, legal form, cash-flow estimates and counterparty risk. The details should be analyzed against currently effective national rules where a real issuer's financial statement is at issue.
Comparative Analysis and Integrated Scholarly Synthesis
Comparing three actual U.S. per-risk agreements
The three original SEC exhibits reveal variation in both numeric design and word choices. The 2008 Homeowners Choice contract covers a $1.5 million excess of $500,000 layer and an overall $3 million term cap. The 2013 United Property & Casualty contract covers $500,000 excess of $500,000 with $2 million available across the term. The 2024 Lemonade/Metromile property contract covers $2.25 million excess of $750,000 with a separate $6.75 million occurrence cap and $20.25 million term aggregate. Each limit must be read with the definition of one risk and the applicable recoverable-loss wording. Their disclosed figures concern different underlying books and periods; they do not form a homogeneous dataset for estimating changes in reinsurance cost. Subsequent Lemonade quarterly SEC reporting states that the per-risk contract effective July 1, 2025 expired June 30, 2026 and was not renewed (Lemonade, 2026). This confirms the 2024 exhibit is examined as historical contract evidence, rather than a representation of an in-force arrangement in November 2026.
A normalized comparison is possible by dividing the exhaustion point by the retention and the aggregate by the per-risk limit, but ratios alone cannot capture exposure. In 2008 the $1.5 million cover above $500,000 exhausts at $2 million, while its $3 million term aggregate equals twice its individual layer limit. The 2013 $500,000 above $500,000 layer exhausts at $1 million, while its $2 million term aggregate is four times the individual layer limit. In 2024 the $2.25 million layer above $750,000 exhausts at $3 million, with occurrence and term limits equal to three and nine times the individual layer limit, respectively. These ratios establish structural differences. They do not prove that any particular arrangement has a greater expected payout or superior cost efficiency, because the number and distribution of insured risks and hazard characteristics are unknown.
The 2024 event cap also makes transparent the interaction between two distinct grouping criteria. Individual risk losses are first calculated as eligible per-risk layer recoveries; then a single occurrence combines those amounts under a separate ceiling. This contractual layering permits the reinsurer to limit its concentration from correlated weather or other multi-risk events. It makes catastrophe exposure a concern even in a treaty conventionally categorized as "per risk." The choice of contractual grouping, not the contract's label, determines the tail capacity delivered.
Comparative regulation: standards, enacted law, and recognition
At the international level, IAIS ICP 13 establishes supervisory expectations concerning effective reinsurance risk management and transfer (IAIS, 2024). IAIS standards are applied through supervisory systems; they are not themselves universal private contract clauses. An analysis of any country must therefore determine the domestic authority implementing those principles and the actual legal treatment of the treaty. The 2026 IAIS peer review provides evidence about supervisory approaches and gaps at the level reported by the IAIS, not individual treaty enforceability (IAIS, 2026).
In the European Union, the Solvency II Delegated Regulation contains technical conditions for recognition of risk-mitigation techniques in the solvency capital calculation, including requirements that the arrangement be legally effective and enforceable, and for dealing with counterparty exposure (European Commission, 2015). The regulation's provisions apply as part of the EU legal and supervisory framework and may have been amended or supplemented by the date of a specific financial calculation. Whether a per-risk treaty reduces an EU insurer's solvency requirement therefore turns on the relevant operative provisions, the substance of risk transfer and credit quality, not simply the notation "A."
In the United States, reinsurance credit is substantially a matter of state insurance law and applicable regulations. The National Association of Insurance Commissioners maintains a model laws repository, including model legislation on credit for reinsurance; models guide state adoption but do not themselves impose uniform obligations unless enacted or otherwise implemented by the relevant jurisdiction (NAIC, n.d.). The NAIC Reinsurance Task Force publishes working materials concerning credit for reinsurance and related oversight (NAIC, n.d.). A responsible contract review must identify the cedent's domiciliary state, the reinsurer's authorization or reciprocal-jurisdiction status, any applicable collateral requirements, and the reporting date.
The regulatory comparison yields a stable principle of interpretation: the existence of a binding private right to be reimbursed, the amount recoverable according to the treaty, and the amount recognized for statutory or regulatory purposes are related but separate determinations. A reinsurance agreement may define a claim without fully satisfying a particular capital recognition test; conversely, a supervisor's framework does not rewrite the loss calculation agreed between the parties. The analyst should keep a contract calculation, legal-enforceability analysis, and solvency-accounting analysis in separate documentary files before integrating them.
Comparative law: follow-the-settlements and coverage limits
A cedent handles claims and makes settlements under direct policies; reinsurers seek to avoid duplicating all primary claims adjustment. Follow-the-fortunes and follow-the-settlements clauses are common mechanisms for allocating this relationship. However, their force and limits depend on the contract and governing law. The New York Court of Appeals' Travelers v. Certain Underwriters (2001) is a jurisdictionally bounded illustration: follow-the-fortunes language does not necessarily displace the reinsurance contract's own liability limit. The case supports a more precise reading of the settlement clause alongside the cap, rather than an automatic unlimited reimbursement for any good-faith primary settlement.
An original insured's claims right ordinarily runs against its original insurer, while the cedent's reinsurance right runs under the separate reinsurance agreement. The U.S. Third Circuit's Travelers v. INA opinion (2010) discusses this structural distinction in its particular judicial and factual setting. Exceptions and direct-rights provisions must be resolved under the applicable law and contract. A statement of basic contractual separateness should therefore not be expanded into a universal conclusion about insolvency proceedings, cut-through clauses or direct third-party enforcement.
A comparative legal review of per-risk cover should address at least the applicable law and dispute forum; contract formation and priority of endorsements; coverage following original policies; the definition of a risk; the scope of ultimate net loss; allocated expense and extracontractual liability; settlement discretion; notice and reporting obligations; event aggregation; exclusions; recovery and inuring clauses; maximum limits and reinstatements; claims timing; governing insolvency protections; and allocation of legal expenses. Each question can affect the amount or collectability of a single treaty claim. Because privately negotiated treaty language varies, none of the three public contracts can be used as a substitute for reviewing the particular contract being interpreted.
Comparative pricing and methodological limits
Experience rating is strongest where detailed claims experience can be reconstructed under the prospective treaty definitions and where the underlying exposure mix has remained reasonably stable. Exposure rating can contribute information where historical losses have not reached the prospective attachment point, provided the selected severity or exposure curve is justified and sensitivity tested. Snoussi and Walhin (2012) examine this methodological interface specifically for property per-risk excess. Their literature can inform model choices but cannot give an empirical rate for the publicly filed 2024 treaty from its disclosed limit structure.
A source-based estimate should classify uncertainty into identifiable components: which claims qualify; which insured structure constitutes one risk; how multiple losses are clustered by event; how development and social inflation affect severity; how expenses and recoveries are netted; how contract-period caps are exhausted; whether capacity is replenished; and how counterparty performance affects actual recovery. These are distinct sources of uncertainty with different possible mitigations. Some may be addressed by better wording, some by more granular data, some by stochastic modeling, and others by collateral, diversification or liquidity planning.
The threshold formula alone can lead to a dangerously narrow actuarial comparison. Two treaties with identical per-risk limits may produce different recoveries where one defines a whole location as one risk and the other defines buildings individually. Two treaties with identical preliminary recoveries may produce different paid amounts because their occurrence caps differ. Two treaties with identical expected indemnities may impose different effective costs when one charges reinstatement premiums and the other does not. Two treaties with identical legal recoverables may offer different realized financial protection because their reinsurers have different counterparty and liquidity characteristics. Each pair establishes a separate mechanism requiring explicit attention.
Integrated analytical findings
The document and mathematical evidence support five integrated findings.
First, contractual "risk" is a measurement convention with real distributional consequences. The unit determines how many retentions are charged, how many per-risk limits can be reached, and how reported losses must be aggregated. When the agreement gives one party judgment over risk classification, that judgment remains bounded by the actual wording, claims facts and applicable law.
Second, a reinsured layer is a contingent payment function of a defined subject-loss amount. Gross insurance loss, ultimate net loss, and collectable reinsurance are successive calculations requiring distinct evidence. Nonlinear attachment and cap operations make it unsafe to move salvage deductions, proportional cessions or event aggregation arbitrarily through the waterfall.
Third, event and term constraints can reshape the financial protection far more than a nominal per-risk limit implies. A property per-risk treaty with an occurrence cap can surrender most apparent per-risk capacity precisely when several risks are damaged by one catastrophe. This is a direct mathematical result of truncating the sum of per-risk recoveries and does not depend on claiming a new theory.
Fourth, any reinstatement provision should be valued as a contingent extension of capacity with possible additional premium, conditions and timing. Reinstatement is an economic decision within the agreement, not a synonym for all layers automatically renewing after loss. It should be expressed through a separate capacity ledger and tested under realistic sequences of claims.
Fifth, evaluation is incomplete without legal and institutional realization. Private indemnity rights, claims procedures, regulatory credit and accounting recognition, and actual payment liquidity occupy connected but different stages. A robust cedent assessment therefore traces losses from the original insured risk to a reinsurance cash receipt, keeping each contractual and supervisory condition visible.
These findings are original in their integrated comparison of verified contract structures, actuarial formulas and supervisory and judicial evidence. They are analytical results about established contracts and methods. The study does not claim to have introduced the concept of a per-risk excess layer, a new pricing theory, a new doctrine of risk aggregation or a statistical estimate of current market prices.
Limitations and future empirical questions
Three public SEC exhibits offer direct contract evidence but cannot establish a representative market survey. The 2024 exhibit contains redactions, and the other contracts belong to past periods. A credible cross-market empirical estimate would need a governed database of actual risk-by-risk claims, location attributes, policy terms, exposure schedules, treaty endorsements, reinsurer shares, loss-adjustment expense, catastrophe grouping, recoverable development, reinstatement premiums, credit characteristics and paid cash flow. The data would have to preserve confidentiality, contractual differences and comparability across years.
Further research could measure the sensitivity of paid reinsurance to alternative building-versus-location risk definitions, or estimate how event cap exhaustion changes under spatially correlated hazards. It could also test the incremental value of exposure rating where severe loss experience is thin, and quantify the liquidity impact of settlement delays or counterparty downgrade. Such studies require data, validation and methods suited to the question. No numerical results of those unperformed studies are asserted here.
Conclusion
Per-risk excess of loss reinsurance transfers a contractually defined band of loss associated with an individually defined insured risk. Its apparent simplicity arises from a familiar formula: an attachment point, a limited recovery interval, and an exhaustion point. The complete financial mechanism is more extensive. It requires the identification of the covered risk, the applicable time and peril, the original insurance obligation, the components of ultimate net loss, the ordering of preceding and inuring recoveries, the share of subscribing reinsurers, any occurrence and term aggregate caps, and any permitted reinstatement. Each element is capable of changing the cedent's realized protection.
The principal contractual finding is that the definition of one risk governs both the unit of attachment and the allocation of losses across retentions and limits. The principal mathematical finding is that a per-risk layer and an occurrence or term cap are successive nonlinear operations. A set of individually qualifying losses does not necessarily produce the sum of their maximum per-risk recoveries. The 2024 publicly filed property treaty provides a direct example: its $2.25 million layer above $750,000 per risk coexists with a $6.75 million occurrence cap and a $20.25 million contract-term cap. These figures identify different rights and constraints that must be computed separately.
The principal actuarial finding is that expected layer loss is an integral of the subject-loss survival distribution over the attachment band. That formula is a starting point for pricing, while real treaty value also depends on risk dependence, severity development, accumulation, exclusions, expense treatment, aggregate exhaustion, reinstatement premiums and counterparty performance. The legal and regulatory comparison establishes that enforceable reinsurance indemnity, solvency recognition and accounting recovery are related determinations governed by distinct documents and requirements.
The analytical synthesis yields a practical criterion of adequacy: a credible assessment of per-risk excess reinsurance can reconstruct, from original risk and claim documentation, the complete path from covered ground-up loss to net paid reinsurance under every binding layer, event and term constraint. It must reproduce not only the theoretical limit but the conditions under which the amount becomes payable and collectable. This standard of contractual and empirical traceability is a conclusion drawn from the verified evidence, not the announcement of a new theoretical system.
The study is therefore complete as a source-grounded doctrinal, actuarial and institutional analysis of per-risk excess of loss reinsurance. Its empirical reach remains bounded by the disclosed agreements, official materials and published scholarship examined here. No first-publication date attaches to this manuscript archive; publication remains a separate authorized editorial action.
References
Aon. (2025). Client Training: Introduction to Reinsurance Seminar — Seminar Overview [training outline]. Aon. https://www.aon.com/reinsurance/getmedia/4688005c-3ff1-46a6-83f1-6cdd0bb93633/rs_us_client_training_2025_overview_and_schedule.pdf
Clark, D. R. (1996). Basics of Reinsurance Pricing. Casualty Actuarial Society. https://www.casact.org/abstract/basics-reinsurance-pricing
European Commission. (2015). Commission Delegated Regulation (EU) 2015/35 of 10 October 2014 supplementing Directive 2009/138/EC. Official Journal of the European Union. https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX%3A32015R0035
Federal Insurance Office. (2014). The Breadth and Scope of the Global Reinsurance Market and the Critical Role Such Market Plays in Supporting Insurance in the United States. U.S. Department of the Treasury. https://home.treasury.gov/system/files/311/FIO%20-Reinsurance%20Report.pdf
HM Revenue & Customs. (2016; updated 2023). General Insurance Manual: GIM8070 — Reinsurance and Other Forms of Risk Transfer: Types of Reinsurance: Non-Proportional Reinsurance: Example of Excess of Loss. HMRC, United Kingdom. https://www.gov.uk/hmrc-internal-manuals/general-insurance-manual/gim8070
Homeowners Choice Property and Casualty Insurance Company, Inc. (2008). Excess Per Risk Reinsurance Contract [SEC-filed contract exhibit]. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1400810/000119312508095817/dex1015.htm
International Association of Insurance Supervisors. (2024). Insurance Core Principle 13: Reinsurance and Other Forms of Risk Transfer. IAIS online standards. https://www.iais.org/icp-online-tool/13524-icp-13reinsurance-and-other-forms-of-risk-transfer/
International Association of Insurance Supervisors. (2026). Aggregate Report on the Peer Review Process: ICP 13, Reinsurance and Other Forms of Risk Transfer. IAIS, April. https://www.iais.org/uploads/2026/04/Aggregate-report-of-peer-review-of-reinsurance-and-other-forms-of-risk-transfer-relative-to-ICP-13.pdf
IFRS Foundation. (2018). IFRS 17 Pocket Guide on Reinsurance Contracts Held. IFRS Foundation. https://www.ifrs.org/content/dam/ifrs/supporting-implementation/ifrs-17/ifrs-17-pocket-guide-on-reinsurance-contracts-held.pdf
IFRS Foundation. (2023). IFRS 17 Insurance Contracts [standard overview and supporting materials]. IFRS Foundation. https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/
Lemonade, Inc. (2026). Quarterly Report on Form 10-Q for the period ended June 30, 2026 [property per risk excess of loss reinsurance disclosure]. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1691421/000169142126000055/lmnd-20260630.htm
Lemonade Insurance Company and Metromile Insurance Company. (2024). Property Per Risk Excess of Loss Reinsurance Contract [SEC-filed Exhibit 10.2; specified material redacted]. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1691421/000169142124000108/exhibit102-pprcontractju.htm
National Association of Insurance Commissioners. (n.d.). Model Laws. NAIC. https://content.naic.org/model-laws
National Association of Insurance Commissioners. (n.d.). Reinsurance (E) Task Force. NAIC. https://content.naic.org/committees/e/reinsurance-tf
Patrik, G. S. (2001). Reinsurance. In Foundations of Casualty Actuarial Science (4th ed., chap. 7). Casualty Actuarial Society. https://www.casact.org/sites/default/files/2021-03/7_Patrik.pdf
Reinsurance Association of America. (n.d.). Glossary of Reinsurance Terms. RAA. https://www.reinsurance.org/RAA/RAA/About-the-RAA/Glossary/Glossary%20of%20Reinsurance%20Terms.aspx
Snoussi, M., & Walhin, J.-F. (2012). Experience and exposure rating for property per risk excess of loss reinsurance revisited. ASTIN Bulletin: The Journal of the IAA, 42(1), 233–270. https://www.cambridge.org/core/journals/astin-bulletin-journal-of-the-iaa/article/abs/experience-and-exposure-rating-for-property-per-risk-excess-of-loss-reinsurance-revisited/2F1A730DD0C9798A211A354C41B7F45D
Swiss Re. (n.d.). Proportional and Non-Proportional Reinsurance [technical publication]. Swiss Re. https://www.swissre.com/dam/jcr%3A1ec8b4cd-dafc-411f-aafd-385d6517ccfd/pub_proportional_non_proportional_en.pdf
Travelers Casualty & Surety Co. v. Certain Underwriters at Lloyd's of London. (2001). New York Court of Appeals decision. https://www.law.cornell.edu/nyctap/I01_0108.htm
Travelers Casualty & Surety Co. v. Insurance Company of North America. (2010). 609 F.3d 143 (3d Cir.). https://law.justia.com/cases/federal/3d-cir/2010/609-f3d-143.html
United Property & Casualty Insurance Company. (2013). Property Per Risk Excess of Loss Reinsurance Contract [SEC-filed Exhibit 10.5]. U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1401521/000140152113000044/exh105perriskreinsuranceco.htm